cache images from fist commit in the PR and re-use it until end of PR

Update the quick-start and CLI reference sections to reflect
per-relay cached images (localchat-test0, localchat-test1),
parallel deploy, DNS readiness checks, and revised memory
limits (256 MiB for DNS container).  Add mention of section
timing summary printed at the end of lxc-test.

.dockerignore

@@ -1,18 +0,0 @@
-data/
-venv/
-__pycache__
-*.pyc
-*.orig
-*.ini
-.pytest_cache
-.env
-
-# Slim build context — .git/ alone can be 100s of MB
-.git
-.github/
-docs/
-tests/
-
-# Exclude markdown files but keep www/src/*.md (used by WebsiteDeployer)
-*.md
-!www/**/*.md

.github/workflows/ci.yaml

@@ -15,28 +15,102 @@ jobs:
         with:
           ref: ${{ github.event.pull_request.head.sha }}
       - name: download filtermail
-        run: curl -L https://github.com/chatmail/filtermail/releases/download/v0.3.0/filtermail-x86_64 -o /usr/local/bin/filtermail && chmod +x /usr/local/bin/filtermail
-      - name: run chatmaild tests 
+        run: curl -L https://github.com/chatmail/filtermail/releases/download/v0.5.2/filtermail-x86_64 -o /usr/local/bin/filtermail && chmod +x /usr/local/bin/filtermail
+      - name: run chatmaild tests
         working-directory: chatmaild
         run: pipx run tox
 
   scripts:
-    name: deploy-chatmail tests 
+    name: deploy-chatmail tests
     runs-on: ubuntu-latest
     steps:
       - uses: actions/checkout@v4
 
-      - name: initenv 
+      - name: initenv
         run: scripts/initenv.sh
 
       - name: append venv/bin to PATH
         run: echo venv/bin >>$GITHUB_PATH
 
-      - name: run formatting checks 
-        run: cmdeploy fmt -v 
+      - name: run formatting checks
+        run: cmdeploy fmt -v
 
-      - name: run deploy-chatmail offline tests 
-        run: pytest --pyargs cmdeploy 
+      - name: run deploy-chatmail offline tests
+        run: pytest --pyargs cmdeploy
 
-      # all other cmdeploy commands require a staging server 
-      # see https://github.com/deltachat/chatmail/issues/100
+  lxc-test:
+    name: LXC deploy and test
+    runs-on: ubuntu-24.04
+    timeout-minutes: 30
+    steps:
+      - uses: actions/checkout@v4
+        with:
+          ref: ${{ github.event.pull_request.head.sha }}
+
+      - name: install incus
+        run: |
+          # zabbly is the official incus community packages source
+          curl -fsSL https://pkgs.zabbly.com/key.asc \
+            | sudo gpg --dearmor -o /etc/apt/keyrings/zabbly.gpg
+          sudo sh -c 'cat <<EOF > /etc/apt/sources.list.d/zabbly-incus-stable.sources
+          Enabled: yes
+          Types: deb
+          URIs: https://pkgs.zabbly.com/incus/stable
+          Suites: $(. /etc/os-release && echo ${VERSION_CODENAME})
+          Components: main
+          Architectures: $(dpkg --print-architecture)
+          Signed-By: /etc/apt/keyrings/zabbly.gpg
+          EOF'
+          sudo apt-get update
+          sudo apt-get install -y incus
+
+      - name: initialise incus
+        run: |
+          sudo systemctl stop docker.socket docker || true
+          sudo iptables -P FORWARD ACCEPT
+          sudo sysctl -w fs.inotify.max_user_instances=65535
+          sudo sysctl -w fs.inotify.max_user_watches=65535
+          sudo incus admin init --minimal
+          sudo usermod -aG incus-admin "$USER"
+
+      - name: initenv
+        run: scripts/initenv.sh
+
+      - name: append venv/bin to PATH
+        run: echo venv/bin >>$GITHUB_PATH
+
+      - name: restore cached images
+        id: cache-images
+        uses: actions/cache@v4
+        with:
+          path: |
+            /tmp/localchat-base.tar.gz
+            /tmp/localchat-ns.tar.gz
+            /tmp/localchat-test0.tar.gz
+            /tmp/localchat-test1.tar.gz
+            lxconfigs/id_localchat*
+          key: incus-images-${{ runner.os }}-${{ github.ref_name }}
+          restore-keys: |
+            incus-images-${{ runner.os }}-${{ github.ref_name }}-
+            incus-images-${{ runner.os }}-main-
+            incus-images-${{ runner.os }}-
+
+      - name: import cached images
+        run: |
+          for alias in localchat-base localchat-ns localchat-test0 localchat-test1; do
+            if [ -f /tmp/$alias.tar.gz ]; then
+              sg incus-admin -c "incus image import /tmp/$alias.tar.gz --alias $alias" || true
+            fi
+          done
+
+      - name: lxc-test
+        run: sg incus-admin -c 'cmdeploy lxc-test'
+
+      - name: export images for cache
+        if: always()
+        run: |
+          for alias in localchat-base localchat-ns localchat-test0 localchat-test1; do
+            if ! [ -f /tmp/$alias.tar.gz ]; then
+              sg incus-admin -c "incus image export $alias /tmp/$alias" || true
+            fi
+          done

.github/workflows/docker-build.yaml

@@ -1,76 +0,0 @@
-name: Docker Build
-
-on:
-  pull_request:
-    paths:
-      - 'docker/**'
-      - 'docker-compose.yaml'
-      - '.dockerignore'
-      - 'chatmaild/**'
-      - 'cmdeploy/**'
-      - '.github/workflows/docker-build.yaml'
-  push:
-    branches:
-      - main
-      - j4n/docker
-    paths:
-      - 'docker/**'
-      - 'docker-compose.yaml'
-      - '.dockerignore'
-      - 'chatmaild/**'
-      - 'cmdeploy/**'
-      - '.github/workflows/docker-build.yaml'
-    tags:
-      - 'v*'
-
-env:
-  REGISTRY: ghcr.io
-  IMAGE_NAME: ${{ github.repository }}
-
-jobs:
-  build:
-    name: Build Docker image
-    runs-on: ubuntu-latest
-    permissions:
-      contents: read
-      packages: write
-    steps:
-      - uses: actions/checkout@v4
-
-      - name: Set up Docker Buildx
-        uses: docker/setup-buildx-action@v3
-
-      - name: Login to GHCR
-        if: github.event_name == 'push'
-        uses: docker/login-action@v3
-        with:
-          registry: ${{ env.REGISTRY }}
-          username: ${{ github.actor }}
-          password: ${{ secrets.GITHUB_TOKEN }}
-
-      - name: Extract metadata (tags, labels)
-        id: meta
-        uses: docker/metadata-action@v5
-        with:
-          images: ${{ env.REGISTRY }}/${{ env.IMAGE_NAME }}
-          tags: |
-            # Tagged releases: v1.2.3 → :1.2.3, :1.2, :latest
-            type=semver,pattern={{version}}
-            type=semver,pattern={{major}}.{{minor}}
-            # Branch pushes: j4n/docker → :j4n-docker
-            type=ref,event=branch
-            # Always: :sha-<hash>
-            type=sha
-
-      - name: Build and push
-        uses: docker/build-push-action@v6
-        with:
-          context: .
-          file: docker/chatmail_relay.dockerfile
-          push: ${{ github.event_name == 'push' }}
-          tags: ${{ steps.meta.outputs.tags }}
-          labels: ${{ steps.meta.outputs.labels }}
-          cache-from: type=gha
-          cache-to: type=gha,mode=max
-          build-args: |
-            GIT_HASH=${{ github.sha }}

.github/workflows/staging-ipv4.testrun.org-default.zone

@@ -1,20 +0,0 @@
-;; Zone file for staging-ipv4.testrun.org
-
-$ORIGIN staging-ipv4.testrun.org.
-$TTL 300
-
-@  IN  SOA     ns.testrun.org. root.nine.testrun.org (
-  2023010101 ; Serial
-  7200       ; Refresh
-  3600       ; Retry
-  1209600    ; Expire
-  3600       ; Negative response caching TTL
-)
-
-;; Nameservers.
-@  IN NS ns.testrun.org.
-
-;; DNS records.
-@       IN      A       37.27.95.249
-mta-sts.staging-ipv4.testrun.org.    CNAME   staging-ipv4.testrun.org.
-www.staging-ipv4.testrun.org.        CNAME   staging-ipv4.testrun.org.

.github/workflows/staging.testrun.org-default.zone

@@ -1,21 +0,0 @@
-;; Zone file for staging2.testrun.org
-
-$ORIGIN staging2.testrun.org.
-$TTL 300
-
-@  IN  SOA     ns.testrun.org. root.nine.testrun.org (
-  2023010101 ; Serial
-  7200       ; Refresh
-  3600       ; Retry
-  1209600    ; Expire
-  3600       ; Negative response caching TTL
-)
-
-;; Nameservers.
-@  IN NS ns.testrun.org.
-
-;; DNS records.
-@       IN      A       37.27.24.139
-mta-sts.staging2.testrun.org.    CNAME   staging2.testrun.org.
-www.staging2.testrun.org.        CNAME   staging2.testrun.org.
-

.github/workflows/test-and-deploy-ipv4only.yaml

@@ -1,95 +0,0 @@
-name: deploy on staging-ipv4.testrun.org, and run tests
-
-on:
-  push:
-    branches:
-      - main
-  pull_request:
-    paths-ignore:
-      - 'scripts/**'
-      - '**/README.md'
-      - 'CHANGELOG.md'
-      - 'LICENSE'
-
-jobs:
-  deploy:
-    name: deploy on staging-ipv4.testrun.org, and run tests
-    runs-on: ubuntu-latest
-    timeout-minutes: 30
-    environment:
-      name: staging-ipv4.testrun.org
-      url: https://staging-ipv4.testrun.org/
-    concurrency: staging-ipv4.testrun.org
-    steps:
-      - uses: actions/checkout@v4
-
-      - name: prepare SSH
-        run: |
-          mkdir ~/.ssh
-          echo "${{ secrets.STAGING_SSH_KEY }}" >> ~/.ssh/id_ed25519
-          chmod 600 ~/.ssh/id_ed25519
-          ssh-keyscan staging-ipv4.testrun.org > ~/.ssh/known_hosts
-          # save previous acme & dkim state
-          rsync -avz root@staging-ipv4.testrun.org:/var/lib/acme acme-ipv4 || true
-          rsync -avz root@staging-ipv4.testrun.org:/etc/dkimkeys dkimkeys-ipv4 || true
-          # store previous acme & dkim state on ns.testrun.org, if it contains useful certs
-          if [ -f dkimkeys-ipv4/dkimkeys/opendkim.private ]; then rsync -avz -e "ssh -o StrictHostKeyChecking=accept-new" dkimkeys-ipv4 root@ns.testrun.org:/tmp/ || true; fi
-          if [ "$(ls -A acme-ipv4/acme/certs)" ]; then rsync -avz -e "ssh -o StrictHostKeyChecking=accept-new" acme-ipv4 root@ns.testrun.org:/tmp/ || true; fi
-          # make sure CAA record isn't set
-          scp -o StrictHostKeyChecking=accept-new .github/workflows/staging-ipv4.testrun.org-default.zone root@ns.testrun.org:/etc/nsd/staging-ipv4.testrun.org.zone
-          ssh root@ns.testrun.org sed -i '/CAA/d' /etc/nsd/staging-ipv4.testrun.org.zone
-          ssh root@ns.testrun.org nsd-checkzone staging-ipv4.testrun.org /etc/nsd/staging-ipv4.testrun.org.zone
-          ssh root@ns.testrun.org systemctl reload nsd
-
-      - name: rebuild staging-ipv4.testrun.org to have a clean VPS
-        run: |
-            curl -X POST \
-            -H "Authorization: Bearer ${{ secrets.HETZNER_API_TOKEN }}" \
-            -H "Content-Type: application/json" \
-            -d '{"image":"debian-12"}' \
-            "https://api.hetzner.cloud/v1/servers/${{ secrets.STAGING_IPV4_SERVER_ID }}/actions/rebuild"
-
-      - run: scripts/initenv.sh
-
-      - name: append venv/bin to PATH
-        run: echo venv/bin >>$GITHUB_PATH
-
-      - name: upload TLS cert after rebuilding
-        run: |
-          echo " --- wait until staging-ipv4.testrun.org VPS is rebuilt --- "
-          rm ~/.ssh/known_hosts
-          while ! ssh -o ConnectTimeout=180 -o StrictHostKeyChecking=accept-new -v root@staging-ipv4.testrun.org id -u ; do sleep 1 ; done
-          ssh -o StrictHostKeyChecking=accept-new -v root@staging-ipv4.testrun.org id -u
-          # download acme & dkim state from ns.testrun.org
-          rsync -e "ssh -o StrictHostKeyChecking=accept-new" -avz root@ns.testrun.org:/tmp/acme-ipv4/acme acme-restore || true
-          rsync -avz root@ns.testrun.org:/tmp/dkimkeys-ipv4/dkimkeys dkimkeys-restore || true
-          # restore acme & dkim state to staging2.testrun.org
-          rsync -avz acme-restore/acme root@staging-ipv4.testrun.org:/var/lib/ || true
-          rsync -avz dkimkeys-restore/dkimkeys root@staging-ipv4.testrun.org:/etc/ || true
-          ssh -o StrictHostKeyChecking=accept-new -v root@staging-ipv4.testrun.org chown root:root -R /var/lib/acme || true
-
-      - name: run deploy-chatmail offline tests 
-        run: pytest --pyargs cmdeploy 
-
-      - run: |
-          cmdeploy init staging-ipv4.testrun.org
-          sed -i 's#disable_ipv6 = False#disable_ipv6 = True#' chatmail.ini
-          sed -i 's/#\s*mtail_address/mtail_address/' chatmail.ini
-          cmdeploy run --verbose --skip-dns-check
-
-      - name: set DNS entries
-        run: |
-          ssh -o StrictHostKeyChecking=accept-new -v root@staging-ipv4.testrun.org chown opendkim:opendkim -R /etc/dkimkeys
-          cmdeploy dns --zonefile staging-generated.zone
-          cat staging-generated.zone >> .github/workflows/staging-ipv4.testrun.org-default.zone
-          cat .github/workflows/staging-ipv4.testrun.org-default.zone
-          scp .github/workflows/staging-ipv4.testrun.org-default.zone root@ns.testrun.org:/etc/nsd/staging-ipv4.testrun.org.zone
-          ssh root@ns.testrun.org nsd-checkzone staging-ipv4.testrun.org /etc/nsd/staging-ipv4.testrun.org.zone
-          ssh root@ns.testrun.org systemctl reload nsd
-
-      - name: cmdeploy test
-        run: CHATMAIL_DOMAIN2=ci-chatmail.testrun.org cmdeploy test --slow
-
-      - name: cmdeploy dns
-        run: cmdeploy dns -v
-

.github/workflows/test-and-deploy.yaml

@@ -1,98 +0,0 @@
-name: deploy on staging2.testrun.org, and run tests
-
-on:
-  push:
-    branches:
-      - main
-  pull_request:
-    paths-ignore:
-      - 'scripts/**'
-      - '**/README.md'
-      - 'CHANGELOG.md'
-      - 'LICENSE'
-
-jobs:
-  deploy:
-    name: deploy on staging2.testrun.org, and run tests
-    runs-on: ubuntu-latest
-    timeout-minutes: 30
-    environment:
-      name: staging2.testrun.org
-      url: https://staging2.testrun.org/
-    concurrency: staging2.testrun.org
-    steps:
-      - uses: actions/checkout@v4
-
-      - name: prepare SSH
-        run: |
-          mkdir ~/.ssh
-          echo "${{ secrets.STAGING_SSH_KEY }}" >> ~/.ssh/id_ed25519
-          chmod 600 ~/.ssh/id_ed25519
-          ssh-keyscan staging2.testrun.org > ~/.ssh/known_hosts
-          # save previous acme & dkim state
-          rsync -avz root@staging2.testrun.org:/var/lib/acme . || true
-          rsync -avz root@staging2.testrun.org:/etc/dkimkeys . || true
-          # store previous acme & dkim state on ns.testrun.org, if it contains useful certs
-          if [ -f dkimkeys/opendkim.private ]; then rsync -avz -e "ssh -o StrictHostKeyChecking=accept-new" dkimkeys root@ns.testrun.org:/tmp/ || true; fi
-          if [ "$(ls -A acme/certs)" ]; then rsync -avz -e "ssh -o StrictHostKeyChecking=accept-new" acme root@ns.testrun.org:/tmp/ || true; fi
-          # make sure CAA record isn't set
-          scp -o StrictHostKeyChecking=accept-new .github/workflows/staging.testrun.org-default.zone root@ns.testrun.org:/etc/nsd/staging2.testrun.org.zone
-          ssh root@ns.testrun.org sed -i '/CAA/d' /etc/nsd/staging2.testrun.org.zone
-          ssh root@ns.testrun.org nsd-checkzone staging2.testrun.org /etc/nsd/staging2.testrun.org.zone
-          ssh root@ns.testrun.org systemctl reload nsd
-
-      - name: rebuild staging2.testrun.org to have a clean VPS
-        run: |
-            curl -X POST \
-            -H "Authorization: Bearer ${{ secrets.HETZNER_API_TOKEN }}" \
-            -H "Content-Type: application/json" \
-            -d '{"image":"debian-12"}' \
-            "https://api.hetzner.cloud/v1/servers/${{ secrets.STAGING_SERVER_ID }}/actions/rebuild"
-
-      - run: scripts/initenv.sh
-
-      - name: append venv/bin to PATH
-        run: echo venv/bin >>$GITHUB_PATH
-
-      - name: upload TLS cert after rebuilding
-        run: |
-          echo " --- wait until staging2.testrun.org VPS is rebuilt --- "
-          rm ~/.ssh/known_hosts
-          while ! ssh -o ConnectTimeout=180 -o StrictHostKeyChecking=accept-new -v root@staging2.testrun.org id -u ; do sleep 1 ; done
-          ssh -o StrictHostKeyChecking=accept-new -v root@staging2.testrun.org id -u
-          # download acme & dkim state from ns.testrun.org
-          rsync -e "ssh -o StrictHostKeyChecking=accept-new" -avz root@ns.testrun.org:/tmp/acme acme-restore || true
-          rsync -avz root@ns.testrun.org:/tmp/dkimkeys dkimkeys-restore || true
-          # restore acme & dkim state to staging2.testrun.org
-          rsync -avz acme-restore/acme root@staging2.testrun.org:/var/lib/ || true
-          rsync -avz dkimkeys-restore/dkimkeys root@staging2.testrun.org:/etc/ || true
-          ssh -o StrictHostKeyChecking=accept-new -v root@staging2.testrun.org chown root:root -R /var/lib/acme || true
-
-      - name: add hpk42 key to staging server
-        run: ssh root@staging2.testrun.org 'curl -s https://github.com/hpk42.keys >> .ssh/authorized_keys'
-
-      - name: run deploy-chatmail offline tests 
-        run: pytest --pyargs cmdeploy 
-
-      - run: |
-          cmdeploy init staging2.testrun.org
-          sed -i 's/#\s*mtail_address/mtail_address/' chatmail.ini
-
-      - run: cmdeploy run --verbose --skip-dns-check
-
-      - name: set DNS entries
-        run: |
-          ssh -o StrictHostKeyChecking=accept-new root@staging2.testrun.org chown opendkim:opendkim -R /etc/dkimkeys
-          cmdeploy dns --zonefile staging-generated.zone --verbose
-          cat staging-generated.zone >> .github/workflows/staging.testrun.org-default.zone
-          cat .github/workflows/staging.testrun.org-default.zone
-          scp .github/workflows/staging.testrun.org-default.zone root@ns.testrun.org:/etc/nsd/staging2.testrun.org.zone
-          ssh root@ns.testrun.org nsd-checkzone staging2.testrun.org /etc/nsd/staging2.testrun.org.zone
-          ssh root@ns.testrun.org systemctl reload nsd
-
-      - name: cmdeploy test
-        run: CHATMAIL_DOMAIN2=ci-chatmail.testrun.org cmdeploy test --slow
-
-      - name: cmdeploy dns
-        run: cmdeploy dns -v
-

.github/workflows/test-tls-external.yaml

@@ -1,37 +0,0 @@
-name: test tls_external_cert_and_key on staging2.testrun.org
-
-on:
-  workflow_run:
-    workflows:
-      - "deploy on staging2.testrun.org, and run tests"
-    types:
-      - completed
-
-jobs:
-  test-tls-external:
-    name: test tls_external_cert_and_key
-    runs-on: ubuntu-latest
-    timeout-minutes: 30
-    concurrency: staging2.testrun.org
-    environment:
-      name: staging2.testrun.org
-
-    steps:
-      - uses: actions/checkout@v4
-
-      - name: prepare SSH
-        run: |
-          mkdir -p ~/.ssh
-          echo "${{ secrets.STAGING_SSH_KEY }}" >> ~/.ssh/id_ed25519
-          chmod 600 ~/.ssh/id_ed25519
-          ssh-keyscan staging2.testrun.org >> ~/.ssh/known_hosts 2>/dev/null
-
-      - run: scripts/initenv.sh
-
-      - name: append venv/bin to PATH
-        run: echo venv/bin >>$GITHUB_PATH
-
-      - name: run tls_external e2e test
-        run: |
-          python -m cmdeploy.tests.setup_tls_external \
-            staging2.testrun.org

.gitignore

@@ -4,7 +4,8 @@ __pycache__/
 *$py.class
 *.swp
 *qr-*.png
-chatmail.ini
+chatmail*.ini
+lxconfigs/
 
 
 # C extensions
@@ -164,9 +165,3 @@ cython_debug/
 #.idea/
 
 chatmail.zone
-
-# docker
-/data/
-/custom/
-docker-compose.override.yaml
-.env

chatmaild/pyproject.toml

@@ -24,7 +24,6 @@ where = ['src']
 [project.scripts]
 doveauth = "chatmaild.doveauth:main"
 chatmail-metadata = "chatmaild.metadata:main"
-chatmail-metrics = "chatmaild.metrics:main"
 chatmail-expire = "chatmaild.expire:main"
 chatmail-fsreport = "chatmaild.fsreport:main"
 lastlogin = "chatmaild.lastlogin:main"

chatmaild/src/chatmaild/config.py

@@ -75,8 +75,7 @@ class Config:
                     " paths: CERT_PATH KEY_PATH"
                 )
             self.tls_cert_mode = "external"
-            self.tls_cert_path = parts[0]
-            self.tls_key_path = parts[1]
+            self.tls_cert_path, self.tls_key_path = parts
         elif self.mail_domain.startswith("_"):
             self.tls_cert_mode = "self"
             self.tls_cert_path = "/etc/ssl/certs/mailserver.pem"

chatmaild/src/chatmaild/doveauth.py

@@ -1,8 +1,11 @@
 import json
 import logging
 import os
+import re
 import sys
 
+import filelock
+
 try:
     import crypt_r
 except ImportError:
@@ -13,6 +16,7 @@ from .dictproxy import DictProxy
 from .migrate_db import migrate_from_db_to_maildir
 
 NOCREATE_FILE = "/etc/chatmail-nocreate"
+VALID_LOCALPART_RE = re.compile(r"^[a-z0-9._-]+$")
 
 
 def encrypt_password(password: str):
@@ -52,6 +56,10 @@ def is_allowed_to_create(config: Config, user, cleartext_password) -> bool:
         )
         return False
 
+    if not VALID_LOCALPART_RE.match(localpart):
+        logging.warning("localpart %r contains invalid characters", localpart)
+        return False
+
     return True
 
 
@@ -140,8 +148,13 @@ class AuthDictProxy(DictProxy):
         if not is_allowed_to_create(self.config, addr, cleartext_password):
             return
 
-        user.set_password(encrypt_password(cleartext_password))
-        print(f"Created address: {addr}", file=sys.stderr)
+        lock = filelock.FileLock(str(user.password_path) + ".lock", timeout=5)
+        with lock:
+            userdata = user.get_userdb_dict()
+            if userdata:
+                return userdata
+            user.set_password(encrypt_password(cleartext_password))
+            print(f"Created address: {addr}", file=sys.stderr)
         return user.get_userdb_dict()
 
 

chatmaild/src/chatmaild/fsreport.py

@@ -13,9 +13,20 @@ to show storage summaries only for first 1000 mailboxes
 
     python -m chatmaild.fsreport /path/to/chatmail.ini --maxnum 1000
 
+to write Prometheus textfile for node_exporter
+
+    python -m chatmaild.fsreport --textfile /var/lib/prometheus/node-exporter/
+
+writes to /var/lib/prometheus/node-exporter/fsreport.prom
+
+to also write legacy metrics.py style output (default: /var/www/html/metrics):
+
+    python -m chatmaild.fsreport --textfile /var/lib/prometheus/node-exporter/ --legacy-metrics
+
 """
 
 import os
+import tempfile
 from argparse import ArgumentParser
 from datetime import datetime
 
@@ -48,7 +59,19 @@ class Report:
         self.num_ci_logins = self.num_all_logins = 0
         self.login_buckets = {x: 0 for x in (1, 10, 30, 40, 80, 100, 150)}
 
-        self.message_buckets = {x: 0 for x in (0, 160000, 500000, 2000000)}
+        KiB = 1024
+        MiB = 1024 * KiB
+        self.message_size_thresholds = (
+            0,
+            100 * KiB,
+            MiB // 2,
+            1 * MiB,
+            2 * MiB,
+            5 * MiB,
+            10 * MiB,
+        )
+        self.message_buckets = {x: 0 for x in self.message_size_thresholds}
+        self.message_count_buckets = {x: 0 for x in self.message_size_thresholds}
 
     def process_mailbox_stat(self, mailbox):
         # categorize login times
@@ -68,9 +91,10 @@ class Report:
             for size in self.message_buckets:
                 for msg in mailbox.messages:
                     if msg.size >= size:
-                        if self.mdir and not msg.relpath.startswith(self.mdir):
+                        if self.mdir and f"/{self.mdir}/" not in msg.path:
                             continue
                         self.message_buckets[size] += msg.size
+                        self.message_count_buckets[size] += 1
 
         self.size_messages += sum(entry.size for entry in mailbox.messages)
         self.size_extra += sum(entry.size for entry in mailbox.extrafiles)
@@ -93,9 +117,10 @@ class Report:
 
         pref = f"[{self.mdir}] " if self.mdir else ""
         for minsize, sumsize in self.message_buckets.items():
+            count = self.message_count_buckets[minsize]
             percent = (sumsize / all_messages * 100) if all_messages else 0
             print(
-                f"{pref}larger than {HSize(minsize)}: {HSize(sumsize)} ({percent:.2f}%)"
+                f"{pref}larger than {HSize(minsize)}: {HSize(sumsize)} ({percent:.2f}%), {count} msgs"
             )
 
         user_logins = self.num_all_logins - self.num_ci_logins
@@ -111,6 +136,75 @@ class Report:
         for days, active in self.login_buckets.items():
             print(f"last {days:3} days: {HSize(active)} {p(active)}")
 
+    def _write_atomic(self, filepath, content):
+        """Atomically write content to filepath via tmp+rename."""
+        dirpath = os.path.dirname(os.path.abspath(filepath))
+        fd, tmppath = tempfile.mkstemp(dir=dirpath, suffix=".tmp")
+        try:
+            with os.fdopen(fd, "w") as f:
+                f.write(content)
+            os.chmod(tmppath, 0o644)
+            os.rename(tmppath, filepath)
+        except BaseException:
+            try:
+                os.unlink(tmppath)
+            except OSError:
+                pass
+            raise
+
+    def dump_textfile(self, filepath):
+        """Dump metrics in Prometheus exposition format."""
+        lines = []
+
+        lines.append("# HELP chatmail_storage_bytes Mailbox storage in bytes.")
+        lines.append("# TYPE chatmail_storage_bytes gauge")
+        lines.append(f'chatmail_storage_bytes{{kind="messages"}} {self.size_messages}')
+        lines.append(f'chatmail_storage_bytes{{kind="extra"}} {self.size_extra}')
+        total = self.size_extra + self.size_messages
+        lines.append(f'chatmail_storage_bytes{{kind="total"}} {total}')
+
+        lines.append("# HELP chatmail_messages_bytes Sum of msg bytes >= threshold.")
+        lines.append("# TYPE chatmail_messages_bytes gauge")
+        for minsize, sumsize in self.message_buckets.items():
+            lines.append(f'chatmail_messages_bytes{{min_size="{minsize}"}} {sumsize}')
+
+        lines.append("# HELP chatmail_messages_count Number of msgs >= size threshold.")
+        lines.append("# TYPE chatmail_messages_count gauge")
+        for minsize, count in self.message_count_buckets.items():
+            lines.append(f'chatmail_messages_count{{min_size="{minsize}"}} {count}')
+
+        lines.append("# HELP chatmail_accounts Number of accounts.")
+        lines.append("# TYPE chatmail_accounts gauge")
+        user_logins = self.num_all_logins - self.num_ci_logins
+        lines.append(f'chatmail_accounts{{kind="all"}} {self.num_all_logins}')
+        lines.append(f'chatmail_accounts{{kind="ci"}} {self.num_ci_logins}')
+        lines.append(f'chatmail_accounts{{kind="user"}} {user_logins}')
+
+        lines.append(
+            "# HELP chatmail_accounts_active Non-CI accounts active within N days."
+        )
+        lines.append("# TYPE chatmail_accounts_active gauge")
+        for days, active in self.login_buckets.items():
+            lines.append(f'chatmail_accounts_active{{days="{days}"}} {active}')
+
+        self._write_atomic(filepath, "\n".join(lines) + "\n")
+
+    def dump_compat_textfile(self, filepath):
+        """Dump legacy metrics.py style metrics."""
+        user_logins = self.num_all_logins - self.num_ci_logins
+        lines = [
+            "# HELP total number of accounts",
+            "# TYPE accounts gauge",
+            f"accounts {self.num_all_logins}",
+            "# HELP number of CI accounts",
+            "# TYPE ci_accounts gauge",
+            f"ci_accounts {self.num_ci_logins}",
+            "# HELP number of non-CI accounts",
+            "# TYPE nonci_accounts gauge",
+            f"nonci_accounts {user_logins}",
+        ]
+        self._write_atomic(filepath, "\n".join(lines) + "\n")
+
 
 def main(args=None):
     """Report about filesystem storage usage of all mailboxes and messages"""
@@ -127,19 +221,21 @@ def main(args=None):
         "--days",
         default=0,
         action="store",
-        help="assume date to be days older than now",
+        help="assume date to be DAYS older than now",
     )
     parser.add_argument(
         "--min-login-age",
         default=0,
+        metavar="DAYS",
         dest="min_login_age",
         action="store",
-        help="only sum up message size if last login is at least min-login-age days old",
+        help="only sum up message size if last login is at least DAYS days old",
     )
     parser.add_argument(
         "--mdir",
+        metavar="{cur,new,tmp}",
         action="store",
-        help="only consider 'cur' or 'new' or 'tmp' messages for summary",
+        help="only consider messages in specified Maildir subdirectory for summary",
     )
 
     parser.add_argument(
@@ -148,6 +244,21 @@ def main(args=None):
         action="store",
         help="maximum number of mailboxes to iterate on",
     )
+    parser.add_argument(
+        "--textfile",
+        metavar="PATH",
+        default=None,
+        help="write Prometheus textfile to PATH (directory or file); "
+        "if PATH is a directory, writes 'fsreport.prom' inside it",
+    )
+    parser.add_argument(
+        "--legacy-metrics",
+        metavar="FILENAME",
+        nargs="?",
+        const="/var/www/html/metrics",
+        default=None,
+        help="write legacy metrics.py textfile (default: /var/www/html/metrics)",
+    )
 
     args = parser.parse_args(args)
 
@@ -161,7 +272,15 @@ def main(args=None):
     rep = Report(now=now, min_login_age=int(args.min_login_age), mdir=args.mdir)
     for mbox in iter_mailboxes(str(config.mailboxes_dir), maxnum=maxnum):
         rep.process_mailbox_stat(mbox)
-    rep.dump_summary()
+    if args.textfile:
+        path = args.textfile
+        if os.path.isdir(path):
+            path = os.path.join(path, "fsreport.prom")
+        rep.dump_textfile(path)
+    if args.legacy_metrics:
+        rep.dump_compat_textfile(args.legacy_metrics)
+    if not args.textfile and not args.legacy_metrics:
+        rep.dump_summary()
 
 
 if __name__ == "__main__":

chatmaild/src/chatmaild/metadata.py

@@ -101,7 +101,11 @@ class MetadataDictProxy(DictProxy):
                 # Handle `GETMETADATA "" /shared/vendor/deltachat/irohrelay`
                 return f"O{self.iroh_relay}\n"
             elif keyname == "vendor/vendor.dovecot/pvt/server/vendor/deltachat/turn":
-                res = turn_credentials()
+                try:
+                    res = turn_credentials()
+                except Exception:
+                    logging.exception("failed to get TURN credentials")
+                    return "N\n"
                 port = 3478
                 return f"O{self.turn_hostname}:{port}:{res}\n"
 

chatmaild/src/chatmaild/metrics.py

@@ -1,32 +0,0 @@
-#!/usr/bin/env python3
-import sys
-from pathlib import Path
-
-
-def main(vmail_dir=None):
-    if vmail_dir is None:
-        vmail_dir = sys.argv[1]
-
-    accounts = 0
-    ci_accounts = 0
-
-    for path in Path(vmail_dir).iterdir():
-        if not path.joinpath("cur").is_dir():
-            continue
-        accounts += 1
-        if path.name[:3] in ("ci-", "ac_"):
-            ci_accounts += 1
-
-    print("# HELP total number of accounts")
-    print("# TYPE accounts gauge")
-    print(f"accounts {accounts}")
-    print("# HELP number of CI accounts")
-    print("# TYPE ci_accounts gauge")
-    print(f"ci_accounts {ci_accounts}")
-    print("# HELP number of non-CI accounts")
-    print("# TYPE nonci_accounts gauge")
-    print(f"nonci_accounts {accounts - ci_accounts}")
-
-
-if __name__ == "__main__":
-    main()

chatmaild/src/chatmaild/newemail.py

@@ -3,7 +3,6 @@
 """CGI script for creating new accounts."""
 
 import json
-import random
 import secrets
 import string
 from urllib.parse import quote
@@ -16,7 +15,9 @@ ALPHANUMERIC_PUNCT = string.ascii_letters + string.digits + string.punctuation
 
 
 def create_newemail_dict(config: Config):
-    user = "".join(random.choices(ALPHANUMERIC, k=config.username_max_length))
+    user = "".join(
+        secrets.choice(ALPHANUMERIC) for _ in range(config.username_max_length)
+    )
     password = "".join(
         secrets.choice(ALPHANUMERIC_PUNCT)
         for _ in range(config.password_min_length + 3)

chatmaild/src/chatmaild/tests/test_config.py

@@ -110,6 +110,7 @@ def test_config_tls_external_overrides_underscore(make_config):
     )
     assert config.tls_cert_mode == "external"
     assert config.tls_cert_path == "/certs/fullchain.pem"
+    assert config.tls_key_path == "/certs/privkey.pem"
 
 
 def test_config_tls_external_bad_format(make_config):

chatmaild/src/chatmaild/tests/test_doveauth.py

@@ -120,6 +120,60 @@ def test_handle_dovecot_protocol_iterate(gencreds, example_config):
     assert not lines[2]
 
 
+def test_invalid_localpart_characters(make_config):
+    """Test that is_allowed_to_create rejects localparts with invalid characters."""
+    config = make_config("chat.example.org", {"username_min_length": "3"})
+    password = "zequ0Aimuchoodaechik"
+    domain = config.mail_domain
+
+    # valid localparts
+    assert is_allowed_to_create(config, f"abc123@{domain}", password)
+    assert is_allowed_to_create(config, f"a.b-c_d@{domain}", password)
+
+    # uppercase rejected
+    assert not is_allowed_to_create(config, f"Abc123@{domain}", password)
+    assert not is_allowed_to_create(config, f"ABCDEFG@{domain}", password)
+
+    # spaces and special chars rejected
+    assert not is_allowed_to_create(config, f"a b cde@{domain}", password)
+    assert not is_allowed_to_create(config, f"abc+def@{domain}", password)
+    assert not is_allowed_to_create(config, f"abc!def@{domain}", password)
+    assert not is_allowed_to_create(config, f"ab@cdef@{domain}", password)
+    assert not is_allowed_to_create(config, f"abc/def@{domain}", password)
+    assert not is_allowed_to_create(config, f"abc\\def@{domain}", password)
+
+
+def test_concurrent_creation_same_account(dictproxy):
+    """Test that concurrent creation of the same account doesn't corrupt password."""
+    addr = "racetest1@chat.example.org"
+    password = "zequ0Aimuchoodaechik"
+    num_threads = 10
+    results = queue.Queue()
+
+    def create():
+        try:
+            res = dictproxy.lookup_passdb(addr, password)
+            results.put(("ok", res))
+        except Exception:
+            results.put(("err", traceback.format_exc()))
+
+    threads = [threading.Thread(target=create, daemon=True) for _ in range(num_threads)]
+    for t in threads:
+        t.start()
+    for t in threads:
+        t.join(timeout=10)
+
+    passwords_seen = set()
+    for _ in range(num_threads):
+        status, res = results.get()
+        if status == "err":
+            pytest.fail(f"concurrent creation failed\n{res}")
+        passwords_seen.add(res["password"])
+
+    # all threads must see the same password hash
+    assert len(passwords_seen) == 1
+
+
 def test_50_concurrent_lookups_different_accounts(gencreds, dictproxy):
     num_threads = 50
     req_per_thread = 5

chatmaild/src/chatmaild/tests/test_expire.py

@@ -112,6 +112,43 @@ def test_report(mbox1, example_config):
     report_main(args)
 
 
+def test_report_mdir_filters_by_path(mbox1, example_config):
+    """Test that Report with mdir='cur' only counts messages in cur/ subdirectory."""
+    from chatmaild.fsreport import Report
+
+    now = datetime.utcnow().timestamp()
+
+    # Set password mtime to old enough so min_login_age check passes
+    password = Path(mbox1.basedir).joinpath("password")
+    old_time = now - 86400 * 10  # 10 days ago
+    os.utime(password, (old_time, old_time))
+
+    # Reload mailbox with updated mtime
+    from chatmaild.expire import MailboxStat
+
+    mbox = MailboxStat(mbox1.basedir)
+
+    # Report without mdir — should count all messages
+    rep_all = Report(now=now, min_login_age=1, mdir=None)
+    rep_all.process_mailbox_stat(mbox)
+    total_all = rep_all.message_buckets[0]
+
+    # Report with mdir='cur' — should only count cur/ messages
+    rep_cur = Report(now=now, min_login_age=1, mdir="cur")
+    rep_cur.process_mailbox_stat(mbox)
+    total_cur = rep_cur.message_buckets[0]
+
+    # Report with mdir='new' — should only count new/ messages
+    rep_new = Report(now=now, min_login_age=1, mdir="new")
+    rep_new.process_mailbox_stat(mbox)
+    total_new = rep_new.message_buckets[0]
+
+    # cur has 500-byte msg, new has 600-byte msg (from fill_mbox)
+    assert total_cur == 500
+    assert total_new == 600
+    assert total_all == 500 + 600
+
+
 def test_expiry_cli_basic(example_config, mbox1):
     args = (str(example_config._inipath),)
     expiry_main(args)

chatmaild/src/chatmaild/tests/test_filtermail_blackbox.py

@@ -47,6 +47,8 @@ def test_one_mail(
     make_config, make_popen, smtpserver, maildata, filtermail_mode, monkeypatch
 ):
     monkeypatch.setenv("PYTHONUNBUFFERED", "1")
+    # DKIM is tested by cmdeploy tests.
+    monkeypatch.setenv("FILTERMAIL_SKIP_DKIM", "1")
     smtp_inject_port = 20025
     if filtermail_mode == "outgoing":
         settings = dict(
@@ -64,6 +66,10 @@ def test_one_mail(
 
     popen = make_popen(["filtermail", path, filtermail_mode])
     line = popen.stderr.readline().strip()
+
+    # skip a warning that FILTERMAIL_SKIP_DKIM shouldn't be used in prod
+    if b"DKIM verification DISABLED!" in line:
+        line = popen.stderr.readline().strip()
     if b"loop" not in line:
         print(line.decode("ascii"), file=sys.stderr)
         pytest.fail("starting filtermail failed")

chatmaild/src/chatmaild/tests/test_metadata.py

@@ -314,6 +314,51 @@ def test_persistent_queue_items(tmp_path, testaddr, token):
     assert not queue_item < item2 and not item2 < queue_item
 
 
+def test_turn_credentials_exception_returns_N(notifier, metadata, monkeypatch):
+    """Test that turn_credentials() failure returns N\\n instead of crashing."""
+    import chatmaild.metadata
+
+    dictproxy = MetadataDictProxy(
+        notifier=notifier,
+        metadata=metadata,
+        turn_hostname="turn.example.org",
+    )
+
+    def mock_turn_credentials():
+        raise ConnectionRefusedError("socket not available")
+
+    monkeypatch.setattr(chatmaild.metadata, "turn_credentials", mock_turn_credentials)
+
+    transactions = {}
+    res = dictproxy.handle_dovecot_request(
+        "Lshared/0123/vendor/vendor.dovecot/pvt/server/vendor/deltachat/turn"
+        "\tuser@example.org",
+        transactions,
+    )
+    assert res == "N\n"
+
+
+def test_turn_credentials_success(notifier, metadata, monkeypatch):
+    """Test that valid turn_credentials() returns TURN URI."""
+    import chatmaild.metadata
+
+    dictproxy = MetadataDictProxy(
+        notifier=notifier,
+        metadata=metadata,
+        turn_hostname="turn.example.org",
+    )
+
+    monkeypatch.setattr(chatmaild.metadata, "turn_credentials", lambda: "user:pass")
+
+    transactions = {}
+    res = dictproxy.handle_dovecot_request(
+        "Lshared/0123/vendor/vendor.dovecot/pvt/server/vendor/deltachat/turn"
+        "\tuser@example.org",
+        transactions,
+    )
+    assert res == "Oturn.example.org:3478:user:pass\n"
+
+
 def test_iroh_relay(dictproxy):
     rfile = io.BytesIO(
         b"\n".join(

chatmaild/src/chatmaild/tests/test_metrics.py

@@ -1,24 +0,0 @@
-from chatmaild.metrics import main
-
-
-def test_main(tmp_path, capsys):
-    paths = []
-    for x in ("ci-asllkj", "ac_12l3kj", "qweqwe", "ci-l1k2j31l2k3"):
-        p = tmp_path.joinpath(x)
-        p.mkdir()
-        p.joinpath("cur").mkdir()
-        paths.append(p)
-
-    tmp_path.joinpath("nomailbox").mkdir()
-
-    main(tmp_path)
-    out, _ = capsys.readouterr()
-    d = {}
-    for line in out.split("\n"):
-        if line.strip() and not line.startswith("#"):
-            name, num = line.split()
-            d[name] = int(num)
-
-    assert d["accounts"] == 4
-    assert d["ci_accounts"] == 3
-    assert d["nonci_accounts"] == 1

chatmaild/src/chatmaild/tests/test_turnserver.py

@@ -0,0 +1,73 @@
+import socket
+import threading
+import time
+from unittest.mock import patch
+
+import pytest
+
+from chatmaild.turnserver import turn_credentials
+
+SOCKET_PATH = "/run/chatmail-turn/turn.socket"
+
+
+@pytest.fixture
+def turn_socket(tmp_path):
+    """Create a real Unix socket server at a temp path."""
+    sock_path = str(tmp_path / "turn.socket")
+    server = socket.socket(socket.AF_UNIX, socket.SOCK_STREAM)
+    server.bind(sock_path)
+    server.listen(1)
+    yield sock_path, server
+    server.close()
+
+
+def _call_turn_credentials(sock_path):
+    """Call turn_credentials but connect to sock_path instead of hardcoded path."""
+    original_connect = socket.socket.connect
+
+    def patched_connect(self, address):
+        if address == SOCKET_PATH:
+            address = sock_path
+        return original_connect(self, address)
+
+    with patch.object(socket.socket, "connect", patched_connect):
+        return turn_credentials()
+
+
+def test_turn_credentials_timeout(turn_socket):
+    """Server accepts but never responds — must raise socket.timeout."""
+    sock_path, server = turn_socket
+
+    def accept_and_hang():
+        conn, _ = server.accept()
+        time.sleep(30)
+        conn.close()
+
+    t = threading.Thread(target=accept_and_hang, daemon=True)
+    t.start()
+
+    with pytest.raises(socket.timeout):
+        _call_turn_credentials(sock_path)
+
+
+def test_turn_credentials_connection_refused(tmp_path):
+    """Socket file doesn't exist — must raise ConnectionRefusedError or FileNotFoundError."""
+    missing = str(tmp_path / "nonexistent.socket")
+    with pytest.raises((ConnectionRefusedError, FileNotFoundError)):
+        _call_turn_credentials(missing)
+
+
+def test_turn_credentials_success(turn_socket):
+    """Server responds with credentials — must return stripped string."""
+    sock_path, server = turn_socket
+
+    def respond():
+        conn, _ = server.accept()
+        conn.sendall(b"testuser:testpass\n")
+        conn.close()
+
+    t = threading.Thread(target=respond, daemon=True)
+    t.start()
+
+    result = _call_turn_credentials(sock_path)
+    assert result == "testuser:testpass"

chatmaild/src/chatmaild/turnserver.py

@@ -4,6 +4,7 @@ import socket
 
 def turn_credentials() -> str:
     with socket.socket(socket.AF_UNIX, socket.SOCK_STREAM) as client_socket:
+        client_socket.settimeout(5)
         client_socket.connect("/run/chatmail-turn/turn.socket")
         with client_socket.makefile("rb") as file:
             return file.readline().decode("utf-8").strip()

cmdeploy/pyproject.toml

@@ -20,6 +20,7 @@ dependencies = [
   "pytest-xdist", 
   "execnet",
   "imap_tools",
+  "deltachat-rpc-client",
 ]
 
 [project.scripts]

cmdeploy/src/cmdeploy/acmetool/__init__.py

@@ -67,7 +67,7 @@ class AcmetoolDeployer(Deployer):
         )
         files.template(
             src=importlib.resources.files(__package__).joinpath("desired.yaml.j2"),
-            dest=f"/var/lib/acme/desired/{self.domains[0]}", # 0 is mailhost TLD
+            dest=f"/var/lib/acme/desired/{self.domains[0]}",  # 0 is mailhost TLD
             user="root",
             group="root",
             mode="644",

cmdeploy/src/cmdeploy/acmetool/acmetool-redirector.service

@@ -3,7 +3,7 @@ Description=acmetool HTTP redirector
 
 [Service]
 Type=notify
-ExecStart=/usr/bin/acmetool redirector --service.uid=daemon
+ExecStart=/usr/bin/acmetool redirector --service.uid=daemon --bind=127.0.0.1:402
 Restart=always
 RestartSec=30
 

cmdeploy/src/cmdeploy/chatmail.zone.j2

@@ -1,32 +0,0 @@
-;
-; Required DNS entries for chatmail servers 
-;
-{% if A %}
-{{ mail_domain }}.                   A     {{ A }}
-{% endif %}
-{% if AAAA %}
-{{ mail_domain }}.                   AAAA  {{ AAAA }}
-{% endif %}
-{{ mail_domain }}.                   MX 10 {{ mail_domain }}.
-{% if strict_tls %}
-_mta-sts.{{ mail_domain }}.          TXT "v=STSv1; id={{ sts_id }}"
-mta-sts.{{ mail_domain }}.           CNAME {{ mail_domain }}.
-{% endif %}
-www.{{ mail_domain }}.               CNAME {{ mail_domain }}.
-{{ dkim_entry }}
-
-;
-; Recommended DNS entries for interoperability and security-hardening
-;
-{{ mail_domain }}.                   TXT "v=spf1 a ~all"
-_dmarc.{{ mail_domain }}.            TXT "v=DMARC1;p=reject;adkim=s;aspf=s"
-
-{% if acme_account_url %}
-{{ mail_domain }}.                   CAA 0 issue "letsencrypt.org;accounturi={{ acme_account_url }}"
-{% endif %}
-_adsp._domainkey.{{ mail_domain }}.  TXT "dkim=discardable"
-
-_submission._tcp.{{ mail_domain }}.  SRV 0 1 587 {{ mail_domain }}.
-_submissions._tcp.{{ mail_domain }}. SRV 0 1 465 {{ mail_domain }}.
-_imap._tcp.{{ mail_domain }}.        SRV 0 1 143 {{ mail_domain }}.
-_imaps._tcp.{{ mail_domain }}.       SRV 0 1 993 {{ mail_domain }}.

cmdeploy/src/cmdeploy/cmdeploy.py

@@ -5,12 +5,13 @@ along with command line option and subcommand parsing.
 
 import argparse
 import importlib.resources
-import importlib.util
 import os
 import pathlib
 import shutil
 import subprocess
 import sys
+import time
+from contextlib import contextmanager
 from pathlib import Path
 
 import pyinfra
@@ -19,7 +20,24 @@ from packaging import version
 from termcolor import colored
 
 from . import dns, remote
-from .sshexec import LocalExec, SSHExec
+from .lxc.cli import (  # noqa: F401
+    lxc_start_cmd,
+    lxc_start_cmd_options,
+    lxc_status_cmd,
+    lxc_status_cmd_options,
+    lxc_stop_cmd,
+    lxc_stop_cmd_options,
+    lxc_test_cmd,
+    lxc_test_cmd_options,
+)
+from .sshexec import (
+    LocalExec,
+    SSHExec,
+    resolve_host_from_ssh_config,
+    resolve_key_from_ssh_config,
+)
+from .util import build_chatmaild_sdist
+from .www import main as webdev_main
 
 #
 # cmdeploy sub commands and options
@@ -83,18 +101,21 @@ def run_cmd_options(parser):
         help="disable checks nslookup for dns",
     )
     add_ssh_host_option(parser)
+    add_ssh_config_option(parser)
 
 
 def run_cmd(args, out):
     """Deploy chatmail services on the remote server."""
 
     ssh_host = args.ssh_host if args.ssh_host else args.config.mail_domain
-    sshexec = get_sshexec(ssh_host)
+    sshexec = get_sshexec(ssh_host, ssh_config=args.ssh_config)
     require_iroh = args.config.enable_iroh_relay
     strict_tls = args.config.tls_cert_mode == "acme"
     if not args.dns_check_disabled:
         remote_data = dns.get_initial_remote_data(sshexec, args.config.mail_domain)
-        if not dns.check_initial_remote_data(remote_data, strict_tls=strict_tls, print=out.red):
+        if not dns.check_initial_remote_data(
+            remote_data, strict_tls=strict_tls, print=out.red
+        ):
             return 1
 
     env = os.environ.copy()
@@ -102,6 +123,9 @@ def run_cmd(args, out):
     env["CHATMAIL_WEBSITE_ONLY"] = "True" if args.website_only else ""
     env["CHATMAIL_DISABLE_MAIL"] = "True" if args.disable_mail else ""
     env["CHATMAIL_REQUIRE_IROH"] = "True" if require_iroh else ""
+
+    if not args.website_only:
+        build_chatmaild_sdist()
     if not args.dns_check_disabled:
         env["CHATMAIL_ADDR_V4"] = remote_data.get("A") or ""
         env["CHATMAIL_ADDR_V6"] = remote_data.get("AAAA") or ""
@@ -109,6 +133,18 @@ def run_cmd(args, out):
     pyinf = "pyinfra --dry" if args.dry_run else "pyinfra"
 
     cmd = f"{pyinf} --ssh-user root {ssh_host} {deploy_path} -y"
+    ssh_config = args.ssh_config
+    if ssh_config:
+        ssh_config = str(Path(ssh_config).resolve())
+
+        # Use pyinfra's native SSH data keys to configure the connection directly
+        # rather than relying on paramiko config parsing (see also sshexec.py)
+        ip = resolve_host_from_ssh_config(ssh_host, ssh_config)
+        key = resolve_key_from_ssh_config(ssh_host, ssh_config)
+        data_args = f"--data ssh_hostname={ip} --data ssh_known_hosts_file=/dev/null"
+        if key:
+            data_args += f" --data ssh_key={key}"
+        cmd = f"{pyinf} --ssh-user root {ssh_host} {deploy_path} -y {data_args}"
     if ssh_host in ["localhost", "@docker"]:
         if ssh_host == "@docker":
             env["CHATMAIL_NOPORTCHECK"] = "True"
@@ -120,24 +156,22 @@ def run_cmd(args, out):
         return 1
 
     try:
-        retcode = out.check_call(cmd, env=env)
+        out.check_call(cmd, env=env)
         if args.website_only:
-            if retcode == 0:
-                out.green("Website deployment completed.")
-            else:
-                out.red("Website deployment failed.")
-        elif retcode == 0:
-            out.green("Deploy completed, call `cmdeploy dns` next.")
-        elif not args.dns_check_disabled and strict_tls and not remote_data["acme_account_url"]:
+            out.green("Website deployment completed.")
+        elif (
+            not args.dns_check_disabled
+            and strict_tls
+            and not remote_data["acme_account_url"]
+        ):
             out.red("Deploy completed but letsencrypt not configured")
             out.red("Run 'cmdeploy run' again")
-            retcode = 0
         else:
-            out.red("Deploy failed")
+            out.green("Deploy completed, call `cmdeploy dns` next.")
+        return 0
     except subprocess.CalledProcessError:
         out.red("Deploy failed")
-        retcode = 1
-    return retcode
+        return 1
 
 
 def dns_cmd_options(parser):
@@ -146,15 +180,16 @@ def dns_cmd_options(parser):
         dest="zonefile",
         type=pathlib.Path,
         default=None,
-        help="write out a zonefile",
+        help="write DNS records in standard BIND format to the given file",
     )
     add_ssh_host_option(parser)
+    add_ssh_config_option(parser)
 
 
 def dns_cmd(args, out):
     """Check DNS entries and optionally generate dns zone file."""
     ssh_host = args.ssh_host if args.ssh_host else args.config.mail_domain
-    sshexec = get_sshexec(ssh_host, verbose=args.verbose)
+    sshexec = get_sshexec(ssh_host, verbose=args.verbose, ssh_config=args.ssh_config)
     tls_cert_mode = args.config.tls_cert_mode
     strict_tls = tls_cert_mode == "acme"
     remote_data = dns.get_initial_remote_data(sshexec, args.config.mail_domain)
@@ -185,13 +220,14 @@ def dns_cmd(args, out):
 
 def status_cmd_options(parser):
     add_ssh_host_option(parser)
+    add_ssh_config_option(parser)
 
 
 def status_cmd(args, out):
     """Display status for online chatmail instance."""
 
     ssh_host = args.ssh_host if args.ssh_host else args.config.mail_domain
-    sshexec = get_sshexec(ssh_host, verbose=args.verbose)
+    sshexec = get_sshexec(ssh_host, verbose=args.verbose, ssh_config=args.ssh_config)
 
     out.green(f"chatmail domain: {args.config.mail_domain}")
     if args.config.privacy_mail:
@@ -210,17 +246,15 @@ def test_cmd_options(parser):
         action="store_true",
         help="also run slow tests",
     )
+    add_ssh_host_option(parser)
+    add_ssh_config_option(parser)
 
 
 def test_cmd(args, out):
-    """Run local and online tests for chatmail deployment.
+    """Run local and online tests for chatmail deployment."""
 
-    This will automatically pip-install 'deltachat' if it's not available.
-    """
-
-    x = importlib.util.find_spec("deltachat")
-    if x is None:
-        out.check_call(f"{sys.executable} -m pip install deltachat")
+    env = os.environ.copy()
+    env["CHATMAIL_INI"] = str(args.inipath.resolve())
 
     pytest_path = shutil.which("pytest")
     pytest_args = [
@@ -234,7 +268,11 @@ def test_cmd(args, out):
     ]
     if args.slow:
         pytest_args.append("--slow")
-    ret = out.run_ret(pytest_args)
+    if args.ssh_host:
+        pytest_args.extend(["--ssh-host", args.ssh_host])
+    if args.ssh_config:
+        pytest_args.extend(["--ssh-config", str(Path(args.ssh_config).resolve())])
+    ret = out.run_ret(pytest_args, env=env)
     return ret
 
 
@@ -285,9 +323,7 @@ def bench_cmd(args, out):
 
 def webdev_cmd(args, out):
     """Run local web development loop for static web pages."""
-    from .www import main
-
-    main()
+    webdev_main()
 
 
 #
@@ -296,17 +332,44 @@ def webdev_cmd(args, out):
 
 
 class Out:
-    """Convenience output printer providing coloring."""
+    """Convenience output printer providing coloring and section formatting."""
+
+    SECTION_WIDTH = 72
+
+    def __init__(self):
+        self.section_timings = []
 
     def red(self, msg, file=sys.stderr):
-        print(colored(msg, "red"), file=file)
+        print(colored(msg, "red"), file=file, flush=True)
 
     def green(self, msg, file=sys.stderr):
-        print(colored(msg, "green"), file=file)
+        print(colored(msg, "green"), file=file, flush=True)
+
+    def print(self, msg="", **kwargs):
+        """Print to stdout with automatic flush."""
+        print(msg, flush=True, **kwargs)
+
+    @contextmanager
+    def section(self, title):
+        """Context manager that prints a section header and records elapsed time."""
+        bar = "\u2501" * (self.SECTION_WIDTH - len(title) - 5)
+        self.green(f"\u2501\u2501\u2501 {title} {bar}")
+        t0 = time.time()
+        yield
+        elapsed = time.time() - t0
+        self.section_timings.append((title, elapsed))
+        self.print(f"{'':>{self.SECTION_WIDTH - 10}}({elapsed:.1f}s)")
+        self.print()
+
+    def section_line(self, title):
+        """Print a section header without timing."""
+        bar = "\u2501" * (self.SECTION_WIDTH - len(title) - 5)
+        self.green(f"\u2501\u2501\u2501 {title} {bar}")
+        self.print()
 
     def __call__(self, msg, red=False, green=False, file=sys.stdout):
         color = "red" if red else ("green" if green else None)
-        print(colored(msg, color), file=file)
+        print(colored(msg, color), file=file, flush=True)
 
     def check_call(self, arg, env=None, quiet=False):
         if not quiet:
@@ -330,6 +393,16 @@ def add_ssh_host_option(parser):
     )
 
 
+def add_ssh_config_option(parser):
+    parser.add_argument(
+        "--ssh-config",
+        dest="ssh_config",
+        type=Path,
+        default=None,
+        help="Path to an SSH config file (e.g. lxconfigs/ssh-config).",
+    )
+
+
 def add_config_option(parser):
     parser.add_argument(
         "--config",
@@ -339,6 +412,7 @@ def add_config_option(parser):
         type=Path,
         help="path to the chatmail.ini file",
     )
+
     parser.add_argument(
         "--verbose",
         "-v",
@@ -349,15 +423,16 @@ def add_config_option(parser):
     )
 
 
-def add_subcommand(subparsers, func):
+def add_subcommand(subparsers, func, add_config=True):
     name = func.__name__
     assert name.endswith("_cmd")
-    name = name[:-4]
+    name = name[:-4].replace("_", "-")
     doc = func.__doc__.strip()
     help = doc.split("\n")[0].strip(".")
     p = subparsers.add_parser(name, description=doc, help=help)
     p.set_defaults(func=func)
-    add_config_option(p)
+    if add_config:
+        add_config_option(p)
     return p
 
 
@@ -371,13 +446,15 @@ def get_parser():
     """Return an ArgumentParser for the 'cmdeploy' CLI"""
 
     parser = argparse.ArgumentParser(description=description.strip())
+    parser.set_defaults(func=None, inipath=None)
     subparsers = parser.add_subparsers(title="subcommands")
 
     # find all subcommands in the module namespace
     glob = globals()
     for name, func in glob.items():
         if name.endswith("_cmd"):
-            subparser = add_subcommand(subparsers, func)
+            needs_config = not name.startswith("lxc_")
+            subparser = add_subcommand(subparsers, func, add_config=needs_config)
             addopts = glob.get(name + "_options")
             if addopts is not None:
                 addopts(subparser)
@@ -385,26 +462,27 @@ def get_parser():
     return parser
 
 
-def get_sshexec(ssh_host: str, verbose=True):
+def get_sshexec(ssh_host: str, verbose=True, ssh_config=None):
     if ssh_host in ["localhost", "@local"]:
         return LocalExec(verbose, docker=False)
     elif ssh_host == "@docker":
         return LocalExec(verbose, docker=True)
     if verbose:
         print(f"[ssh] login to {ssh_host}")
-    return SSHExec(ssh_host, verbose=verbose)
+    return SSHExec(ssh_host, verbose=verbose, ssh_config=ssh_config)
 
 
 def main(args=None):
     """Provide main entry point for 'cmdeploy' CLI invocation."""
     parser = get_parser()
     args = parser.parse_args(args=args)
-    if not hasattr(args, "func"):
+    if args.func is None:
         return parser.parse_args(["-h"])
 
     out = Out()
     kwargs = {}
-    if args.func.__name__ not in ("init_cmd", "fmt_cmd"):
+
+    if args.inipath is not None and args.func.__name__ not in ("init_cmd", "fmt_cmd"):
         if not args.inipath.exists():
             out.red(f"expecting {args.inipath} to exist, run init first?")
             raise SystemExit(1)

cmdeploy/src/cmdeploy/deployers.py

@@ -3,24 +3,21 @@ Chat Mail pyinfra deploy.
 """
 
 import os
-import shutil
-import subprocess
-import sys
-from io import StringIO
+from io import BytesIO, StringIO
 from pathlib import Path
 
 from chatmaild.config import read_config
 from pyinfra import facts, host, logger
-from pyinfra.facts import hardware
 from pyinfra.api import FactBase
+from pyinfra.facts import hardware
 from pyinfra.facts.files import Sha256File
 from pyinfra.facts.systemd import SystemdEnabled
 from pyinfra.operations import apt, files, pip, server, systemd
 
 from cmdeploy.cmdeploy import Out
+from cmdeploy.util import get_chatmaild_sdist, get_version_string
 
 from .acmetool import AcmetoolDeployer
-from .external.deployer import ExternalTlsDeployer
 from .basedeploy import (
     Deployer,
     Deployment,
@@ -30,6 +27,7 @@ from .basedeploy import (
     has_systemd,
 )
 from .dovecot.deployer import DovecotDeployer
+from .external.deployer import ExternalTlsDeployer
 from .filtermail.deployer import FiltermailDeployer
 from .mtail.deployer import MtailDeployer
 from .nginx.deployer import NginxDeployer
@@ -54,20 +52,6 @@ class Port(FactBase):
         return output[0]
 
 
-def _build_chatmaild(dist_dir) -> None:
-    dist_dir = Path(dist_dir).resolve()
-    if dist_dir.exists():
-        shutil.rmtree(dist_dir)
-    dist_dir.mkdir()
-    subprocess.check_output(
-        [sys.executable, "-m", "build", "-n"]
-        + ["--sdist", "chatmaild", "--outdir", str(dist_dir)]
-    )
-    entries = list(dist_dir.iterdir())
-    assert len(entries) == 1
-    return entries[0]
-
-
 def remove_legacy_artifacts():
     if not has_systemd():
         return
@@ -83,7 +67,7 @@ def remove_legacy_artifacts():
 
 def _install_remote_venv_with_chatmaild() -> None:
     remove_legacy_artifacts()
-    dist_file = _build_chatmaild(dist_dir=Path("chatmaild/dist"))
+    dist_file = get_chatmaild_sdist()
     remote_base_dir = "/usr/local/lib/chatmaild"
     remote_dist_file = f"{remote_base_dir}/dist/{dist_file.name}"
     remote_venv_dir = f"{remote_base_dir}/venv"
@@ -123,7 +107,6 @@ def _install_remote_venv_with_chatmaild() -> None:
 
 def _configure_remote_venv_with_chatmaild(config) -> None:
     remote_base_dir = "/usr/local/lib/chatmaild"
-    remote_venv_dir = f"{remote_base_dir}/venv"
     remote_chatmail_inipath = f"{remote_base_dir}/chatmail.ini"
     root_owned = dict(user="root", group="root", mode="644")
 
@@ -134,16 +117,13 @@ def _configure_remote_venv_with_chatmaild(config) -> None:
         **root_owned,
     )
 
-    files.template(
-        src=get_resource("metrics.cron.j2"),
-        dest="/etc/cron.d/chatmail-metrics",
-        user="root",
-        group="root",
-        mode="644",
-        config={
-            "mailboxes_dir": config.mailboxes_dir,
-            "execpath": f"{remote_venv_dir}/bin/chatmail-metrics",
-        },
+    files.file(
+        path="/etc/cron.d/chatmail-metrics",
+        present=False,
+    )
+    files.file(
+        path="/var/www/html/metrics",
+        present=False,
     )
 
 
@@ -271,9 +251,18 @@ class WebsiteDeployer(Deployer):
             # if www_folder is a hugo page, build it
             if build_dir:
                 www_path = build_webpages(src_dir, build_dir, self.config)
+                if www_path is None:
+                    logger.warning("Web page build failed, skipping website deployment")
+                    return
             # if it is not a hugo page, upload it as is
-            files.rsync(
-                f"{www_path}/", "/var/www/html", flags=["-avz", "--chown=www-data"]
+            # pyinfra files.rsync (experimental) causes problems with ssh-config configuration
+            # the stable files.sync should do
+            files.sync(
+                src=str(www_path),
+                dest="/var/www/html",
+                user="www-data",
+                group="www-data",
+                delete=True,
             )
 
 
@@ -337,12 +326,12 @@ class TurnDeployer(Deployer):
     def install(self):
         (url, sha256sum) = {
             "x86_64": (
-                "https://github.com/chatmail/chatmail-turn/releases/download/v0.3/chatmail-turn-x86_64-linux",
-                "841e527c15fdc2940b0469e206188ea8f0af48533be12ecb8098520f813d41e4",
+                "https://github.com/chatmail/chatmail-turn/releases/download/v0.4/chatmail-turn-x86_64-linux",
+                "1ec1f5c50122165e858a5a91bcba9037a28aa8cb8b64b8db570aa457c6141a8a",
             ),
             "aarch64": (
-                "https://github.com/chatmail/chatmail-turn/releases/download/v0.3/chatmail-turn-aarch64-linux",
-                "a5fc2d06d937b56a34e098d2cd72a82d3e89967518d159bf246dc69b65e81b42",
+                "https://github.com/chatmail/chatmail-turn/releases/download/v0.4/chatmail-turn-aarch64-linux",
+                "0fb3e792419494e21ecad536464929dba706bb2c88884ed8f1788141d26fc756",
             ),
         }[host.get_fact(facts.server.Arch)]
 
@@ -479,6 +468,14 @@ class ChatmailDeployer(Deployer):
         self.mail_domain = mail_domain
 
     def install(self):
+        files.put(
+            name="Disable installing recommended packages globally",
+            src=BytesIO(b'APT::Install-Recommends "0";\n'),
+            dest="/etc/apt/apt.conf.d/99no-recommends",
+            user="root",
+            group="root",
+            mode="644",
+        )
         apt.update(name="apt update", cache_time=24 * 3600)
         apt.upgrade(name="upgrade apt packages", auto_remove=True)
 
@@ -525,17 +522,9 @@ class FcgiwrapDeployer(Deployer):
 
 class GithashDeployer(Deployer):
     def activate(self):
-        try:
-            git_hash = subprocess.check_output(["git", "rev-parse", "HEAD"]).decode()
-        except Exception:
-            git_hash = "unknown\n"
-        try:
-            git_diff = subprocess.check_output(["git", "diff"]).decode()
-        except Exception:
-            git_diff = ""
         files.put(
             name="Upload chatmail relay git commit hash",
-            src=StringIO(git_hash + git_diff),
+            src=StringIO(get_version_string()),
             dest="/etc/chatmail-version",
             mode="700",
         )
@@ -579,11 +568,17 @@ def deploy_chatmail(config_path: Path, disable_mail: bool, website_only: bool) -
         )
 
     # Check if mtail_address interface is available (if configured)
-    if config.mtail_address and config.mtail_address not in ('127.0.0.1', '::1', 'localhost'):
+    if config.mtail_address and config.mtail_address not in (
+        "127.0.0.1",
+        "::1",
+        "localhost",
+    ):
         ipv4_addrs = host.get_fact(hardware.Ipv4Addrs)
         all_addresses = [addr for addrs in ipv4_addrs.values() for addr in addrs]
         if config.mtail_address not in all_addresses:
-            Out().red(f"Deploy failed: mtail_address {config.mtail_address} is not available (VPN up?).\n")
+            Out().red(
+                f"Deploy failed: mtail_address {config.mtail_address} is not available (VPN up?).\n"
+            )
             exit(1)
 
     if not os.environ.get("CHATMAIL_NOPORTCHECK"):
@@ -592,9 +587,13 @@ def deploy_chatmail(config_path: Path, disable_mail: bool, website_only: bool) -
             ("unbound", 53),
         ]
         if config.tls_cert_mode == "acme":
-            port_services.append(("acmetool", 80))
+            port_services.append(("acmetool", 402))
         port_services += [
             (["imap-login", "dovecot"], 143),
+            # acmetool previously listened on port 80,
+            # so don't complain during upgrade that moved it to port 402
+            # and gave the port to nginx.
+            (["acmetool", "nginx"], 80),
             ("nginx", 443),
             (["master", "smtpd"], 465),
             (["master", "smtpd"], 587),

cmdeploy/src/cmdeploy/dns.py

@@ -1,11 +1,26 @@
 import datetime
-import importlib
-
-from jinja2 import Template
 
 from . import remote
 
 
+def parse_zone_records(text):
+    """Yield ``(name, ttl, rtype, rdata)`` from standard BIND-format text.
+
+    Skips comment lines (starting with ``;``) and blank lines.
+    Each record line must have the format ``name TTL IN type rdata``.
+    """
+    for raw_line in text.splitlines():
+        line = raw_line.strip()
+        if not line or line.startswith(";"):
+            continue
+        try:
+            name, ttl, _in, rtype, rdata = line.split(None, 4)
+        except ValueError:
+            raise ValueError(f"Bad zone record line: {line!r}") from None
+        name = name.rstrip(".")
+        yield name, ttl, rtype.upper(), rdata
+
+
 def get_initial_remote_data(sshexec, mail_domain):
     return sshexec.logged(
         call=remote.rdns.perform_initial_checks, kwargs=dict(mail_domain=mail_domain)
@@ -31,13 +46,36 @@ def get_filled_zone_file(remote_data):
     if not sts_id:
         remote_data["sts_id"] = datetime.datetime.now().strftime("%Y%m%d%H%M")
 
-    template = importlib.resources.files(__package__).joinpath("chatmail.zone.j2")
-    content = template.read_text()
-    zonefile = Template(content).render(**remote_data)
-    lines = [x.strip() for x in zonefile.split("\n") if x.strip()]
+    d = remote_data["mail_domain"]
+    lines = ["; Required DNS entries"]
+    if remote_data.get("A"):
+        lines.append(f"{d}.  3600  IN  A  {remote_data['A']}")
+    if remote_data.get("AAAA"):
+        lines.append(f"{d}.  3600  IN  AAAA  {remote_data['AAAA']}")
+    lines.append(f"{d}.  3600  IN  MX  10 {d}.")
+    if remote_data.get("strict_tls"):
+        lines.append(
+            f'_mta-sts.{d}.  3600  IN  TXT  "v=STSv1; id={remote_data["sts_id"]}"'
+        )
+        lines.append(f"mta-sts.{d}.  3600  IN  CNAME  {d}.")
+    lines.append(f"www.{d}.  3600  IN  CNAME  {d}.")
+    lines.append(remote_data["dkim_entry"])
     lines.append("")
-    zonefile = "\n".join(lines)
-    return zonefile
+    lines.append("; Recommended DNS entries")
+    lines.append(f'{d}.  3600  IN  TXT  "v=spf1 a ~all"')
+    lines.append(f'_dmarc.{d}.  3600  IN  TXT  "v=DMARC1;p=reject;adkim=s;aspf=s"')
+    if remote_data.get("acme_account_url"):
+        lines.append(
+            f"{d}.  3600  IN  CAA  0 issue"
+            f' "letsencrypt.org;accounturi={remote_data["acme_account_url"]}"'
+        )
+    lines.append(f'_adsp._domainkey.{d}.  3600  IN  TXT  "dkim=discardable"')
+    lines.append(f"_submission._tcp.{d}.  3600  IN  SRV  0 1 587 {d}.")
+    lines.append(f"_submissions._tcp.{d}.  3600  IN  SRV  0 1 465 {d}.")
+    lines.append(f"_imap._tcp.{d}.  3600  IN  SRV  0 1 143 {d}.")
+    lines.append(f"_imaps._tcp.{d}.  3600  IN  SRV  0 1 993 {d}.")
+    lines.append("")
+    return "\n".join(lines)
 
 
 def check_full_zone(sshexec, remote_data, out, zonefile) -> int:

cmdeploy/src/cmdeploy/dovecot/deployer.py

@@ -1,4 +1,5 @@
 import os
+import urllib.request
 
 from chatmaild.config import Config
 from pyinfra import host
@@ -41,7 +42,9 @@ class DovecotDeployer(Deployer):
         restart = False if self.disable_mail else self.need_restart
 
         systemd.service(
-            name="Disable dovecot for now" if self.disable_mail else "Start and enable Dovecot",
+            name="Disable dovecot for now"
+            if self.disable_mail
+            else "Start and enable Dovecot",
             service="dovecot.service",
             running=False if self.disable_mail else True,
             enabled=False if self.disable_mail else True,
@@ -51,10 +54,21 @@ class DovecotDeployer(Deployer):
         self.need_restart = False
 
 
+def _pick_url(primary, fallback):
+    try:
+        req = urllib.request.Request(primary, method="HEAD")
+        urllib.request.urlopen(req, timeout=10)
+        return primary
+    except Exception:
+        return fallback
+
+
 def _install_dovecot_package(package: str, arch: str):
     arch = "amd64" if arch == "x86_64" else arch
     arch = "arm64" if arch == "aarch64" else arch
-    url = f"https://download.delta.chat/dovecot/dovecot-{package}_2.3.21%2Bdfsg1-3_{arch}.deb"
+    primary_url = f"https://download.delta.chat/dovecot/dovecot-{package}_2.3.21%2Bdfsg1-3_{arch}.deb"
+    fallback_url = f"https://github.com/chatmail/dovecot/releases/download/upstream%2F2.3.21%2Bdfsg1/dovecot-{package}_2.3.21%2Bdfsg1-3_{arch}.deb"
+    url = _pick_url(primary_url, fallback_url)
     deb_filename = "/root/" + url.split("/")[-1]
 
     match (package, arch):

cmdeploy/src/cmdeploy/external/deployer.py

@@ -1,4 +1,8 @@
-from pyinfra.operations import files, server, systemd
+import io
+
+from pyinfra import host
+from pyinfra.facts.files import File
+from pyinfra.operations import files, systemd
 
 from cmdeploy.basedeploy import Deployer, get_resource
 
@@ -17,19 +21,18 @@ class ExternalTlsDeployer(Deployer):
         self.key_path = key_path
 
     def configure(self):
-        server.shell(
-            name="Verify external TLS certificate and key exist",
-            commands=[
-                f"test -f {self.cert_path} && test -f {self.key_path}",
-            ],
-        )
+        # Verify cert and key exist on the remote host using pyinfra facts.
+        for path in (self.cert_path, self.key_path):
+            info = host.get_fact(File, path=path)
+            if info is None:
+                raise Exception(f"External TLS file not found on server: {path}")
 
         # Deploy the .path unit (templated with the cert path).
+        # pkg=__package__ is required here because the resource files
+        # live in cmdeploy.external, not the default cmdeploy package.
         source = get_resource("tls-cert-reload.path.f", pkg=__package__)
         content = source.read_text().format(cert_path=self.cert_path).encode()
 
-        import io
-
         path_unit = files.put(
             name="Upload tls-cert-reload.path",
             src=io.BytesIO(content),
@@ -60,10 +63,5 @@ class ExternalTlsDeployer(Deployer):
             restarted=self.need_restart,
             daemon_reload=self.need_restart,
         )
-        # Always trigger a reload so services pick up the current cert.
-        # The path unit handles future changes via inotify.
-        server.shell(
-            name="Reload TLS services for current certificate",
-            commands=["systemctl start tls-cert-reload.service"],
-        )
-
+        # No explicit reload needed here: dovecot/nginx read the cert
+        # on startup, and the .path watcher handles live changes.

cmdeploy/src/cmdeploy/external/tls-cert-reload.path.f

@@ -1,6 +1,10 @@
 # Watch the TLS certificate file for changes.
 # When the cert is updated (e.g. renewed by an external process),
-# this triggers tls-cert-reload.service to restart the affected services.
+# this triggers tls-cert-reload.service to reload the affected services.
+#
+# NOTE: changes to the certificates are not detected if they cross bind-mount boundaries. 
+# After cert renewal, you must then trigger the reload explicitly:
+#   systemctl start tls-cert-reload.service
 [Unit]
 Description=Watch TLS certificate for changes
 

cmdeploy/src/cmdeploy/external/tls-cert-reload.service

@@ -11,5 +11,5 @@ Description=Reload TLS services after certificate change
 
 [Service]
 Type=oneshot
-ExecStart=/bin/systemctl reload dovecot
-ExecStart=/bin/systemctl reload nginx
+ExecStart=/bin/systemctl try-reload-or-restart dovecot
+ExecStart=/bin/systemctl try-reload-or-restart nginx

cmdeploy/src/cmdeploy/filtermail/deployer.py

@@ -14,10 +14,10 @@ class FiltermailDeployer(Deployer):
 
     def install(self):
         arch = host.get_fact(facts.server.Arch)
-        url = f"https://github.com/chatmail/filtermail/releases/download/v0.3.0/filtermail-{arch}"
+        url = f"https://github.com/chatmail/filtermail/releases/download/v0.5.2/filtermail-{arch}"
         sha256sum = {
-            "x86_64": "f14a31323ae2dad3b59d3fdafcde507521da2f951a9478cd1f2fe2b4463df71d",
-            "aarch64": "933770d75046c4fd7084ce8d43f905f8748333426ad839154f0fc654755ef09f",
+            "x86_64": "ce24ca0075aa445510291d775fb3aea8f4411818c7b885ae51a0fe18c5f789ce",
+            "aarch64": "c5d783eefa5332db3d97a0e6a23917d72849e3eb45da3d16ce908a9b4e5a797d",
         }[arch]
         self.need_restart |= files.download(
             name="Download filtermail",

cmdeploy/src/cmdeploy/lxc/cli.py

@@ -0,0 +1,558 @@
+"""lxc-start/stop/status/test subcommands for testing with local containers."""
+
+import os
+import subprocess
+import threading
+import time
+
+from ..util import (
+    collapse,
+    get_git_hash,
+    get_version_string,
+    shell,
+)
+from .incus import Incus, RelayContainer
+
+RELAY_NAMES = ("test0", "test1")
+
+
+# -------------------------------------------------------------------
+# lxc-start
+# -------------------------------------------------------------------
+
+
+def lxc_start_cmd_options(parser):
+    _add_name_args(
+        parser,
+        help_text="User relay name(s) to create (default: test0).",
+    )
+    parser.add_argument(
+        "--ipv4-only",
+        dest="ipv4_only",
+        action="store_true",
+        help="Create an IPv4-only container.",
+    )
+    parser.add_argument(
+        "--run",
+        action="store_true",
+        help="Run 'cmdeploy run' on each container after starting it.",
+    )
+
+
+def lxc_start_cmd(args, out):
+    """Create/Ensure and start LXC relay and DNS containers."""
+    ix = Incus()
+    out.green("Ensuring DNS container (ns-localchat) ...")
+    dns_ct = ix.get_dns_container()
+    dns_ct.ensure()
+    if not ix.find_dns_image():
+        with out.section("LXC: publishing DNS image"):
+            dns_ct.publish_as_dns_image()
+    out.print(f"  DNS container IP: {dns_ct.ipv4}")
+
+    names = args.names if args.names else RELAY_NAMES
+    relays = list(ix.get_container(n) for n in names)
+    for ct in relays:
+        out.green(f"Ensuring container {ct.name!r} ({ct.domain}) ...")
+        ct.ensure()
+        ip = ct.ipv4
+
+        out.print("  Configuring container hostname ...")
+        ct.configure_hosts(ip)
+
+        out.print(f"  Writing {ct.ini.name} ...")
+        ct.write_ini(disable_ipv6=args.ipv4_only)
+        out.print(f"  Config: {ct.ini}")
+        if args.ipv4_only:
+            ct.disable_ipv6()
+            ipv6 = None
+        else:
+            output = ct.bash(
+                "ip -6 addr show scope global -deprecated"
+                " | grep -oP '(?<=inet6 )[^/]+'",
+                check=False,
+            )
+            ipv6 = output.strip() if output else None
+        out.print(f"  {_format_addrs(ip, ipv6)}")
+
+        out.green(f"  Container {ct.name!r} ready: {ct.domain} -> {ip}")
+        out.print()
+
+    # Reset DNS zones only for the containers we just started
+    started_cnames = {ct.name for ct in relays}
+    managed = ix.list_managed()
+    started = [c for c in managed if c["name"] in started_cnames]
+
+    if started:
+        out.print(
+            f"Resetting DNS zones for {len(started)}"
+            " domain(s) (A + AAAA records) ..."
+        )
+        dns_ct.reset_dns_records(dns_ct.ipv4, started)
+
+        for ct in relays:
+            if ct.name in started_cnames:
+                out.print(f"  Configuring and testing DNS in {ct.name} ...")
+                ct.configure_dns(dns_ct.ipv4)
+                if not ct.check_dns():
+                    out.red(
+                        f"  DNS check failed for {ct.name}"
+                        ": cannot resolve external hosts"
+                    )
+                    return 1
+
+    # Generate the unified SSH config
+    out.green("Writing ssh-config ...")
+    ssh_cfg = ix.write_ssh_config()
+    out.print(f"  {ssh_cfg}")
+
+    # Verify SSH via the generated config
+    for ct in relays:
+        out.print(f"  Verifying SSH to {ct.name} via ssh-config ...")
+        if ct.verify_ssh(ssh_cfg):
+            out.print(f"  SSH OK: ssh -F lxconfigs/ssh-config {ct.domain}")
+        else:
+            out.red(f"  WARNING: SSH verification failed for {ct.name}")
+
+    # Print integration suggestions
+    ssh_cfg = ix.ssh_config_path
+    if not ix.check_ssh_include():
+        out.green(
+            "\n  (Optional) To use containers from any SSH client, add to ~/.ssh/config:"
+        )
+        out.green(f"      Include {ssh_cfg}")
+
+    # Optionally run cmdeploy run on each relay
+    if args.run:
+        for ct in relays:
+            with out.section(f"cmdeploy run: {ct.sname} ({ct.domain})"):
+                ret = _run_cmdeploy("run", ct, ix, out, extra=["--skip-dns-check"])
+                if ret:
+                    out.red(f"Deploy to {ct.sname} failed (exit {ret})")
+                    return ret
+
+
+# -------------------------------------------------------------------
+# lxc-stop
+# -------------------------------------------------------------------
+
+
+def lxc_stop_cmd_options(parser):
+    parser.add_argument(
+        "--destroy",
+        action="store_true",
+        help="Delete containers and their config files after stopping.",
+    )
+    parser.add_argument(
+        "--destroy-all",
+        dest="destroy_all",
+        action="store_true",
+        help="Like --destroy, but also remove the ns-localchat DNS container.",
+    )
+    _add_name_args(
+        parser,
+        help_text="Container name(s) to stop (default: test0 + test1).",
+    )
+
+
+def lxc_stop_cmd(args, out):
+    """Stop (and optionally destroy) local LXC relay containers."""
+    ix = Incus()
+    names = args.names or RELAY_NAMES
+    destroy = args.destroy or args.destroy_all
+
+    for ct in map(ix.get_container, names):
+        if destroy:
+            out.green(f"Destroying container {ct.name!r} ...")
+            ct.destroy()
+            if hasattr(ct, "image_alias"):
+                out.green(f"  Deleting cached image {ct.image_alias!r} ...")
+                ix.run(["image", "delete", ct.image_alias], check=False)
+        else:
+            out.green(f"Stopping container {ct.name!r} ...")
+            ct.stop(force=True)
+
+    if args.destroy_all:
+        dns_ct = ix.get_dns_container()
+        out.green(f"Destroying DNS container {dns_ct.name!r} ...")
+        dns_ct.destroy()
+        ix.delete_images()
+
+    if destroy:
+        ix.write_ssh_config()
+        out.green("LXC containers destroyed.")
+    else:
+        out.green("LXC containers stopped.")
+
+
+# -------------------------------------------------------------------
+# lxc-test
+# -------------------------------------------------------------------
+
+
+def lxc_test_cmd_options(parser):
+    parser.add_argument(
+        "--one",
+        action="store_true",
+        help="Only deploy and test against test0 (skip test1).",
+    )
+
+
+def lxc_test_cmd(args, out):
+    """Run full LXC pipeline: start, deploy, DNS, zone files, and tests.
+
+    All commands run directly on the host using
+    ``--ssh-config lxconfigs/ssh-config`` for SSH access.
+    """
+    ix = Incus()
+    t_total = time.time()
+    relay_names = list(RELAY_NAMES)
+    if args.one:
+        relay_names = relay_names[:1]
+
+    local_hash = get_git_hash()
+
+    # Per-relay: start containers, then deploy in parallel.
+    ipv4_only_flags = {RELAY_NAMES[0]: False, RELAY_NAMES[1]: True}
+
+    # Phase 1: start all containers (sequential, fast)
+    for ct in map(ix.get_container, relay_names):
+        name = ct.sname
+        ipv4_only = ipv4_only_flags.get(name, False)
+        label = "IPv4-only" if ipv4_only else "dual-stack"
+
+        with out.section(f"LXC: lxc-start {name} ({label})"):
+            args.names = [name]
+            args.ipv4_only = ipv4_only
+            args.run = False
+            ret = lxc_start_cmd(args, out)
+            if ret:
+                return ret
+
+    # Phase 2: deploy all relays in parallel
+    to_deploy = []
+    for ct in map(ix.get_container, relay_names):
+        status = _deploy_status(ct, local_hash, ix)
+        if "IN-SYNC" in status:
+            out.section_line(f"cmdeploy run: {ct.sname}: {status}, skipping")
+        else:
+            to_deploy.append(ct)
+
+    if to_deploy:
+        with out.section("cmdeploy run (parallel)"):
+            ret = _run_cmdeploy_parallel(
+                "run", to_deploy, ix, out, extra=["--skip-dns-check"]
+            )
+            if ret:
+                return ret
+
+    # Phase 3: publish images (sequential, fast)
+    for ct in map(ix.get_container, relay_names):
+        if ct.publish_image():
+            out.section_line(f"LXC: published {ct.sname} image")
+        else:
+            out.section_line(
+                f"LXC: publish {ct.sname} image: skipped, cached",
+            )
+
+    for ct in map(ix.get_container, relay_names):
+        with out.section(f"cmdeploy dns: {ct.sname} ({ct.domain})"):
+            ret = _run_cmdeploy("dns", ct, ix, out, extra=["--zonefile", str(ct.zone)])
+            if ret:
+                out.red(f"DNS for {ct.sname} failed (exit {ret})")
+                return ret
+
+    with out.section("LXC: PowerDNS zone update"):
+        dns_ct = ix.get_dns_container()
+        for ct in map(ix.get_container, relay_names):
+            if ct.zone.exists():
+                zone_data = ct.zone.read_text()
+                out.print(f"  Loading {ct.zone} into PowerDNS ...")
+                dns_ct.set_dns_records(zone_data)
+
+    # Run tests in both directions when two relays are available.
+    test_pairs = [(0, 1), (1, 0)] if len(relay_names) > 1 else [(0,)]
+    for pair in test_pairs:
+        first = ix.get_container(relay_names[pair[0]])
+        label = first.sname
+        env = None
+        if len(pair) > 1:
+            second = ix.get_container(relay_names[pair[1]])
+            label = f"{first.sname} \u2194 {second.sname}"
+            env = os.environ.copy()
+            env["CHATMAIL_DOMAIN2"] = second.domain
+
+        with out.section(f"cmdeploy test: {label}"):
+            ret = _run_cmdeploy("test", first, ix, out, **({"env": env} if env else {}))
+            if ret:
+                out.red(f"Tests failed (exit {ret})")
+                return ret
+
+    elapsed = time.time() - t_total
+    out.section_line(f"lxc-test complete ({elapsed:.1f}s)")
+    if out.section_timings:
+        out.print("Section timings:")
+        for name, secs in out.section_timings:
+            out.print(f"  {name:.<50s} {secs:5.1f}s")
+        out.print(f"  {'total':.<50s} {elapsed:5.1f}s")
+    out.section_timings.clear()
+    return 0
+
+
+# -------------------------------------------------------------------
+# lxc-status
+# -------------------------------------------------------------------
+
+
+def lxc_status_cmd_options(parser):
+    pass
+
+
+def lxc_status_cmd(args, out):
+    """Show status of local LXC chatmail containers."""
+    ix = Incus()
+    containers = ix.list_managed()
+    if not containers:
+        out.red("No LXC containers found.  Run 'cmdeploy lxc-start' first.")
+        return 1
+
+    local_hash = get_git_hash()
+
+    # Get storage pool path for display
+    storage_path = None
+    data = ix.run_json(["storage", "show", "default"], check=False)
+    if data:
+        storage_path = data.get("config", {}).get("source")
+    if storage_path:
+        out.green(f"Containers: ({storage_path})")
+    else:
+        out.green("Containers:")
+
+    dns_ip = None
+    for c in containers:
+        _print_container_status(out, c, ix, local_hash)
+        if c["name"] == ix.get_dns_container().name:
+            dns_ip = c["ip"]
+
+    _print_ssh_status(out, ix)
+    _print_dns_forwarding_status(out, dns_ip)
+    return 0
+
+
+def _print_container_status(out, c, ix, local_hash):
+    """Print name/status, domain/IPs, and RAM for one container."""
+    cname = c["name"]
+    is_running = c.get("status") == "Running"
+    ct = ix.get_container(cname)
+
+    # First line: name + running/STOPPED + deploy status
+    if not is_running:
+        tag = "STOPPED"
+    elif not isinstance(ct, RelayContainer):
+        tag = "running"
+    else:
+        tag = f"running {_deploy_status(ct, local_hash, ix)}"
+    out.print(f"  {cname:20s} {tag}")
+
+    # Second line: domain, IPv4, IPv6
+    domain = c.get("domain", "")
+    ip = c.get("ip") or "?"
+    ipv6 = c.get("ipv6")
+    out.print(f"  {domain:20s} {_format_addrs(ip, ipv6)}")
+
+    # Third line: RAM (RSS), config
+    indent = " " * 21
+    try:
+        used, total = ct.rss_mib()
+    except Exception:
+        ram_str = "RSS ?"
+    else:
+        ram_str = f"RSS {used}/{total} MiB ({used * 100 // total}%)"
+
+    if isinstance(ct, RelayContainer):
+        detail = f"{ram_str}, config: {os.path.relpath(ct.ini)}"
+    else:
+        detail = ram_str
+
+    out.print(f"  {indent}{detail}")
+    out.print()
+
+
+def _print_ssh_status(out, ix):
+    """Print SSH integration status."""
+    out.print()
+    ssh_cfg = ix.ssh_config_path
+    if ix.check_ssh_include():
+        out.green("SSH: ~/.ssh/config includes lxconfigs/ssh-config ✓")
+    else:
+        out.red("SSH: ~/.ssh/config does NOT include lxconfigs/ssh-config")
+        out.print("  Add to ~/.ssh/config:")
+        out.print(f"      Include {ssh_cfg}")
+
+
+def _print_dns_forwarding_status(out, dns_ip):
+    """Print host DNS forwarding status for .localchat."""
+    if not dns_ip:
+        out.red("DNS: ns-localchat container not found")
+        return
+    try:
+        rv = shell("resolvectl status incusbr0", timeout=5)
+        dns_ok = dns_ip in rv.stdout and "localchat" in rv.stdout
+    except (FileNotFoundError, subprocess.TimeoutExpired, OSError):
+        dns_ok = None
+    if dns_ok is True:
+        out.green(f"DNS: .localchat forwarding to {dns_ip} ✓")
+    elif dns_ok is False:
+        out.red("DNS: .localchat forwarding NOT configured")
+        out.print("  Run:")
+        out.print(f"      sudo resolvectl dns incusbr0 {dns_ip}")
+        out.print("      sudo resolvectl domain incusbr0 ~localchat")
+    else:
+        out.print("  DNS: .localchat forwarding status UNKNOWN")
+
+
+# -------------------------------------------------------------------
+# Internal helpers
+# -------------------------------------------------------------------
+
+
+def _format_addrs(ip, ipv6=None):
+    parts = [f"IPv4 {ip}"]
+    if ipv6:
+        parts.append(f"IPv6 {ipv6}")
+    return ", ".join(parts)
+
+
+def _deploy_status(ct, local_hash, ix):
+    """Return a human-readable deploy status string.
+
+    Compares the full deployed version (hash + diff) against
+    the local state built by :func:`~cmdeploy.util.get_version_string`.
+    """
+    deployed = ct.deployed_version()
+    if deployed is None:
+        return "NOT DEPLOYED"
+
+    # A container launched from the relay image has the same
+    # git hash but a different domain - always redeploy.
+    deployed_domain = ct.deployed_domain()
+    if deployed_domain and deployed_domain != ct.domain:
+        return f"DOMAIN-MISMATCH (deployed: {deployed_domain})"
+
+    deployed_lines = deployed.splitlines()
+    deployed_hash = deployed_lines[0] if deployed_lines else ""
+    short = deployed_hash[:12]
+
+    if not local_hash:
+        return f"UNKNOWN (deployed: {short})"
+
+    local_short = local_hash[:12]
+    if deployed_hash != local_hash:
+        return f"STALE (deployed: {short}, local: {local_short})"
+
+    # Hash matches - check for uncommitted diffs
+    local_version = get_version_string()
+    if deployed != local_version:
+        return f"DIRTY ({local_short}, undeployed changes)"
+
+    return f"IN-SYNC ({short})"
+
+
+def _add_name_args(parser, help_text=None):
+    """Add optional positional NAME arguments."""
+    parser.add_argument(
+        "names",
+        nargs="*",
+        metavar="NAME",
+        help=help_text or "Relay name(s) to operate on.",
+    )
+
+
+def _build_cmdeploy_cmd(subcmd, ct, ix, extra=None):
+    """Build the ``cmdeploy <subcmd>`` command string."""
+    extra_str = " ".join(extra) if extra else ""
+    return collapse(f"""\
+        cmdeploy {subcmd}
+        --config {ct.ini}
+        --ssh-config {ix.ssh_config_path}
+        --ssh-host {ct.domain}
+        {extra_str}
+    """)
+
+
+def _run_cmdeploy(subcmd, ct, ix, out, extra=None, **kwargs):
+    """Run ``cmdeploy <subcmd>`` with standard --config/--ssh flags.
+
+    *ct* is a Container (uses ``ct.ini`` and ``ct.domain``).
+    Returns the subprocess exit code.
+    """
+    cmd = _build_cmdeploy_cmd(subcmd, ct, ix, extra=extra)
+    if "cwd" not in kwargs:
+        kwargs["cwd"] = str(ix.project_root)
+    out.print(f"  [$ {cmd}]")
+    return shell(cmd, capture_output=False, **kwargs).returncode
+
+
+# Number of tail lines to print on failure.
+_FAIL_CONTEXT_LINES = 40
+
+
+def _run_cmdeploy_parallel(subcmd, containers, ix, out, extra=None):
+    """Run ``cmdeploy <subcmd>`` for every container in parallel.
+
+    Output is captured and filtered: only lines containing
+    ``"Start operation"`` are printed (prefixed with the relay
+    short-name).  On failure the last *_FAIL_CONTEXT_LINES*
+    lines of that process's output are shown.
+    """
+    procs = []  # list of (container, Popen, collected_lines)
+    cwd = str(ix.project_root)
+
+    for ct in containers:
+        cmd = _build_cmdeploy_cmd(subcmd, ct, ix, extra=extra)
+        out.print(f"  [{ct.sname}] $ {cmd}")
+        proc = subprocess.Popen(
+            cmd,
+            shell=True,
+            text=True,
+            stdout=subprocess.PIPE,
+            stderr=subprocess.STDOUT,
+            cwd=cwd,
+        )
+        procs.append((ct, proc, []))
+
+    def _reader(ct, proc, lines):
+        prefix = f"  [{ct.sname}]"
+        for raw in proc.stdout:
+            line = raw.rstrip("\n")
+            lines.append(line)
+            if "Starting operation" in line:
+                out.print(f"{prefix} {line}")
+
+    threads = []
+    for ct, proc, lines in procs:
+        t = threading.Thread(
+            target=_reader,
+            args=(ct, proc, lines),
+            daemon=True,
+        )
+        t.start()
+        threads.append(t)
+
+    for t in threads:
+        t.join()
+    for _, proc, _ in procs:
+        proc.wait()
+
+    # Check results
+    first_failure = 0
+    for ct, proc, lines in procs:
+        if proc.returncode:
+            out.red(f"Deploy to {ct.sname} failed " f"(exit {proc.returncode})")
+            tail = lines[-_FAIL_CONTEXT_LINES:]
+            for tl in tail:
+                out.print(f"  [{ct.sname}] {tl}")
+            if not first_failure:
+                first_failure = proc.returncode
+
+    return first_failure

cmdeploy/src/cmdeploy/lxc/incus.py

@@ -0,0 +1,754 @@
+"""Core Incus operations for local chatmail LXC containers."""
+
+import json
+import subprocess
+import textwrap
+import time
+from pathlib import Path
+
+from ..util import shell
+
+LABEL_KEY = "user.localchat-managed"
+SSH_KEY_NAME = "id_localchat"
+DOMAIN_SUFFIX = ".localchat"
+UPSTREAM_IMAGE = "images:debian/12"
+BASE_IMAGE_ALIAS = "localchat-base"
+BASE_SETUP_NAME = "localchat-base-setup"
+DNS_IMAGE_ALIAS = "localchat-ns"
+
+DNS_CONTAINER_NAME = "ns-localchat"
+DNS_DOMAIN = "ns.localchat"
+
+BRIDGE_IPV4 = "10.200.200.1/24"
+DNS_IP = "10.200.200.2"
+RELAY_IPS = {
+    "test0": "10.200.200.10",
+    "test1": "10.200.200.11",
+    "test2": "10.200.200.12",
+}
+
+
+def _extract_ip(net_data, family="inet"):
+    """Extract the first global-scope IP of *family* from network state data.
+
+    *net_data* is the ``state.network`` dict from ``incus list --format=json``.
+    *family* is ``"inet"`` for IPv4 or ``"inet6"`` for IPv6.
+    Returns the address string, or None.
+    """
+    for iface_name, iface in net_data.items():
+        if iface_name == "lo":
+            continue
+        for addr in iface.get("addresses", []):
+            if addr["family"] == family and addr["scope"] == "global":
+                return addr["address"]
+    return None
+
+
+class Incus:
+    """Gateway for all Incus container operations.
+
+    Instantiated once per CLI command and passed around so that
+    all modules share a single entry point for Incus interactions.
+    """
+
+    def __init__(self):
+        self.project_root = Path(__file__).resolve().parent.parent.parent.parent.parent
+        self.lxconfigs_dir = self.project_root / "lxconfigs"
+        self.lxconfigs_dir.mkdir(exist_ok=True)
+        self.ssh_key_path = self.lxconfigs_dir / SSH_KEY_NAME
+        if not self.ssh_key_path.exists():
+            shell(
+                f"ssh-keygen -t ed25519 -f {self.ssh_key_path} -N '' -C localchat",
+                check=True,
+            )
+        self.ssh_config_path = self.lxconfigs_dir / "ssh-config"
+
+    def write_ssh_config(self):
+        """Write ``lxconfigs/ssh-config`` mapping all containers to their IPs.
+
+        Each Host block maps the container name, the domain name, and the
+        short relay name (e.g. ``_test0``) to the container's IP, using the
+        shared localchat SSH key.  Returns the path to the file.
+        """
+        containers = self.list_managed()
+        key_path = self.ssh_key_path
+        lines = ["# Auto-generated by cmdeploy lxc-start - do not edit\n"]
+        for c in containers:
+            hosts = [c["name"]]
+            domain = c.get("domain", "")
+            if domain and domain != c["name"]:
+                hosts.append(domain)
+                short = domain.split(".")[0]
+                if short and short not in hosts:
+                    hosts.append(short)
+            lines.append(f"\nHost {' '.join(hosts)}\n")
+            lines.append(f"    Hostname {c['ip']}\n")
+            lines.append("    User root\n")
+            lines.append(f"    IdentityFile {key_path}\n")
+            lines.append("    IdentitiesOnly yes\n")
+            lines.append("    StrictHostKeyChecking accept-new\n")
+            lines.append("    UserKnownHostsFile /dev/null\n")
+            lines.append("    LogLevel ERROR\n")
+        path = self.ssh_config_path
+        path.write_text("".join(lines))
+        return path
+
+    def check_ssh_include(self):
+        """Check if the user's ~/.ssh/config already includes our ssh-config."""
+        user_ssh_config = Path.home() / ".ssh" / "config"
+        if not user_ssh_config.exists():
+            return False
+        lines = filter(None, map(str.strip, user_ssh_config.open("r")))
+        return f"Include {self.ssh_config_path}" in lines
+
+    def run(self, args, check=True, capture=True, input=None):
+        """Run an incus command."""
+        cmd = ["incus"] + list(args)
+        kwargs = dict(check=check, text=True, input=input)
+        if capture:
+            kwargs["capture_output"] = True
+        else:
+            kwargs["stdout"] = None
+            kwargs["stderr"] = None
+        return subprocess.run(cmd, **kwargs)  # noqa: PLW1510
+
+    def run_json(self, args, check=True):
+        """Run an incus command with ``--format=json``.
+
+        Returns the parsed JSON on success.
+        When *check* is True raises ``subprocess.CalledProcessError``
+        on non-zero exit; when False returns *None* instead.
+        """
+        result = self.run(
+            list(args) + ["--format=json"],
+            check=check,
+        )
+        if result.returncode != 0:
+            return None
+        return json.loads(result.stdout)
+
+    def run_output(self, args, check=True):
+        """Run an incus command and return its stripped stdout.
+
+        When *check* is False, returns *None* on non-zero exit
+        instead of raising.
+        """
+        result = self.run(args, check=check)
+        if result.returncode != 0:
+            return None
+        return result.stdout.strip()
+
+    def _find_image(self, alias):
+        """Return *alias* if an image with that alias exists, else None."""
+        images = self.run_json(["image", "list"], check=False) or []
+        for img in images:
+            for a in img.get("aliases", []):
+                if a.get("name") == alias:
+                    return alias
+        return None
+
+    def find_dns_image(self):
+        """Return the DNS image alias if it exists, else None."""
+        return self._find_image(DNS_IMAGE_ALIAS)
+
+    def delete_images(self):
+        """Delete all cached localchat images."""
+        for alias in (DNS_IMAGE_ALIAS, BASE_IMAGE_ALIAS):
+            self.run(["image", "delete", alias], check=False)
+        for name in RELAY_IPS:
+            self.run(["image", "delete", f"localchat-{name}"], check=False)
+
+    def list_managed(self):
+        """Return list of dicts with name, ip, ipv6, domain, status, memory_usage."""
+        containers = []
+        for ct in self.run_json(["list"]):
+            config = ct.get("config", {})
+            if config.get(LABEL_KEY) != "true":
+                continue
+            name = ct["name"]
+            state = ct.get("state", {})
+            net = state.get("network") or {}
+            containers.append(
+                {
+                    "name": name,
+                    "ip": _extract_ip(net, "inet"),
+                    "ipv6": _extract_ip(net, "inet6"),
+                    "domain": config.get(
+                        "user.localchat-domain", f"{name}{DOMAIN_SUFFIX}"
+                    ),
+                    "status": ct.get("status", "Unknown"),
+                    "memory_usage": state.get("memory", {}).get("usage", 0),
+                }
+            )
+        return containers
+
+    def ensure_base_image(self):
+        """Build and cache a base image with openssh and the SSH key.
+
+        The image is published as a local incus image with alias
+        'localchat-base'.  Subsequent container launches use this
+        image instead of the upstream Debian 12, skipping the
+        slow apt-get install step.
+        Returns the image alias.
+        """
+        if self._find_image(BASE_IMAGE_ALIAS):
+            return BASE_IMAGE_ALIAS
+
+        print("  Building base image (one-time setup) ...")
+
+        self.run(["delete", BASE_SETUP_NAME, "--force"], check=False)
+        self.run(["image", "delete", BASE_IMAGE_ALIAS], check=False)
+        self.run(
+            ["launch", UPSTREAM_IMAGE, BASE_SETUP_NAME, "-c", "limits.memory=512MiB"]
+        )
+
+        ct = Container(self, BASE_SETUP_NAME, memory="512MiB")
+        ct.wait_ready()
+
+        key_path = self.ssh_key_path
+        pub_key = key_path.with_suffix(".pub").read_text().strip()
+        print("  ── apt-get install (base image) ──")
+        ct.bash(
+            f"""\
+            systemctl disable --now systemd-resolved 2>/dev/null || true
+            rm -f /etc/resolv.conf
+            echo 'nameserver 9.9.9.9' > /etc/resolv.conf
+            while fuser /var/lib/apt/lists/lock >/dev/null 2>&1 ; do
+                echo "Waiting for other apt-get instance to finish..."
+                sleep 5
+            done
+            apt-get -o DPkg::Lock::Timeout=60 update
+            DEBIAN_FRONTEND=noninteractive apt-get install -y openssh-server python3
+            systemctl enable ssh
+            apt-get clean
+            mkdir -p /root/.ssh
+            chmod 700 /root/.ssh
+            echo '{pub_key}' > /root/.ssh/authorized_keys
+            chmod 600 /root/.ssh/authorized_keys
+        """,
+            capture=False,
+        )
+        print("  ── base image install done ──")
+
+        self.run(["stop", BASE_SETUP_NAME])
+        self.run(["publish", BASE_SETUP_NAME, f"--alias={BASE_IMAGE_ALIAS}"])
+        self.run(["delete", BASE_SETUP_NAME, "--force"])
+        print(f"  Base image '{BASE_IMAGE_ALIAS}' ready.")
+        return BASE_IMAGE_ALIAS
+
+    def ensure_bridge(self):
+        """Ensure incusbr0 exists and uses our fixed IPv4 subnet."""
+        bridge = self.run_json(["network", "show", "incusbr0"], check=False)
+        if bridge and bridge.get("config", {}).get("ipv4.address") == BRIDGE_IPV4:
+            return
+
+        print(f"  Configuring incusbr0 with static subnet {BRIDGE_IPV4} ...")
+        if not bridge:
+            self.run(["network", "create", "incusbr0"], check=False)
+
+        self.run(
+            [
+                "network",
+                "set",
+                "incusbr0",
+                f"ipv4.address={BRIDGE_IPV4}",
+                "ipv4.nat=true",
+                "ipv6.address=none",
+                "dns.mode=none",
+            ]
+        )
+
+    def get_container(self, name):
+        """Return a container handle for the given name.
+
+        Accepts both short relay names (``test0``) and full Incus
+        container names (``test0-localchat``).  Returns
+        ``DNSContainer`` for the DNS container and
+        ``RelayContainer`` for everything else.
+        """
+        if name == DNS_CONTAINER_NAME:
+            return DNSContainer(self)
+        return RelayContainer(self, name.removesuffix("-localchat"))
+
+    def get_dns_container(self):
+        """Return a DNSContainer handle."""
+        return DNSContainer(self)
+
+
+class Container:
+    """Lightweight handle for an Incus container.
+
+    Carries the container *name* and provides convenience methods
+    for running commands, managing lifecycle, and extracting state
+    so callers don't repeat the name everywhere.
+    """
+
+    def __init__(self, incus, name, domain=None, memory="200MiB", ipv4=None):
+        self.incus = incus
+        self.name = name
+        self.domain = domain or f"{name}{DOMAIN_SUFFIX}"
+        self.memory = memory
+        self.ipv4 = ipv4
+        self.ipv6 = None
+
+    def bash(self, script, check=True, capture=True):
+        """Returns stdout from executing ``bash -ec <script>`` inside this container.
+
+        *script* is dedented and stripped so callers can use triple-quoted strings.
+        When *check* is False, returns *None* on non-zero exit instead of raising.
+        When *capture* is False, output streams to the terminal and None is returned.
+        """
+        cmd = ["exec", self.name, "--", "bash", "-ec", textwrap.dedent(script).strip()]
+        if not capture:
+            self.incus.run(cmd, check=check, capture=False)
+            return None
+        return self.incus.run_output(cmd, check=check)
+
+    def run_cmd(self, *args, check=True):
+        """Return stdout from running a command directly in the container (no shell).
+
+        When *check* is False, returns *None* on non-zero exit instead of raising.
+        """
+        return self.incus.run_output(
+            ["exec", self.name, "--", *args],
+            check=check,
+        )
+
+    def start(self):
+        self.incus.run(["start", self.name])
+
+    def stop(self, force=False):
+        cmd = ["stop", self.name]
+        if force:
+            cmd.append("--force")
+        self.incus.run(cmd, check=False)
+
+    def launch(self, image=None):
+        """Launch from the specified image, or the base image if None."""
+        self.incus.ensure_bridge()
+        if image is None:
+            image = self.incus.ensure_base_image()
+        cfg = []
+        cfg += ("-c", f"{LABEL_KEY}=true")
+        cfg += ("-c", f"user.localchat-domain={self.domain}")
+        cfg += ("-c", f"limits.memory={self.memory}")
+        self.incus.run(["init", image, self.name, *cfg])
+        if self.ipv4:
+            self.incus.run(
+                [
+                    "config",
+                    "device",
+                    "override",
+                    self.name,
+                    "eth0",
+                    f"ipv4.address={self.ipv4}",
+                ]
+            )
+        self.incus.run(["start", self.name])
+        return image
+
+    def ensure(self):
+        """Create/start this container from the cached base image.
+
+        On first call, builds the base image (~30s).
+        Subsequent containers launch in ~2s from the cached image.
+        Returns ``self`` for chaining.
+        """
+        data = self.incus.run_json(["list", self.name], check=False) or []
+
+        existing = [c for c in data if c["name"] == self.name]
+        image = None
+        if existing:
+            status = existing[0]["status"]
+            if status != "Running":
+                print(f"  Starting stopped {self.name} container ...")
+                self.start()
+            else:
+                print(f"  {self.name} already running")
+        else:
+            image = self.launch()
+        self.wait_ready()
+        if image:
+            print(f"  Ensured {self.name} (launched from {image!r} image)")
+        return self
+
+    def destroy(self):
+        """Stop, delete, and clean up config files."""
+        self.stop(force=True)
+        self.incus.run(["delete", self.name, "--force"], check=False)
+
+    def push_file_content(self, dest_path, content):
+        """Write *content* to *dest_path* inside the container.
+
+        *content* is dedented and stripped so callers can use
+        indented triple-quoted strings.
+        """
+        content = textwrap.dedent(content).strip() + "\n"
+        self.incus.run(
+            ["file", "push", "-", f"{self.name}{dest_path}"],
+            input=content,
+        )
+        self.bash(f"chmod 644 {dest_path}")
+
+    def wait_ready(self, timeout=60):
+        """Wait until the container is running with an IPv4 address.
+
+        Sets ``self.ipv4`` and ``self.ipv6`` (may be *None*),
+        or raises ``TimeoutError``.
+        """
+        deadline = time.time() + timeout
+        while time.time() < deadline:
+            data = self.incus.run_json(
+                ["list", self.name],
+                check=False,
+            )
+            if data and data[0].get("status") == "Running":
+                net = data[0].get("state", {}).get("network", {})
+                self.ipv4 = _extract_ip(net, "inet")
+                self.ipv6 = _extract_ip(net, "inet6")
+                if self.ipv4:
+                    return
+            time.sleep(1)
+        raise TimeoutError(
+            f"Container {self.name!r} did not become ready within {timeout}s"
+        )
+
+    def rss_mib(self):
+        """Return ``(used, total)`` memory from container (or None if unobtainable)."""
+        output = self.bash("free -m", check=False)
+        if output:
+            for line in output.splitlines():
+                if line.startswith("Mem:"):
+                    parts = line.split()
+                    return int(parts[2]), int(parts[1])
+
+
+class RelayContainer(Container):
+    """Container handle for a chatmail relay.
+
+    Accepts the short relay name (e.g. ``test0``) and derives
+    the Incus container name and mail domain automatically.
+    """
+
+    def __init__(self, incus, name):
+        super().__init__(
+            incus,
+            f"{name}-localchat",
+            domain=f"_{name}{DOMAIN_SUFFIX}",
+            memory="500MiB",
+            ipv4=RELAY_IPS.get(name),
+        )
+        self.sname = name
+        self.image_alias = f"localchat-{name}"
+        self.ini = incus.lxconfigs_dir / f"chatmail-{name}.ini"
+        self.zone = incus.lxconfigs_dir / f"{name}.zone"
+
+    def launch(self):
+        """Launch from a cached per-relay image if available, else from base."""
+        cached = self.incus._find_image(self.image_alias)
+        if cached:
+            print(f"  Using cached image {cached!r}")
+        else:
+            print("  No cached image, building from base")
+        image = super().launch(image=cached)
+        self.bash("rm -f /etc/chatmail-version")
+        return image
+
+    def destroy(self):
+        """Stop, delete, and clean up config files."""
+        super().destroy()
+        if self.ini.exists():
+            self.ini.unlink()
+
+    def disable_ipv6(self):
+        """Disable IPv6 inside the container via sysctl."""
+        self.bash("""\
+            sysctl -w net.ipv6.conf.all.disable_ipv6=1
+            sysctl -w net.ipv6.conf.default.disable_ipv6=1
+            mkdir -p /etc/sysctl.d
+            printf 'net.ipv6.conf.all.disable_ipv6=1\\n
+            net.ipv6.conf.default.disable_ipv6=1\\n'
+            > /etc/sysctl.d/99-disable-ipv6.conf
+        """)
+
+    def configure_hosts(self, ip):
+        """Set hostname and /etc/hosts inside the container."""
+        self.bash(f"""
+            echo '{self.name}' > /etc/hostname
+            hostname {self.name}
+            sed -i '/ {self.domain}$/d' /etc/hosts
+            echo '{ip} {self.name} {self.domain}' >> /etc/hosts
+        """)
+
+    def publish_image(self):
+        """Publish this container as a reusable per-relay image.
+
+        Returns True if an image was published,
+        False if a cached image already existed.
+        """
+        if self.incus._find_image(self.image_alias):
+            return False
+        self.bash("apt-get clean && rm -rf /var/lib/apt/lists/*")
+        print(f"  Publishing {self.name!r} as {self.image_alias!r} image ...")
+        self.incus.run(
+            ["publish", self.name, f"--alias={self.image_alias}", "--force"],
+            capture=False,
+        )
+        self.wait_ready()
+        print(f"  Image {self.image_alias!r} ready.")
+        return True
+
+    def deployed_version(self):
+        """Read /etc/chatmail-version, or None if absent."""
+        return self.bash("cat /etc/chatmail-version", check=False)
+
+    def deployed_domain(self):
+        """Read the domain deployed on the container (postfix myhostname)."""
+        return self.bash(
+            "postconf -h myhostname 2>/dev/null",
+            check=False,
+        )
+
+    def verify_ssh(self, ssh_config):
+        """Verify SSH connectivity to this container."""
+        cmd = f"ssh -F {ssh_config} -o ConnectTimeout=10 root@{self.domain} hostname"
+        return shell(cmd, timeout=15).returncode == 0
+
+    def configure_dns(self, dns_ip):
+        """Point this container's resolver at *dns_ip*.
+
+        Disables systemd-resolved to free port 53 and writes
+        a static /etc/resolv.conf.  Also configures unbound
+        (if present) to forward .localchat queries.
+        """
+        self.bash(f"""\
+            systemctl disable --now systemd-resolved 2>/dev/null || true
+            rm -f /etc/resolv.conf
+            echo 'nameserver {dns_ip}' > /etc/resolv.conf
+            mkdir -p /etc/unbound/unbound.conf.d
+            printf 'server:\\n  domain-insecure: "localchat"\\n\\n
+            forward-zone:\\n  name: "localchat"\\n
+              forward-addr: {dns_ip}\\n'
+            > /etc/unbound/unbound.conf.d/localchat-forward.conf
+            systemctl restart unbound 2>/dev/null || true
+        """)
+
+    def check_dns(self, retries=5, delay=2):
+        """Verify that external DNS resolution works inside the container."""
+        for i in range(retries):
+            result = self.bash(
+                "getent hosts pypi.org",
+                check=False,
+            )
+            if result:
+                return True
+            if i < retries - 1:
+                time.sleep(delay)
+        return False
+
+    def write_ini(self, disable_ipv6=False):
+        """Generate a chatmail.ini config file in lxconfigs/."""
+        from chatmaild.config import write_initial_config
+
+        overrides = {
+            "max_user_send_per_minute": 600,
+            "max_user_send_burst_size": 100,
+            "mtail_address": "127.0.0.1",
+        }
+        if disable_ipv6:
+            overrides["disable_ipv6"] = "True"
+        write_initial_config(self.ini, self.domain, overrides)
+        return self.ini
+
+
+class DNSContainer(Container):
+    """Specialised container handle for the PowerDNS name server."""
+
+    def __init__(self, incus):
+        super().__init__(
+            incus, DNS_CONTAINER_NAME, domain=DNS_DOMAIN, memory="256MiB", ipv4=DNS_IP
+        )
+
+    def launch(self):
+        """Launch from cached DNS image if available, else from base image."""
+        cached = self.incus._find_image(DNS_IMAGE_ALIAS)
+        if cached:
+            print(f"  Using cached image {cached!r}")
+        else:
+            print("  No cached image, building from base")
+        return super().launch(image=cached)
+
+    def publish_as_dns_image(self):
+        """Publish this container as a reusable DNS image."""
+        if self.incus._find_image(DNS_IMAGE_ALIAS):
+            return
+        self.bash("apt-get clean && rm -rf /var/lib/apt/lists/*")
+        print(f"  Publishing {self.name!r} as {DNS_IMAGE_ALIAS!r} image ...")
+        self.incus.run(
+            ["publish", self.name, f"--alias={DNS_IMAGE_ALIAS}", "--force"],
+            capture=False,
+        )
+        self.wait_ready()
+        print(f"  DNS image {DNS_IMAGE_ALIAS!r} ready.")
+
+    def pdnsutil(self, *args, check=True):
+        """Run ``pdnsutil <args>`` inside the DNS container."""
+        return self.run_cmd("pdnsutil", *args, check=check)
+
+    def replace_rrset(self, zone, name, rtype, ttl, rdata):
+        """Shortcut for ``pdnsutil replace-rrset``."""
+        self.pdnsutil("replace-rrset", zone, name, rtype, ttl, rdata)
+
+    def restart_services(self):
+        """Restart pdns and pdns-recursor."""
+        self.bash("""\
+            systemctl restart pdns
+            systemctl restart pdns-recursor || true
+        """)
+
+    def ensure(self):
+        """Create the DNS container with PowerDNS if needed.
+
+        Calls ``super().ensure()`` to create/start the container
+        and set up SSH, then installs PowerDNS and configures
+        the Incus bridge to use this container as DNS.
+        """
+        super().ensure()
+        self._install_powerdns()
+        self.incus.run(
+            ["network", "set", "incusbr0", "dns.mode=none"],
+            check=False,
+        )
+        self.incus.run(
+            ["network", "set", "incusbr0", f"raw.dnsmasq=dhcp-option=6,{self.ipv4}"],
+            check=False,
+        )
+
+    def _install_powerdns(self):
+        """Install and configure PowerDNS if not already present."""
+        if self.run_cmd("which", "pdns_server", check=False) is not None:
+            return
+
+        self.bash("""\
+            systemctl disable --now systemd-resolved 2>/dev/null || true
+            rm -f /etc/resolv.conf
+            echo 'nameserver 9.9.9.9' > /etc/resolv.conf
+            apt-get -o DPkg::Lock::Timeout=60 update
+            DEBIAN_FRONTEND=noninteractive apt-get install -y \
+                pdns-server pdns-backend-sqlite3 sqlite3 pdns-recursor dnsutils
+            systemctl stop pdns pdns-recursor || true
+            mkdir -p /var/lib/powerdns
+            sqlite3 /var/lib/powerdns/pdns.sqlite3 \
+                </usr/share/doc/pdns-backend-sqlite3/schema.sqlite3.sql
+            chown -R pdns:pdns /var/lib/powerdns
+        """)
+
+        self.push_file_content(
+            "/etc/powerdns/pdns.conf",
+            """\
+            launch=gsqlite3
+            gsqlite3-database=/var/lib/powerdns/pdns.sqlite3
+            local-address=127.0.0.1
+            local-port=5353
+            """,
+        )
+
+        self.push_file_content(
+            "/etc/powerdns/recursor.conf",
+            """\
+            local-address=0.0.0.0
+            local-port=53
+            forward-zones=localchat=127.0.0.1:5353
+            forward-zones-recurse=.=9.9.9.9;149.112.112.112
+            allow-from=0.0.0.0/0
+            dont-query=
+            dnssec=off
+            """,
+        )
+
+        self.bash("""\
+            systemctl start pdns
+            systemctl start pdns-recursor
+            echo 'nameserver 127.0.0.1' > /etc/resolv.conf
+        """)
+
+    def reset_dns_records(self, dns_ip, domains):
+        """Create DNS zones with initial A records via pdnsutil.
+
+        Only sets SOA, NS, and A records as the minimal set
+        needed for SSH connectivity.  Full records (MX, TXT, SRV,
+        CNAME, DKIM) are added later by ``cmdeploy dns``.
+
+        Args:
+            dns_ip: IP of the DNS container
+            domains: list of dicts with 'name', 'domain', 'ip'
+        """
+        for d in domains:
+            domain = d["domain"]
+            ip = d["ip"]
+            print(f"  {domain} -> {ip}")
+
+            # Delete and recreate zone fresh (removes stale records)
+            self.pdnsutil("delete-zone", domain, check=False)
+            self.pdnsutil("create-zone", domain, f"ns.{domain}")
+
+            serial = str(int(time.time()))
+            soa = f"ns.{domain} hostmaster.{domain} {serial} 3600 900 604800 300"
+            self.replace_rrset(domain, ".", "SOA", "3600", soa)
+            self.replace_rrset(domain, ".", "NS", "3600", f"ns.{domain}.")
+            self.replace_rrset(domain, ".", "A", "3600", ip)
+            self.replace_rrset(domain, "ns", "A", "3600", dns_ip)
+
+            # AAAA (domain -> container IPv6, if available)
+            ipv6 = d.get("ipv6")
+            if ipv6:
+                self.replace_rrset(domain, ".", "AAAA", "3600", ipv6)
+                print(f"    zone reset: SOA, NS, A, AAAA  ({ip}, {ipv6})")
+            else:
+                # Remove any stale AAAA record
+                self.pdnsutil("delete-rrset", domain, ".", "AAAA", check=False)
+                print(f"    zone reset: SOA, NS, A  ({ip}, IPv4-only)")
+
+        self.restart_services()
+
+    def set_dns_records(self, text):
+        """Add or overwrite DNS records from standard BIND format.
+
+        Uses ``cmdeploy.dns.parse_zone_records`` to parse.
+        Zones are created automatically from the record names.
+        """
+        from ..dns import parse_zone_records
+
+        zones_seen = set()
+
+        for name, ttl, rtype, rdata in parse_zone_records(text):
+            # Derive zone from name: find top-level .localchat domain
+            name_parts = name.split(".")
+            zone = name  # fallback
+            for i in range(len(name_parts) - 1):
+                if name_parts[i + 1 :] == ["localchat"]:
+                    zone = ".".join(name_parts[i:])
+                    break
+
+            # Create zone if first time seeing it
+            if zone not in zones_seen:
+                self.pdnsutil(
+                    "create-zone",
+                    zone,
+                    f"ns.{zone}",
+                    check=False,
+                )
+                zones_seen.add(zone)
+
+            # Figure out the record name relative to zone
+            if name == zone:
+                relative = "."
+            elif name.endswith(f".{zone}"):
+                relative = name[: -(len(zone) + 1)]
+            else:
+                relative = name
+
+            self.replace_rrset(zone, relative, rtype, ttl, rdata)
+
+        if zones_seen:
+            self.restart_services()

cmdeploy/src/cmdeploy/metrics.cron.j2

@@ -1 +0,0 @@
-*/5 * * * * root {{ config.execpath }} {{ config.mailboxes_dir }} >/var/www/html/metrics

cmdeploy/src/cmdeploy/nginx/nginx.conf.j2

@@ -54,7 +54,7 @@ http {
 	include /etc/nginx/mime.types;
 	default_type application/octet-stream;
 
-	ssl_protocols TLSv1 TLSv1.1 TLSv1.2 TLSv1.3;
+	ssl_protocols TLSv1.2 TLSv1.3;
 	ssl_prefer_server_ciphers on;
 	ssl_certificate {{ config.tls_cert_path }};
 	ssl_certificate_key {{ config.tls_key_path }};
@@ -79,10 +79,6 @@ http {
 			try_files $uri $uri/ =404;
 		}
 
-		location /metrics {
-			default_type text/plain;
-		}
-
 		location /new {
 {% if config.tls_cert_mode != "self" %}
 			if ($request_method = GET) {
@@ -145,4 +141,25 @@ http {
 		return 301 $scheme://{{ config.mail_domain }}$request_uri;
 		access_log syslog:server=unix:/dev/log,facility=local7;
 	}
+
+	server {
+		listen 80;
+		{% if not disable_ipv6 %}
+		listen [::]:80;
+		{% endif %}
+
+		{% if config.tls_cert_mode == "acme" %}
+		location /.well-known/acme-challenge/ {
+			proxy_pass http://acmetool;
+		}
+		{% endif %}
+
+		return 301 https://$host$request_uri;
+	}
+
+	{% if config.tls_cert_mode == "acme" %}
+	upstream acmetool {
+		server 127.0.0.1:402;
+	}
+	{% endif %}
 }

cmdeploy/src/cmdeploy/opendkim/deployer.py

@@ -37,21 +37,15 @@ class OpendkimDeployer(Deployer):
         )
         need_restart |= main_config.changed
 
-        screen_script = files.put(
-            src=get_resource("opendkim/screen.lua"),
-            dest="/etc/opendkim/screen.lua",
-            user="root",
-            group="root",
-            mode="644",
+        screen_script = files.file(
+            path="/etc/opendkim/screen.lua",
+            present=False,
         )
         need_restart |= screen_script.changed
 
-        final_script = files.put(
-            src=get_resource("opendkim/final.lua"),
-            dest="/etc/opendkim/final.lua",
-            user="root",
-            group="root",
-            mode="644",
+        final_script = files.file(
+            path="/etc/opendkim/final.lua",
+            present=False,
         )
         need_restart |= final_script.changed
 
@@ -109,6 +103,13 @@ class OpendkimDeployer(Deployer):
         )
         need_restart |= service_file.changed
 
+        files.file(
+            name="chown opendkim: /etc/dkimkeys/opendkim.private",
+            path="/etc/dkimkeys/opendkim.private",
+            user="opendkim",
+            group="opendkim",
+        )
+
         self.need_restart = need_restart
 
     def activate(self):

cmdeploy/src/cmdeploy/opendkim/final.lua

@@ -1,42 +0,0 @@
-mtaname = odkim.get_mtasymbol(ctx, "{daemon_name}")
-if mtaname == "ORIGINATING" then
-	-- Outgoing message will be signed,
-	-- no need to look for signatures.
-	return nil
-end
-
-nsigs = odkim.get_sigcount(ctx)
-if nsigs == nil then
-	return nil
-end
-
-local valid = false
-local error_msg = "No valid DKIM signature found."
-for i = 1, nsigs do
-	sig = odkim.get_sighandle(ctx, i - 1)
-	sigres = odkim.sig_result(sig)
-
-	-- All signatures that do not correspond to From: 
-	-- were ignored in screen.lua and return sigres -1.
-	-- 
-	-- Any valid signature that was not ignored like this
-	-- means the message is acceptable.
-	if sigres == 0 then
-		valid = true
-    else
-        error_msg = "DKIM signature is invalid, error code " .. tostring(sigres) .. ", search https://github.com/trusteddomainproject/OpenDKIM/blob/master/libopendkim/dkim.h#L108"
-	end
-end
-
-if valid then
-	-- Strip all DKIM-Signature headers after successful validation
-	-- Delete in reverse order to avoid index shifting.
-	for i = nsigs, 1, -1 do
-		odkim.del_header(ctx, "DKIM-Signature", i)
-	end
-else
-	odkim.set_reply(ctx, "554", "5.7.1", error_msg)
-	odkim.set_result(ctx, SMFIS_REJECT)
-end
-
-return nil

cmdeploy/src/cmdeploy/opendkim/opendkim.conf

@@ -45,12 +45,6 @@ SignHeaders *,+autocrypt,+content-type
 # Default is empty.
 OversignHeaders from,reply-to,subject,date,to,cc,resent-date,resent-from,resent-sender,resent-to,resent-cc,in-reply-to,references,list-id,list-help,list-unsubscribe,list-subscribe,list-post,list-owner,list-archive,autocrypt
 
-# Script to ignore signatures that do not correspond to the From: domain.
-ScreenPolicyScript /etc/opendkim/screen.lua
-
-# Script to reject mails without a valid DKIM signature.
-FinalPolicyScript /etc/opendkim/final.lua
-
 # In Debian, opendkim runs as user "opendkim". A umask of 007 is required when
 # using a local socket with MTAs that access the socket as a non-privileged
 # user (for example, Postfix). You may need to add user "postfix" to group

cmdeploy/src/cmdeploy/opendkim/screen.lua

@@ -1,21 +0,0 @@
--- Ignore signatures that do not correspond to the From: domain.
-
-from_domain = odkim.get_fromdomain(ctx)
-if from_domain == nil then
-	return nil
-end
-
-n = odkim.get_sigcount(ctx)
-if n == nil then
-	return nil
-end
-
-for i = 1, n do
-	sig = odkim.get_sighandle(ctx, i - 1)
-	sig_domain = odkim.sig_getdomain(sig)
-	if from_domain ~= sig_domain then
-		odkim.sig_ignore(sig)
-	end
-end
-
-return nil

cmdeploy/src/cmdeploy/postfix/deployer.py

@@ -97,7 +97,9 @@ class PostfixDeployer(Deployer):
             server.shell(
                 name="Validate postfix configuration",
                 # Extract stderr and quit with error if non-zero
-                commands=["""bash -c 'w=$(postconf 2>&1 >/dev/null); [[ -z "$w" ]] || { echo "$w"; false; }'"""],
+                commands=[
+                    """bash -c 'w=$(postconf 2>&1 >/dev/null); [[ -z "$w" ]] || { echo "$w"; false; }'"""
+                ],
             )
         self.need_restart = need_restart
 

cmdeploy/src/cmdeploy/postfix/master.cf.j2

@@ -86,7 +86,6 @@ filter    unix -        n       n       -       -       lmtp
 # Local SMTP server for reinjecting incoming filtered mail 
 127.0.0.1:{{ config.postfix_reinject_port_incoming }} inet  n       -       n       -       100      smtpd
   -o syslog_name=postfix/reinject_incoming
-  -o smtpd_milters=unix:opendkim/opendkim.sock
 
 # Cleanup `Received` headers for authenticated mail
 # to avoid leaking client IP.

cmdeploy/src/cmdeploy/remote/rdns.py

@@ -53,13 +53,13 @@ def get_dkim_entry(mail_domain, pre_command, dkim_selector):
             print=log_progress,
         )
     except CalledProcessError:
-        return
+        return None, None
     dkim_value_raw = f"v=DKIM1;k=rsa;p={dkim_pubkey};s=email;t=s"
     dkim_value = '" "'.join(re.findall(".{1,255}", dkim_value_raw))
     web_dkim_value = "".join(re.findall(".{1,255}", dkim_value_raw))
     return (
-        f'{dkim_selector}._domainkey.{mail_domain}. TXT "{dkim_value}"',
-        f'{dkim_selector}._domainkey.{mail_domain}. TXT "{web_dkim_value}"',
+        f'{dkim_selector}._domainkey.{mail_domain}.  3600  IN  TXT  "{dkim_value}"',
+        f'{dkim_selector}._domainkey.{mail_domain}.  3600  IN  TXT  "{web_dkim_value}"',
     )
 
 
@@ -94,9 +94,11 @@ def check_zonefile(zonefile, verbose=True):
         if not zf_line.strip() or zf_line.startswith(";"):
             continue
         print(f"dns-checking {zf_line!r}") if verbose else log_progress("")
-        zf_domain, zf_typ, zf_value = zf_line.split(maxsplit=2)
-        zf_domain = zf_domain.rstrip(".")
-        zf_value = zf_value.strip()
+        parts = zf_line.split(None, 4)
+        zf_domain = parts[0].rstrip(".")
+        # parts[1]=TTL, parts[2]=IN, parts[3]=type, parts[4]=rdata
+        zf_typ = parts[3]
+        zf_value = parts[4].strip()
         query_value = query_dns(zf_typ, zf_domain)
         if zf_value != query_value:
             assert zf_typ in ("A", "AAAA", "CNAME", "CAA", "SRV", "MX", "TXT"), zf_line

cmdeploy/src/cmdeploy/remote/rshell.py

@@ -40,5 +40,5 @@ def dovecot_recalc_quota(user):
     #
     for line in output.split("\n"):
         parts = line.split()
-        if parts[2] == "STORAGE":
+        if len(parts) >= 6 and parts[2] == "STORAGE":
             return dict(value=int(parts[3]), limit=int(parts[4]), percent=int(parts[5]))

cmdeploy/src/cmdeploy/selfsigned/deployer.py

@@ -12,13 +12,27 @@ def openssl_selfsigned_args(domain, cert_path, key_path, days=36500):
     ``www.<domain>`` and ``mta-sts.<domain>``.
     """
     return [
-        "openssl", "req", "-x509",
-        "-newkey", "ec", "-pkeyopt", "ec_paramgen_curve:P-256",
-        "-noenc", "-days", str(days),
-        "-keyout", str(key_path),
-        "-out", str(cert_path),
-        "-subj", f"/CN={domain}",
-        "-addext", "extendedKeyUsage=serverAuth,clientAuth",
+        "openssl",
+        "req",
+        "-x509",
+        "-newkey",
+        "ec",
+        "-pkeyopt",
+        "ec_paramgen_curve:P-256",
+        "-noenc",
+        "-days",
+        str(days),
+        "-keyout",
+        str(key_path),
+        "-out",
+        str(cert_path),
+        "-subj",
+        f"/CN={domain}",
+        # Mark as end-entity cert so it cannot be used as a CA to sign others.
+        "-addext",
+        "basicConstraints=critical,CA:FALSE",
+        "-addext",
+        "extendedKeyUsage=serverAuth,clientAuth",
         "-addext",
         f"subjectAltName=DNS:{domain},DNS:www.{domain},DNS:mta-sts.{domain}",
     ]
@@ -40,7 +54,9 @@ class SelfSignedTlsDeployer(Deployer):
 
     def configure(self):
         args = openssl_selfsigned_args(
-            self.mail_domain, self.cert_path, self.key_path,
+            self.mail_domain,
+            self.cert_path,
+            self.key_path,
         )
         cmd = shlex.join(args)
         server.shell(

cmdeploy/src/cmdeploy/service/chatmail-fsreport.service.f

@@ -5,5 +5,5 @@ After=network.target
 [Service]
 Type=oneshot
 User=vmail
-ExecStart=/usr/local/lib/chatmaild/venv/bin/chatmail-fsreport /usr/local/lib/chatmaild/chatmail.ini 
+ExecStart=/usr/local/lib/chatmaild/venv/bin/chatmail-fsreport /usr/local/lib/chatmaild/chatmail.ini
 

cmdeploy/src/cmdeploy/service/chatmail-metadata.service.f

@@ -4,7 +4,7 @@ Description=Chatmail dict proxy for IMAP METADATA
 [Service]
 ExecStart={execpath} /run/chatmail-metadata/metadata.socket {config_path}
 Restart=always
-RestartSec=30
+RestartSec=5
 User=vmail
 RuntimeDirectory=chatmail-metadata
 UMask=0077

cmdeploy/src/cmdeploy/sshexec.py

@@ -49,8 +49,13 @@ class SSHExec:
     RemoteError = execnet.RemoteError
     FuncError = FuncError
 
-    def __init__(self, host, verbose=False, python="python3", timeout=60):
-        self.gateway = execnet.makegateway(f"ssh=root@{host}//python={python}")
+    def __init__(
+        self, host, verbose=False, python="python3", timeout=60, ssh_config=None
+    ):
+        spec = f"ssh=root@{host}//python={python}"
+        if ssh_config:
+            spec += f"//ssh_config={ssh_config}"
+        self.gateway = execnet.makegateway(spec)
         self._remote_cmdloop_channel = bootstrap_remote(self.gateway, remote)
         self.timeout = timeout
         self.verbose = verbose
@@ -85,6 +90,8 @@ class SSHExec:
 
 
 class LocalExec:
+    FuncError = FuncError
+
     def __init__(self, verbose=False, docker=False):
         self.verbose = verbose
         self.docker = docker
@@ -95,11 +102,62 @@ class LocalExec:
         return call(**kwargs)
 
     def logged(self, call, kwargs: dict):
+        title = call.__doc__
+        if not title:
+            title = call.__name__
         where = "locally"
         if self.docker:
             if call == remote.rdns.perform_initial_checks:
                 kwargs["pre_command"] = "docker exec chatmail "
                 where = "in docker"
         if self.verbose:
-            print(f"Running {where}: {call.__name__}(**{kwargs})")
-        return call(**kwargs)
+            print_stderr(f"Running {where}: {title}(**{kwargs})")
+            return self(call, kwargs, log_callback=print_stderr)
+        else:
+            print_stderr(title, end="")
+            res = self(call, kwargs, log_callback=remote.rshell.log_progress)
+            print_stderr()
+            return res
+
+
+# pyinfra exposes a ``ssh_config_file`` data key that *should* let
+# paramiko parse an SSH config file directly.  In practice it silently
+# fails to connect (zero hosts / zero operations), so we resolve the
+# hostname and identity-file ourselves and pass them via
+# ``--data ssh_hostname`` / ``--data ssh_key`` instead.
+# Execnet uses ssh natively (and not paramiko) and doesn't have this problem.
+
+
+def _get_from_ssh_config(host, ssh_config_path, key):
+    """Internal helper to parse a value for a specific key from ssh-config."""
+    current_hosts = []
+    found_value = None
+    with open(ssh_config_path) as f:
+        for raw_line in f:
+            line = raw_line.strip()
+            if not line or line.startswith("#"):
+                continue
+            parts = line.split(None, 1)
+            if not parts:
+                continue
+            directive = parts[0].lower()
+            if directive == "host":
+                if host in current_hosts and found_value:
+                    return found_value
+                current_hosts = parts[1].split()
+                found_value = None
+            elif directive == key.lower():
+                found_value = parts[1]
+    if host in current_hosts and found_value:
+        return found_value
+    return None
+
+
+def resolve_host_from_ssh_config(host, ssh_config_path):
+    """Resolve a host alias to its IP from an ssh-config file."""
+    return _get_from_ssh_config(host, ssh_config_path, "Hostname") or host
+
+
+def resolve_key_from_ssh_config(host, ssh_config_path):
+    """Resolve a host alias to its IdentityFile from an ssh-config file."""
+    return _get_from_ssh_config(host, ssh_config_path, "IdentityFile")

cmdeploy/src/cmdeploy/tests/data/zftest.zone

@@ -1,17 +1,18 @@
-; Required DNS entries for chatmail servers
-zftest.testrun.org.                   A     135.181.204.127
-zftest.testrun.org.                   AAAA  2a01:4f9:c012:52f4::1
-zftest.testrun.org.                   MX 10 zftest.testrun.org.
-_mta-sts.zftest.testrun.org.          TXT "v=STSv1; id=202403211706"
-mta-sts.zftest.testrun.org.           CNAME zftest.testrun.org.
-www.zftest.testrun.org.               CNAME zftest.testrun.org.
-opendkim._domainkey.zftest.testrun.org. TXT "v=DKIM1;k=rsa;p=MIIBIjANBgkqhkiG9w0BAQEFAAOCAQ8AMIIBCgKCAQEAoYt82CVUyz2ouaqjX2kB+5J80knAyoOU3MGU5aWppmwUwwTvj/oSTSpkc5JMtVTRmKKr8NUDWAL1Yw7dfGqqPHdHfwwjS3BIvDzYx+hzgtz62RnfNgV+/2MAoNpfX7cAFIHdRzEHNtwugc3RDLquqPoupAE3Y2YRw2T5zG5fILh4vwIcJZL5Uq6B92j8wwJqOex" "33n+vm1NKQ9rxo/UsHAmZlJzpooXcG/4igTBxJyJlamVSRR6N7Nul1v//YJb7J6v2o0iPHW6uE0StzKaPPNC2IVosSRFbD9H2oqppltptFSNPlI0E+t0JBWHem6YK7xcugiO3ImMCaaU8g6Jt/wIDAQAB;s=email;t=s"
+; Required DNS entries
+zftest.testrun.org.  3600  IN  A  135.181.204.127
+zftest.testrun.org.  3600  IN  AAAA  2a01:4f9:c012:52f4::1
+zftest.testrun.org.  3600  IN  MX  10 zftest.testrun.org.
+_mta-sts.zftest.testrun.org.  3600  IN  TXT  "v=STSv1; id=202403211706"
+mta-sts.zftest.testrun.org.  3600  IN  CNAME  zftest.testrun.org.
+www.zftest.testrun.org.  3600  IN  CNAME  zftest.testrun.org.
+opendkim._domainkey.zftest.testrun.org.  3600  IN  TXT  "v=DKIM1;k=rsa;p=MIIBIjANBgkqhkiG9w0BAQEFAAOCAQ8AMIIBCgKCAQEAoYt82CVUyz2ouaqjX2kB+5J80knAyoOU3MGU5aWppmwUwwTvj/oSTSpkc5JMtVTRmKKr8NUDWAL1Yw7dfGqqPHdHfwwjS3BIvDzYx+hzgtz62RnfNgV+/2MAoNpfX7cAFIHdRzEHNtwugc3RDLquqPoupAE3Y2YRw2T5zG5fILh4vwIcJZL5Uq6B92j8wwJqOex" "33n+vm1NKQ9rxo/UsHAmZlJzpooXcG/4igTBxJyJlamVSRR6N7Nul1v//YJb7J6v2o0iPHW6uE0StzKaPPNC2IVosSRFbD9H2oqppltptFSNPlI0E+t0JBWHem6YK7xcugiO3ImMCaaU8g6Jt/wIDAQAB;s=email;t=s"
+
 ; Recommended DNS entries
-_submission._tcp.zftest.testrun.org.  SRV 0 1 587 zftest.testrun.org.
-_submissions._tcp.zftest.testrun.org. SRV 0 1 465 zftest.testrun.org.
-_imap._tcp.zftest.testrun.org.        SRV 0 1 143 zftest.testrun.org.
-_imaps._tcp.zftest.testrun.org.       SRV 0 1 993 zftest.testrun.org.
-zftest.testrun.org.                   CAA 0 issue "letsencrypt.org;accounturi=https://acme-v02.api.letsencrypt.org/acme/acct/1371472956"
-zftest.testrun.org.                   TXT "v=spf1 a:zftest.testrun.org ~all"
-_dmarc.zftest.testrun.org.            TXT "v=DMARC1;p=reject;adkim=s;aspf=s"
-_adsp._domainkey.zftest.testrun.org.  TXT "dkim=discardable"
+zftest.testrun.org.  3600  IN  TXT  "v=spf1 a ~all"
+_dmarc.zftest.testrun.org.  3600  IN  TXT  "v=DMARC1;p=reject;adkim=s;aspf=s"
+zftest.testrun.org.  3600  IN  CAA  0 issue "letsencrypt.org;accounturi=https://acme-v02.api.letsencrypt.org/acme/acct/1371472956"
+_adsp._domainkey.zftest.testrun.org.  3600  IN  TXT  "dkim=discardable"
+_submission._tcp.zftest.testrun.org.  3600  IN  SRV  0 1 587 zftest.testrun.org.
+_submissions._tcp.zftest.testrun.org.  3600  IN  SRV  0 1 465 zftest.testrun.org.
+_imap._tcp.zftest.testrun.org.  3600  IN  SRV  0 1 143 zftest.testrun.org.
+_imaps._tcp.zftest.testrun.org.  3600  IN  SRV  0 1 993 zftest.testrun.org.

cmdeploy/src/cmdeploy/tests/online/benchmark.py

@@ -41,9 +41,9 @@ class TestDC:
 
         def dc_ping_pong():
             chat.send_text("ping")
-            msg = ac2._evtracker.wait_next_incoming_message()
-            msg.chat.send_text("pong")
-            ac1._evtracker.wait_next_incoming_message()
+            msg = ac2.wait_for_incoming_msg()
+            msg.get_snapshot().chat.send_text("pong")
+            ac1.wait_for_incoming_msg()
 
         benchmark(dc_ping_pong, 5)
 
@@ -55,6 +55,6 @@ class TestDC:
             for i in range(10):
                 chat.send_text(f"hello {i}")
             for i in range(10):
-                ac2._evtracker.wait_next_incoming_message()
+                ac2.wait_for_incoming_msg()
 
-        benchmark(dc_send_10_receive_10, 5)
+        benchmark(dc_send_10_receive_10, 5, cooldown="auto")

cmdeploy/src/cmdeploy/tests/online/test_0_qr.py

@@ -20,7 +20,7 @@ def test_fastcgi_working(maildomain, chatmail_config):
 
 
 @pytest.mark.filterwarnings("ignore::urllib3.exceptions.InsecureRequestWarning")
-def test_newemail_configure(maildomain, rpc, chatmail_config):
+def test_newemail_configure(maildomain, maildomain_ip, rpc, chatmail_config):
     """Test configuring accounts by scanning a QR code works."""
     url = f"DCACCOUNT:https://{maildomain}/new"
     for i in range(3):
@@ -30,12 +30,15 @@ def test_newemail_configure(maildomain, rpc, chatmail_config):
             # set_config_from_qr, so fetch credentials via requests instead
             res = requests.post(f"https://{maildomain}/new", verify=False)
             data = res.json()
-            rpc.add_or_update_transport(account_id, {
-                "addr": data["email"],
-                "password": data["password"],
-                "imapServer": maildomain,
-                "smtpServer": maildomain,
-                "certificateChecks": "acceptInvalidCertificates",
-            })
+            rpc.add_or_update_transport(
+                account_id,
+                {
+                    "addr": data["email"],
+                    "password": data["password"],
+                    "imapServer": maildomain_ip,
+                    "smtpServer": maildomain_ip,
+                    "certificateChecks": "acceptInvalidCertificates",
+                },
+            )
         else:
             rpc.add_transport_from_qr(account_id, url)

cmdeploy/src/cmdeploy/tests/online/test_1_basic.py

@@ -7,13 +7,14 @@ import time
 import pytest
 
 from cmdeploy import remote
-from cmdeploy.sshexec import SSHExec
+from cmdeploy.cmdeploy import get_sshexec
 
 
 class TestSSHExecutor:
     @pytest.fixture(scope="class")
-    def sshexec(self, sshdomain):
-        return SSHExec(sshdomain)
+    def sshexec(self, sshdomain, pytestconfig):
+        ssh_config = pytestconfig.getoption("ssh_config")
+        return get_sshexec(sshdomain, ssh_config=ssh_config)
 
     def test_ls(self, sshexec):
         out = sshexec(call=remote.rdns.shell, kwargs=dict(command="ls"))
@@ -27,6 +28,7 @@ class TestSSHExecutor:
         assert res["A"] or res["AAAA"]
 
     def test_logged(self, sshexec, maildomain, capsys):
+        sshexec.verbose = False
         sshexec.logged(
             remote.rdns.perform_initial_checks, kwargs=dict(mail_domain=maildomain)
         )
@@ -52,6 +54,8 @@ class TestSSHExecutor:
                 remote.rdns.perform_initial_checks,
                 kwargs=dict(mail_domain=None),
             )
+        except AssertionError:
+            pass
         except sshexec.FuncError as e:
             assert "rdns.py" in str(e)
             assert "AssertionError" in str(e)
@@ -83,10 +87,8 @@ def test_remote(remote, imap_or_smtp):
 
 
 def test_use_two_chatmailservers(cmfactory, maildomain2):
-    ac1 = cmfactory.new_online_configuring_account(cache=False)
-    cmfactory.switch_maildomain(maildomain2)
-    ac2 = cmfactory.new_online_configuring_account(cache=False)
-    cmfactory.bring_accounts_online()
+    ac1 = cmfactory.get_online_account()
+    ac2 = cmfactory.get_online_account(domain=maildomain2)
     cmfactory.get_accepted_chat(ac1, ac2)
     domain1 = ac1.get_config("addr").split("@")[1]
     domain2 = ac2.get_config("addr").split("@")[1]
@@ -131,11 +133,10 @@ def test_authenticated_from(cmsetup, maildata):
 @pytest.mark.parametrize("from_addr", ["fake@example.org", "fake@testrun.org"])
 def test_reject_missing_dkim(cmsetup, maildata, from_addr):
     domain = cmsetup.maildomain
-    sock = socket.socket(socket.AF_INET, socket.SOCK_STREAM)
-    sock.settimeout(10)
     try:
-        sock.connect((domain, 25))
-    except socket.timeout:
+        sock = socket.create_connection((domain, 25), timeout=10)
+        sock.close()
+    except (socket.timeout, OSError):
         pytest.skip(f"port 25 not reachable for {domain}")
 
     recipient = cmsetup.gen_users(1)[0]
@@ -146,7 +147,7 @@ def test_reject_missing_dkim(cmsetup, maildata, from_addr):
     conn.starttls()
 
     with conn as s:
-        with pytest.raises(smtplib.SMTPDataError, match="No valid DKIM signature"):
+        with pytest.raises(smtplib.SMTPDataError, match="No DKIM signature found"):
             s.sendmail(from_addr=from_addr, to_addrs=recipient.addr, msg=msg)
 
 
@@ -218,7 +219,7 @@ def test_expunged(remote, chatmail_config):
     ]
     outdated_days = int(chatmail_config.delete_large_after) + 1
     find_cmds.append(
-        "find {chatmail_config.mailboxes_dir} -path '*/cur/*' -mtime +{outdated_days} -size +200k -type f"
+        f"find {chatmail_config.mailboxes_dir} -path '*/cur/*' -mtime +{outdated_days} -size +200k -type f"
     )
     for cmd in find_cmds:
         for line in remote.iter_output(cmd):

cmdeploy/src/cmdeploy/tests/online/test_2_deltachat.py

@@ -6,8 +6,8 @@ import imap_tools
 import pytest
 import requests
 
+from cmdeploy.cmdeploy import get_sshexec
 from cmdeploy.remote import rshell
-from cmdeploy.sshexec import SSHExec
 
 
 @pytest.fixture
@@ -27,6 +27,7 @@ class TestMetadataTokens:
 
     def test_set_get_metadata(self, imap_mailbox):
         "set and get metadata token for an account"
+        time.sleep(5)  # make sure Metadata service had a chance to restart
         client = imap_mailbox.client
         client.send(b'a01 SETMETADATA INBOX (/private/devicetoken "1111" )\n')
         res = client.readline()
@@ -62,11 +63,11 @@ class TestEndToEndDeltaChat:
         chat.send_text("message0")
 
         lp.sec("wait for ac2 to receive message")
-        msg2 = ac2._evtracker.wait_next_incoming_message()
-        assert msg2.text == "message0"
+        msg2 = ac2.wait_for_incoming_msg()
+        assert msg2.get_snapshot().text == "message0"
 
     def test_exceed_quota(
-        self, cmfactory, lp, tmpdir, remote, chatmail_config, sshdomain
+        self, cmfactory, lp, tmpdir, remote, chatmail_config, sshdomain, pytestconfig
     ):
         """This is a very slow test as it needs to upload >100MB of mail data
         before quota is exceeded, and thus depends on the speed of the upload.
@@ -91,45 +92,43 @@ class TestEndToEndDeltaChat:
         lp.sec(f"filling remote inbox for {user}")
         fn = f"7743102289.M843172P2484002.c20,S={quota},W=2398:2,"
         path = chatmail_config.mailboxes_dir.joinpath(user, "cur", fn)
-        sshexec = SSHExec(sshdomain)
+        sshexec = get_sshexec(
+            sshdomain, ssh_config=pytestconfig.getoption("ssh_config")
+        )
         sshexec(call=rshell.write_numbytes, kwargs=dict(path=str(path), num=120))
         res = sshexec(call=rshell.dovecot_recalc_quota, kwargs=dict(user=user))
         assert res["percent"] >= 100
 
         lp.sec("ac2: check quota is triggered")
 
-        starting = True
-        for line in remote.iter_output("journalctl -n0 -f -u dovecot"):
-            if starting:
-                chat.send_text("hello")
-                starting = False
+        def send_hello():
+            chat.send_text("hello")
+
+        for line in remote.iter_output(
+            "journalctl -n1 -f -u dovecot", ready=send_hello
+        ):
             if user not in line:
-                # print(line)
                 continue
             if "quota exceeded" in line:
                 return
 
     def test_securejoin(self, cmfactory, lp, maildomain2):
-        ac1 = cmfactory.new_online_configuring_account(cache=False)
-        cmfactory.switch_maildomain(maildomain2)
-        ac2 = cmfactory.new_online_configuring_account(cache=False)
-        cmfactory.bring_accounts_online()
+        ac1 = cmfactory.get_online_account()
+        ac2 = cmfactory.get_online_account(domain=maildomain2)
 
         lp.sec("ac1: create QR code and let ac2 scan it, starting the securejoin")
-        qr = ac1.get_setup_contact_qr()
+        qr = ac1.get_qr_code()
 
         lp.sec("ac2: start QR-code based setup contact protocol")
-        ch = ac2.qr_setup_contact(qr)
+        ch = ac2.secure_join(qr)
         assert ch.id >= 10
-        ac1._evtracker.wait_securejoin_inviter_progress(1000)
+        ac1.wait_for_securejoin_inviter_success()
 
     def test_dkim_header_stripped(self, cmfactory, maildomain2, lp, imap_mailbox):
         """Test that if a DC address receives a message, it has no
         DKIM-Signature and Authentication-Results headers."""
-        ac1 = cmfactory.new_online_configuring_account(cache=False)
-        cmfactory.switch_maildomain(maildomain2)
-        ac2 = cmfactory.new_online_configuring_account(cache=False)
-        cmfactory.bring_accounts_online()
+        ac1 = cmfactory.get_online_account()
+        ac2 = cmfactory.get_online_account(domain=maildomain2)
         chat = cmfactory.get_accepted_chat(ac1, imap_mailbox.dc_ac)
         chat.send_text("message0")
         chat2 = cmfactory.get_accepted_chat(ac2, imap_mailbox.dc_ac)
@@ -146,29 +145,28 @@ class TestEndToEndDeltaChat:
                 assert "dkim-signature" not in msg.headers
 
     def test_read_receipts_between_instances(self, cmfactory, lp, maildomain2):
-        ac1 = cmfactory.new_online_configuring_account(cache=False)
-        cmfactory.switch_maildomain(maildomain2)
-        ac2 = cmfactory.new_online_configuring_account(cache=False)
-        cmfactory.bring_accounts_online()
+        ac1 = cmfactory.get_online_account()
+        ac2 = cmfactory.get_online_account(domain=maildomain2)
 
         lp.sec("setup encrypted comms between ac1 and ac2 on different instances")
-        qr = ac1.get_setup_contact_qr()
-        ch = ac2.qr_setup_contact(qr)
+        qr = ac1.get_qr_code()
+        ch = ac2.secure_join(qr)
         assert ch.id >= 10
-        ac1._evtracker.wait_securejoin_inviter_progress(1000)
+        ac1.wait_for_securejoin_inviter_success()
 
         lp.sec("ac1 sends a message and ac2 marks it as seen")
         chat = ac1.create_chat(ac2)
         msg = chat.send_text("hi")
-        m = ac2._evtracker.wait_next_incoming_message()
+        m = ac2.wait_for_incoming_msg()
         m.mark_seen()
         # we can only indirectly wait for mark-seen to cause an smtp-error
         lp.sec("try to wait for markseen to complete and check error states")
         deadline = time.time() + 3.1
         while time.time() < deadline:
-            msgs = m.chat.get_messages()
+            m_snap = m.get_snapshot()
+            msgs = m_snap.chat.get_messages()
             for msg in msgs:
-                assert "error" not in m.get_message_info()
+                assert "error" not in m.get_info()
             time.sleep(1)
 
 
@@ -180,7 +178,7 @@ def test_hide_senders_ip_address(cmfactory, ssl_context):
     chat = cmfactory.get_accepted_chat(user1, user2)
 
     chat.send_text("testing submission header cleanup")
-    user2._evtracker.wait_next_incoming_message()
+    user2.wait_for_incoming_msg()
     addr = user2.get_config("addr")
     host = addr.split("@")[1]
     pw = user2.get_config("mail_pw")

cmdeploy/src/cmdeploy/tests/online/test_3_status.py

@@ -3,9 +3,16 @@ import os
 from cmdeploy.cmdeploy import main
 
 
-def test_status_cmd(chatmail_config, capsys, request):
+def test_status_cmd(chatmail_config, capsys, request, pytestconfig):
     os.chdir(request.config.invocation_params.dir)
-    assert main(["status"]) == 0
+    command = ["status"]
+    ssh_host = pytestconfig.getoption("ssh_host")
+    if ssh_host:
+        command.extend(["--ssh-host", ssh_host])
+    ssh_config = pytestconfig.getoption("ssh_config")
+    if ssh_config:
+        command.extend(["--ssh-config", ssh_config])
+    assert main(command) == 0
     status_out = capsys.readouterr()
     print(status_out.out)
 

cmdeploy/src/cmdeploy/tests/plugin.py

@@ -1,9 +1,10 @@
 import imaplib
-import io
 import itertools
 import os
 import random
+import re
 import smtplib
+import socket
 import ssl
 import subprocess
 import time
@@ -19,6 +20,76 @@ def pytest_addoption(parser):
     parser.addoption(
         "--slow", action="store_true", default=False, help="also run slow tests"
     )
+    parser.addoption(
+        "--ssh-host",
+        dest="ssh_host",
+        default=None,
+        help="SSH host (overrides mail_domain for SSH operations).",
+    )
+    parser.addoption(
+        "--ssh-config",
+        dest="ssh_config",
+        default=None,
+        help="Path to an SSH config file (e.g. lxconfigs/ssh-config).",
+    )
+
+
+def _parse_ssh_config_hosts(path):
+    """Parse an OpenSSH config file and return a dict of hostname -> IP."""
+    mapping = {}
+    current_names = []
+    for ln in Path(path).read_text().splitlines():
+        line = ln.strip()
+        m = re.match(r"^Host\s+(.+)", line)
+        if m:
+            current_names = m.group(1).split()
+            continue
+        m = re.match(r"^Hostname\s+(\S+)", line)
+        if m and current_names:
+            ip = m.group(1)
+            for name in current_names:
+                mapping[name] = ip
+            current_names = []
+    return mapping
+
+
+_original_getaddrinfo = socket.getaddrinfo
+
+
+def _make_patched_getaddrinfo(host_map):
+    """Return a getaddrinfo that resolves hosts in host_map to their IPs."""
+
+    def patched_getaddrinfo(host, port, family=0, type=0, proto=0, flags=0):
+        if host in host_map:
+            ip = host_map[host]
+            return _original_getaddrinfo(ip, port, family, type, proto, flags)
+        return _original_getaddrinfo(host, port, family, type, proto, flags)
+
+    return patched_getaddrinfo
+
+
+@pytest.fixture(autouse=True, scope="session")
+def _setup_localchat_dns(pytestconfig):
+    """Monkey-patch socket.getaddrinfo to resolve .localchat via ssh-config."""
+    ssh_config = pytestconfig.getoption("ssh_config")
+    if not ssh_config or not Path(ssh_config).exists():
+        yield {}
+        return
+    host_map = _parse_ssh_config_hosts(ssh_config)
+    if not host_map:
+        yield {}
+        return
+    socket.getaddrinfo = _make_patched_getaddrinfo(host_map)
+    try:
+        yield host_map
+    finally:
+        socket.getaddrinfo = _original_getaddrinfo
+
+
+@pytest.fixture(scope="session")
+def ssh_config_host_map(_setup_localchat_dns):
+    """Return the host-name → IP map parsed from ssh-config."""
+    return _setup_localchat_dns
 
 
 def pytest_configure(config):
@@ -35,17 +106,29 @@ def pytest_runtest_setup(item):
             pytest.skip("skipping slow test, use --slow to run")
 
 
-@pytest.fixture(scope="session")
-def chatmail_config(pytestconfig):
-    current = basedir = Path().resolve()
+def _get_chatmail_config():
+    ini = os.environ.get("CHATMAIL_INI")
+    if ini:
+        path = Path(ini).resolve()
+        if path.exists():
+            return read_config(path), path
+    current = Path().resolve()
     while 1:
         path = current.joinpath("chatmail.ini").resolve()
         if path.exists():
-            return read_config(path)
+            return read_config(path), path
         if current == current.parent:
             break
         current = current.parent
+    return None, None
 
+
+@pytest.fixture(scope="session")
+def chatmail_config(pytestconfig):
+    config, path = _get_chatmail_config()
+    if config:
+        return config
+    basedir = Path().resolve()
     pytest.skip(f"no chatmail.ini file found in {basedir} or parent dirs")
 
 
@@ -55,8 +138,14 @@ def maildomain(chatmail_config):
 
 
 @pytest.fixture(scope="session")
-def sshdomain(maildomain):
-    return os.environ.get("CHATMAIL_SSH", maildomain)
+def sshdomain(maildomain, pytestconfig):
+    return pytestconfig.getoption("ssh_host") or maildomain
+
+
+@pytest.fixture(scope="session")
+def maildomain_ip(maildomain, ssh_config_host_map):
+    """Return the IP for maildomain from ssh-config, or maildomain itself."""
+    return ssh_config_host_map.get(maildomain, maildomain)
 
 
 @pytest.fixture
@@ -73,10 +162,17 @@ def sshdomain2(maildomain2):
 
 
 def pytest_report_header():
-    domain = os.environ.get("CHATMAIL_DOMAIN")
-    if domain:
-        text = f"chatmail test instance: {domain}"
-        return ["-" * len(text), text, "-" * len(text)]
+    config, path = _get_chatmail_config()
+    domain2 = os.environ.get("CHATMAIL_DOMAIN2", "NOT SET")
+    domain = config.mail_domain if config else "NOT SET"
+    path = path if path else "NOT SET"
+
+    lines = [
+        f"chatmail.ini {domain} location: {path}",
+        f"chatmail2: {domain2}",
+    ]
+    sep = "-" * max(map(len, lines))
+    return [sep, *lines, sep]
 
 
 @pytest.fixture
@@ -91,15 +187,22 @@ def cm_data(request):
 
 
 @pytest.fixture
-def benchmark(request):
-    def bench(func, num, name=None, reportfunc=None):
+def benchmark(request, chatmail_config):
+    def bench(func, num, name=None, reportfunc=None, cooldown=0.0):
         if name is None:
             name = func.__name__
+        if cooldown == "auto":
+            per_minute = max(chatmail_config.max_user_send_per_minute, 1)
+            cooldown = chatmail_config.max_user_send_burst_size * 60 / per_minute
+
         durations = []
         for i in range(num):
             now = time.time()
             func()
             durations.append(time.time() - now)
+            if cooldown > 0 and i + 1 < num:
+                # Keep post-run cooldown out of measured benchmark duration.
+                time.sleep(cooldown)
         durations.sort()
         request.config._benchresults[name] = (reportfunc, durations)
 
@@ -276,102 +379,146 @@ def gencreds(chatmail_config):
 
 
 #
-# Delta Chat testplugin re-use
+# Delta Chat RPC-based test support
 # use the cmfactory fixture to get chatmail instance accounts
 #
 
+from deltachat_rpc_client import DeltaChat, Rpc
 
-class ChatmailTestProcess:
-    """Provider for chatmail instance accounts as used by deltachat.testplugin.acfactory"""
 
-    def __init__(self, pytestconfig, maildomain, gencreds, chatmail_config):
-        self.pytestconfig = pytestconfig
-        self.maildomain = maildomain
-        assert "." in self.maildomain, maildomain
+class ChatmailACFactory:
+    """RPC-based account factory for chatmail testing."""
+
+    def __init__(
+        self,
+        rpc,
+        maildomain,
+        maildomain_ip,
+        gencreds,
+        chatmail_config,
+        ssh_config_host_map,
+    ):
+        self.dc = DeltaChat(rpc)
+        self.rpc = rpc
+        self._maildomain = maildomain
+        self._maildomain_ip = maildomain_ip
         self.gencreds = gencreds
         self.chatmail_config = chatmail_config
-        self._addr2files = {}
+        self._ssh_config_host_map = ssh_config_host_map
 
-    def get_liveconfig_producer(self):
-        while 1:
-            user, password = self.gencreds(self.maildomain)
-            config = {
-                "addr": user,
-                "mail_pw": password,
-            }
-            # speed up account configuration
-            config["mail_server"] = self.maildomain
-            config["send_server"] = self.maildomain
-            if self.chatmail_config.tls_cert_mode == "self":
-                # Accept self-signed TLS certificates
-                config["imap_certificate_checks"] = "3"
-            yield config
+    def _make_transport(self, domain):
+        """Build a transport config dict for the given domain."""
+        addr, password = self.gencreds(domain)
+        server = self._ssh_config_host_map.get(domain, domain)
+        transport = {
+            "addr": addr,
+            "password": password,
+            "imapServer": server,
+            "smtpServer": server,
+        }
+        if self.chatmail_config.tls_cert_mode == "self":
+            transport["certificateChecks"] = "acceptInvalidCertificates"
+        return transport
 
-    def cache_maybe_retrieve_configured_db_files(self, cache_addr, db_target_path):
-        pass
+    def get_online_account(self, domain=None):
+        """Create, configure and bring online a single account."""
+        return self.get_online_accounts(1, domain)[0]
 
-    def cache_maybe_store_configured_db_files(self, acc):
-        pass
+    def get_online_accounts(self, num, domain=None):
+        """Create multiple online accounts in parallel."""
+        domain = domain or self._maildomain
+        futures = []
+        accounts = []
+        for _ in range(num):
+            account = self.dc.add_account()
+            future = account.add_or_update_transport.future(
+                self._make_transport(domain)
+            )
+            futures.append(future)
+
+            # ensure messages stay in INBOX so that they can be
+            # concurrently fetched via extra IMAP connections during tests
+            account.set_config("delete_server_after", "10")
+            accounts.append(account)
+
+        for future in futures:
+            future()
+
+        for account in accounts:
+            account.bring_online()
+        return accounts
+
+    def get_accepted_chat(self, ac1, ac2):
+        """Create a 1:1 chat between ac1 and ac2 accepted on both sides."""
+        ac2.create_chat(ac1)
+        return ac1.create_chat(ac2)
+
+
+@pytest.fixture(scope="session")
+def rpc(tmp_path_factory):
+    """Start a deltachat-rpc-server process for the test session."""
+
+    # NB: accounts_dir must NOT already exist as directory --
+    # core-rust only creates accounts.toml if the dir doesn't exist yet.
+    accounts_dir = str(tmp_path_factory.mktemp("dc") / "accounts")
+    rpc = Rpc(accounts_dir=accounts_dir)
+    rpc.start()
+    yield rpc
+    rpc.close()
 
 
 @pytest.fixture
-def cmfactory(request, gencreds, tmpdir, maildomain, chatmail_config):
-    # cloned from deltachat.testplugin.amfactory
-    pytest.importorskip("deltachat")
-    from deltachat.testplugin import ACFactory
-
-    testproc = ChatmailTestProcess(
-        request.config, maildomain, gencreds, chatmail_config
+def cmfactory(
+    rpc, gencreds, maildomain, maildomain_ip, chatmail_config, ssh_config_host_map
+):
+    """Return a ChatmailACFactory for creating online Delta Chat accounts."""
+    return ChatmailACFactory(
+        rpc=rpc,
+        maildomain=maildomain,
+        maildomain_ip=maildomain_ip,
+        gencreds=gencreds,
+        chatmail_config=chatmail_config,
+        ssh_config_host_map=ssh_config_host_map,
     )
 
-    class Data:
-        def read_path(self, path):
-            return
-
-    am = ACFactory(request=request, tmpdir=tmpdir, testprocess=testproc, data=Data())
-
-    # Skip upstream's init_imap to prevent extra imap connections not
-    # needed for relay testing
-    am._acsetup.init_imap = lambda acc: None
-
-    # nb. a bit hacky
-    # would probably be better if deltachat's test machinery grows native support
-    def switch_maildomain(maildomain2):
-        am.testprocess.maildomain = maildomain2
-
-    am.switch_maildomain = switch_maildomain
-
-    yield am
-    if hasattr(request.node, "rep_call") and request.node.rep_call.failed:
-        if testproc.pytestconfig.getoption("--extra-info"):
-            logfile = io.StringIO()
-            am.dump_imap_summary(logfile=logfile)
-            print(logfile.getvalue())
-            # request.node.add_report_section("call", "imap-server-state", s)
-
 
 @pytest.fixture
-def remote(sshdomain):
-    return Remote(sshdomain)
+def remote(sshdomain, pytestconfig):
+    return Remote(sshdomain, ssh_config=pytestconfig.getoption("ssh_config"))
 
 
 class Remote:
-    def __init__(self, sshdomain):
+    def __init__(self, sshdomain, ssh_config=None):
         self.sshdomain = sshdomain
+        self.ssh_config = ssh_config
 
-    def iter_output(self, logcmd=""):
+    def iter_output(self, logcmd="", ready=None):
         getjournal = "journalctl -f" if not logcmd else logcmd
+        print(self.sshdomain)
+        match self.sshdomain:
+            case "@local":
+                command = []
+            case "localhost":
+                command = []
+            case _:
+                command = ["ssh"]
+                if self.ssh_config:
+                    command.extend(["-F", self.ssh_config])
+                command.append(f"root@{self.sshdomain}")
+        [command.append(arg) for arg in getjournal.split()]
         self.popen = subprocess.Popen(
-            ["ssh", f"root@{self.sshdomain}", getjournal],
+            command,
             stdout=subprocess.PIPE,
         )
         while 1:
             line = self.popen.stdout.readline()
             res = line.decode().strip().lower()
-            if res:
-                yield res
-            else:
+            if not res:
                 break
+            if ready is not None:
+                ready()
+                ready = None
+            yield res
 
 
 @pytest.fixture

cmdeploy/src/cmdeploy/tests/setup_tls_external.py

@@ -1,362 +0,0 @@
-"""Setup and verify external TLS certificates for a chatmail server.
-
-Generates a self-signed TLS certificate, uploads it to the chatmail
-server via SCP, runs ``cmdeploy run``, and then probes all TLS-enabled
-ports (nginx, postfix, dovecot) to verify the certificate is actually
-served.  After probing, checks remote service logs for errors.
-
-Prerequisites
-~~~~~~~~~~~~~
-- SSH root access to the target server (same as ``cmdeploy run``)
-- ``cmdeploy`` in PATH (activate the venv first)
-
-How to run
-~~~~~~~~~~
-From the repository root::
-
-    # Full run: generate cert, deploy, probe ports, check services
-    python -m cmdeploy.tests.setup_tls_external DOMAIN
-
-    # Re-probe only (after a previous deploy)
-    python -m cmdeploy.tests.setup_tls_external DOMAIN \\
-        --skip-deploy --skip-certgen
-
-    # Override SSH host (e.g. when domain doesn't resolve to the server)
-    python -m cmdeploy.tests.setup_tls_external DOMAIN \\
-        --ssh-host staging-ipv4.testrun.org
-
-Arguments
-~~~~~~~~~
-DOMAIN            mail domain for the chatmail server (SSH root login must work)
-
-Options
-~~~~~~~
---skip-deploy     skip ``cmdeploy run``, only probe ports
---skip-certgen    skip cert generation/upload, use certs already on server
---ssh-host HOST   SSH host override (defaults to DOMAIN)
-"""
-
-import argparse
-import shutil
-import smtplib
-import socket
-import ssl
-import subprocess
-import sys
-import tempfile
-import time
-from pathlib import Path
-
-# Cert paths on the remote server
-REMOTE_CERT = "/etc/ssl/certs/tmp_fullchain.pem"
-REMOTE_KEY = "/etc/ssl/private/tmp_privkey.pem"
-
-
-# ---------------------------------------------------------------------------
-# Config generation
-# ---------------------------------------------------------------------------
-
-
-def generate_config(domain: str, config_dir: Path) -> Path:
-    """Generate a chatmail.ini with tls_external_cert_and_key for *domain*."""
-    from chatmaild.config import write_initial_config
-
-    ini_path = config_dir / "chatmail.ini"
-    write_initial_config(
-        ini_path,
-        domain,
-        overrides={
-            "tls_external_cert_and_key": f"{REMOTE_CERT} {REMOTE_KEY}",
-        },
-    )
-    print(f"[+] Generated chatmail.ini for {domain} in {config_dir}")
-    return ini_path
-
-
-# ---------------------------------------------------------------------------
-# Certificate generation
-# ---------------------------------------------------------------------------
-
-
-def generate_cert(domain: str, cert_dir: Path) -> tuple:
-    """Generate a self-signed TLS cert+key for *domain* with proper SANs."""
-    from cmdeploy.selfsigned.deployer import openssl_selfsigned_args
-
-    cert_path = cert_dir / "fullchain.pem"
-    key_path = cert_dir / "privkey.pem"
-    subprocess.check_call(openssl_selfsigned_args(domain, cert_path, key_path, days=30))
-    print(f"[+] Generated cert for {domain} in {cert_dir}")
-    return cert_path, key_path
-
-
-# ---------------------------------------------------------------------------
-# Upload certs to remote server
-# ---------------------------------------------------------------------------
-
-
-def upload_certs(
-    ssh_host: str,
-    cert_path: Path,
-    key_path: Path,
-) -> None:
-    """SCP cert and key to the remote server."""
-    subprocess.check_call([
-        "scp", str(cert_path), f"root@{ssh_host}:{REMOTE_CERT}",
-    ])
-    subprocess.check_call([
-        "scp", str(key_path), f"root@{ssh_host}:{REMOTE_KEY}",
-    ])
-    # Ensure cert is world-readable and key is readable by ssl-cert group
-    # (dovecot/postfix/nginx need to read these files)
-    subprocess.check_call([
-        "ssh", f"root@{ssh_host}",
-        f"chmod 644 {REMOTE_CERT} && chmod 640 {REMOTE_KEY}"
-        f" && chgrp ssl-cert {REMOTE_KEY}",
-    ])
-    print(f"[+] Uploaded cert/key to {ssh_host}")
-
-
-# ---------------------------------------------------------------------------
-# Deploy
-# ---------------------------------------------------------------------------
-
-
-def run_deploy(ini_path: str) -> None:
-    """Run ``cmdeploy run --skip-dns-check --config <ini>``."""
-    cmd = ["cmdeploy", "run", "--config", str(ini_path), "--skip-dns-check"]
-    print(f"[+] Running: {' '.join(cmd)}")
-    subprocess.check_call(cmd)
-    print("[+] Deploy completed successfully")
-
-
-# ---------------------------------------------------------------------------
-# TLS port probing
-# ---------------------------------------------------------------------------
-
-
-def get_peer_cert_binary(host: str, port: int) -> bytes:
-    """Connect to host:port with TLS and return the DER-encoded peer cert."""
-    ctx = ssl.create_default_context()
-    ctx.check_hostname = False
-    ctx.verify_mode = ssl.CERT_NONE
-    with socket.create_connection((host, port), timeout=15) as sock:
-        with ctx.wrap_socket(sock, server_hostname=host) as ssock:
-            return ssock.getpeercert(binary_form=True)
-
-
-def get_smtp_starttls_cert_binary(host: str, port: int = 587) -> bytes:
-    """Connect via SMTP STARTTLS and return the DER cert."""
-    ctx = ssl.create_default_context()
-    ctx.check_hostname = False
-    ctx.verify_mode = ssl.CERT_NONE
-    with smtplib.SMTP(host, port, timeout=15) as smtp:
-        smtp.starttls(context=ctx)
-        return smtp.sock.getpeercert(binary_form=True)
-
-
-def check_cert_matches(
-    label: str, served_der: bytes, expected_der: bytes,
-) -> bool:
-    """Compare served DER cert against the expected cert."""
-    if served_der == expected_der:
-        print(f"  [OK]   {label}: certificate matches")
-        return True
-    else:
-        print(f"  [FAIL] {label}: certificate does NOT match")
-        return False
-
-
-def load_cert_der(cert_pem_path: Path) -> bytes:
-    """Load a PEM cert file and return its DER encoding."""
-    pem_text = cert_pem_path.read_text()
-    start = pem_text.index("-----BEGIN CERTIFICATE-----")
-    end = pem_text.index("-----END CERTIFICATE-----") + len(
-        "-----END CERTIFICATE-----"
-    )
-    return ssl.PEM_cert_to_DER_cert(pem_text[start:end])
-
-
-def probe_all_ports(host: str, expected_cert_der: bytes) -> bool:
-    """Probe TLS ports and verify the served certificate matches.
-
-    Checks ports 993 (IMAP), 465 (SMTPS), 587 (STARTTLS), and 443
-    (nginx stream).  Port 8443 is skipped as nginx binds it to
-    localhost behind the stream proxy on 443.
-    """
-    print(f"\n[+] Probing TLS ports on {host}...")
-    all_ok = True
-
-    for label, port in [
-        ("IMAP/TLS (993)", 993),
-        ("SMTP/TLS (465)", 465),
-    ]:
-        try:
-            served = get_peer_cert_binary(host, port)
-            if not check_cert_matches(label, served, expected_cert_der):
-                all_ok = False
-        except Exception as e:
-            print(f"  [FAIL] {label}: connection failed: {e}")
-            all_ok = False
-
-    # STARTTLS on port 587
-    try:
-        served = get_smtp_starttls_cert_binary(host, 587)
-        if not check_cert_matches("SMTP/STARTTLS (587)", served, expected_cert_der):
-            all_ok = False
-    except Exception as e:
-        print(f"  [FAIL] SMTP/STARTTLS (587): connection failed: {e}")
-        all_ok = False
-
-    # Port 443 (nginx stream proxy with ALPN routing)
-    try:
-        served = get_peer_cert_binary(host, 443)
-        if not check_cert_matches("nginx/443 (stream)", served, expected_cert_der):
-            all_ok = False
-    except Exception as e:
-        print(f"  [FAIL] nginx/443 (stream): connection failed: {e}")
-        all_ok = False
-
-    return all_ok
-
-
-# ---------------------------------------------------------------------------
-# Post-deploy service health checks
-# ---------------------------------------------------------------------------
-
-SERVICES = ["dovecot", "postfix", "nginx"]
-
-
-def check_remote_services(ssh_host: str, since: str = "") -> bool:
-    """SSH to the server and check for service failures or errors.
-
-    *since* is a ``journalctl --since`` timestamp (e.g. ``"5 min ago"``).
-    If empty, checks the entire boot journal.
-    """
-    print(f"\n[+] Checking remote service health on {ssh_host}...")
-    all_ok = True
-
-    for svc in SERVICES:
-        try:
-            result = subprocess.run(
-                ["ssh", f"root@{ssh_host}",
-                 f"systemctl is-active {svc}.service"],
-                capture_output=True, text=True, timeout=15, check=False,
-            )
-            status = result.stdout.strip()
-            if status == "active":
-                print(f"  [OK]   {svc}: active")
-            else:
-                print(f"  [FAIL] {svc}: {status}")
-                all_ok = False
-        except Exception as e:
-            print(f"  [FAIL] {svc}: check failed: {e}")
-            all_ok = False
-
-    since_arg = f'--since="{since}"' if since else ""
-    print(f"\n[+] Checking journal for errors on {ssh_host}...")
-    for svc in SERVICES:
-        try:
-            result = subprocess.run(
-                ["ssh", f"root@{ssh_host}",
-                 f"journalctl -u {svc}.service {since_arg}"
-                 f" --no-pager -p err -q"],
-                capture_output=True, text=True, timeout=15, check=False,
-            )
-            errors = result.stdout.strip()
-            if errors:
-                print(f"  [WARN] {svc} errors in journal:")
-                for line in errors.splitlines()[:10]:
-                    print(f"         {line}")
-                all_ok = False
-            else:
-                print(f"  [OK]   {svc}: no errors in journal")
-        except Exception as e:
-            print(f"  [FAIL] {svc}: journal check failed: {e}")
-            all_ok = False
-
-    return all_ok
-
-
-# ---------------------------------------------------------------------------
-# Main
-# ---------------------------------------------------------------------------
-
-
-def main():
-    parser = argparse.ArgumentParser(
-        description=__doc__,
-        formatter_class=argparse.RawDescriptionHelpFormatter,
-    )
-    parser.add_argument(
-        "domain",
-        help="mail domain (SSH root login must work to this host)",
-    )
-    parser.add_argument(
-        "--skip-deploy",
-        action="store_true",
-        help="skip cmdeploy run, only probe ports",
-    )
-    parser.add_argument(
-        "--skip-certgen",
-        action="store_true",
-        help="skip cert generation and upload (use existing)",
-    )
-    parser.add_argument(
-        "--ssh-host",
-        help="SSH host override (defaults to DOMAIN)",
-    )
-    args = parser.parse_args()
-
-    domain = args.domain
-    ssh_host = args.ssh_host or domain
-    print(f"[+] Domain: {domain}")
-    print(f"[+] SSH host: {ssh_host}")
-    print(f"[+] Remote cert: {REMOTE_CERT}")
-    print(f"[+] Remote key:  {REMOTE_KEY}")
-
-    work_dir = Path(tempfile.mkdtemp(prefix="tls-external-test-"))
-    try:
-        # Generate chatmail.ini
-        ini_path = generate_config(domain, work_dir)
-
-        if not args.skip_certgen:
-            local_cert, local_key = generate_cert(domain, work_dir)
-            upload_certs(ssh_host, local_cert, local_key)
-        else:
-            local_cert = work_dir / "fullchain.pem"
-            subprocess.check_call([
-                "scp", f"root@{ssh_host}:{REMOTE_CERT}", str(local_cert),
-            ])
-
-        # Record timestamp before deploy for journal filtering
-        deploy_start = time.strftime("%Y-%m-%d %H:%M:%S")
-
-        if not args.skip_deploy:
-            run_deploy(ini_path)
-
-        # Probe TLS ports
-        expected_der = load_cert_der(local_cert)
-        ports_ok = probe_all_ports(domain, expected_der)
-
-        # Check service health (only errors since deploy started)
-        services_ok = check_remote_services(ssh_host, since=deploy_start)
-
-        if ports_ok and services_ok:
-            print(
-                "\n[SUCCESS] All TLS port probes passed and services are healthy"
-            )
-            return 0
-        else:
-            if not ports_ok:
-                print("\n[FAILURE] Some TLS port probes failed", file=sys.stderr)
-            if not services_ok:
-                print(
-                    "\n[FAILURE] Some services have errors", file=sys.stderr
-                )
-            return 1
-    finally:
-        shutil.rmtree(work_dir, ignore_errors=True)
-
-
-if __name__ == "__main__":
-    sys.exit(main())

cmdeploy/src/cmdeploy/tests/test_cmdeploy.py

@@ -23,7 +23,10 @@ class TestCmdline:
         run = parser.parse_args(["run"])
         assert init and run
 
-    def test_init_not_overwrite(self, capsys):
+    def test_init_not_overwrite(self, tmp_path, capsys, monkeypatch):
+        monkeypatch.delenv("CHATMAIL_INI", raising=False)
+        monkeypatch.chdir(tmp_path)
+
         assert main(["init", "chat.example.org"]) == 0
         capsys.readouterr()
 

cmdeploy/src/cmdeploy/tests/test_dns.py

@@ -3,7 +3,7 @@ from copy import deepcopy
 import pytest
 
 from cmdeploy import remote
-from cmdeploy.dns import check_full_zone, check_initial_remote_data
+from cmdeploy.dns import check_full_zone, check_initial_remote_data, parse_zone_records
 
 
 @pytest.fixture
@@ -60,6 +60,29 @@ def mockdns(request, mockdns_base, mockdns_expected):
     return mockdns_base
 
 
+class TestGetDkimEntry:
+    def test_dkim_entry_returns_tuple_on_success(self, mockdns):
+        entry, web_entry = remote.rdns.get_dkim_entry(
+            "some.domain", "", dkim_selector="opendkim"
+        )
+        # May return None,None if openssl not available, but should never crash
+        if entry is not None:
+            assert "opendkim._domainkey.some.domain" in entry
+            assert "opendkim._domainkey.some.domain" in web_entry
+
+    def test_dkim_entry_returns_none_tuple_on_error(self, monkeypatch):
+        """CalledProcessError must return (None, None), not bare None."""
+        from subprocess import CalledProcessError
+
+        def failing_shell(command, fail_ok=False, print=print):
+            raise CalledProcessError(1, command)
+
+        monkeypatch.setattr(remote.rdns, "shell", failing_shell)
+        result = remote.rdns.get_dkim_entry("some.domain", "", dkim_selector="opendkim")
+        assert result == (None, None)
+        assert result[0] is None and result[1] is None
+
+
 class TestPerformInitialChecks:
     def test_perform_initial_checks_ok1(self, mockdns, mockdns_expected):
         remote_data = remote.rdns.perform_initial_checks("some.domain")
@@ -102,18 +125,33 @@ class TestPerformInitialChecks:
         assert not l
 
 
+def test_parse_zone_records():
+    text = """
+    ; This is a comment
+    some.domain. 3600 IN A 1.1.1.1
+
+    ; Another comment
+    www.some.domain. 3600 IN CNAME some.domain.
+    """
+    records = list(parse_zone_records(text))
+    assert records == [
+        ("some.domain", "3600", "A", "1.1.1.1"),
+        ("www.some.domain", "3600", "CNAME", "some.domain."),
+    ]
+
+
+def test_parse_zone_records_invalid_line():
+    text = "invalid line"
+    with pytest.raises(ValueError, match="Bad zone record line"):
+        list(parse_zone_records(text))
+
+
 def parse_zonefile_into_dict(zonefile, mockdns_base, only_required=False):
-    for zf_line in zonefile.split("\n"):
-        if zf_line.startswith("#"):
-            if "Recommended" in zf_line and only_required:
-                return
-            continue
-        if not zf_line.strip():
-            continue
-        zf_domain, zf_typ, zf_value = zf_line.split(maxsplit=2)
-        zf_domain = zf_domain.rstrip(".")
-        zf_value = zf_value.strip()
-        mockdns_base.setdefault(zf_typ, {})[zf_domain] = zf_value
+    if only_required:
+        # Only take records before the "; Recommended" section
+        zonefile = zonefile.split("; Recommended")[0]
+    for name, ttl, rtype, rdata in parse_zone_records(zonefile):
+        mockdns_base.setdefault(rtype, {})[name] = rdata
 
 
 class MockSSHExec:

cmdeploy/src/cmdeploy/tests/test_lxc.py

@@ -0,0 +1,173 @@
+"""Tests for cmdeploy lxc-* subcommands."""
+
+import shutil
+import subprocess
+import sys
+
+import pytest
+
+from cmdeploy.lxc import cli
+from cmdeploy.lxc.incus import Incus
+
+pytestmark = pytest.mark.skipif(
+    not shutil.which("incus") or not shutil.which("lxc"),
+    reason="incus/lxc not installed",
+)
+
+
+# ---------------------------------------------------------------------------
+# Fixtures
+# ---------------------------------------------------------------------------
+
+
+@pytest.fixture
+def ix():
+    return Incus()
+
+
+@pytest.fixture(scope="session")
+def lxc_setup():
+    ix = Incus()
+    ix.get_dns_container().ensure()
+    return ix.list_managed()
+
+
+@pytest.fixture(scope="session")
+def relay_container(lxc_setup):
+    test_names = {f"{n}-localchat" for n in cli.RELAY_NAMES}
+    relays = [c for c in lxc_setup if c["name"] in test_names and c.get("ip")]
+    if not relays:
+        pytest.skip("no test relay containers running")
+    return relays[0]
+
+
+@pytest.fixture
+def cmdeploy():
+    def run(*args):
+        return subprocess.run(
+            [sys.executable, "-m", "cmdeploy.cmdeploy", *args],
+            capture_output=True,
+            text=True,
+            check=False,
+        )
+
+    return run
+
+
+# ---------------------------------------------------------------------------
+# Tests
+# ---------------------------------------------------------------------------
+
+
+@pytest.mark.parametrize(
+    "subcmd, expected, absent",
+    [
+        (None, ["lxc-start", "lxc-stop", "lxc-test", "lxc-status"], ["lxc-destroy"]),
+        ("lxc-start", ["--ipv4-only", "--run"], ["--config"]),
+        ("lxc-stop", ["--destroy", "--destroy-all"], ["--config"]),
+        ("lxc-test", ["--one"], ["--config"]),
+        ("lxc-status", [], ["--config"]),
+        ("run", ["--ssh-config"], ["--lxc"]),
+        ("dns", ["--ssh-config"], []),
+        ("test", ["--ssh-config"], []),
+        ("status", ["--ssh-config"], []),
+    ],
+)
+def test_help_options(cmdeploy, subcmd, expected, absent):
+    args = [subcmd, "--help"] if subcmd else ["--help"]
+    result = cmdeploy(*args)
+    output = result.stdout + result.stderr
+    assert result.returncode == 0
+    for flag in expected:
+        assert flag in output
+    for flag in absent:
+        assert flag not in output
+
+
+class TestSSHConfig:
+    def test_lxconfigs(self, ix, lxc_setup):
+        d = ix.lxconfigs_dir
+        assert d.name == "lxconfigs"
+        assert d.exists()
+        path = ix.ssh_config_path
+        assert path.name == "ssh-config"
+        assert path.parent.name == "lxconfigs"
+
+    def test_write_ssh_config(self, ix, lxc_setup):
+        path = ix.write_ssh_config()
+        assert path.exists()
+        text = path.read_text()
+
+        for c in lxc_setup:
+            if c.get("ip"):
+                assert c["name"] in text
+                assert f"Hostname {c['ip']}" in text
+
+        assert "User root" in text
+        assert "IdentityFile" in text
+        assert "StrictHostKeyChecking accept-new" in text
+
+
+def test_dns(ix, relay_container):
+    def dig(qname, qtype):
+        ct = ix.get_dns_container()
+        return ct.bash(f"dig @127.0.0.1 {qname} {qtype} +short").strip()
+
+    domain = relay_container["domain"]
+    assert dig(domain, "A") == relay_container["ip"]
+    assert domain in dig(domain, "MX")
+    assert "587" in dig(f"_submission._tcp.{domain}", "SRV")
+
+
+class TestLxcStatus:
+    def test_cli_lxc_status_help(self, cmdeploy):
+        result = cmdeploy("lxc-status", "--help")
+        assert result.returncode == 0
+        assert "status" in result.stdout.lower()
+
+    def test_shows_containers(self, lxc_setup, capsys):
+        from cmdeploy.cmdeploy import Out
+
+        class QuietOut(Out):
+            def red(self, msg, **kw):
+                pass
+
+            def green(self, msg, **kw):
+                pass
+
+        ret = cli.lxc_status_cmd(None, QuietOut())
+        assert ret == 0
+        captured = capsys.readouterr().out
+        assert "ns-localchat" in captured
+        assert "running" in captured
+
+    def test_deploy_freshness(self, ix, monkeypatch):
+        ct = ix.get_container("x")
+
+        monkeypatch.setattr(
+            "cmdeploy.lxc.incus.RelayContainer.deployed_version",
+            lambda _self: "abc123def456",
+        )
+        monkeypatch.setattr(
+            "cmdeploy.lxc.incus.RelayContainer.deployed_domain",
+            lambda _self: ct.domain,
+        )
+        monkeypatch.setattr(
+            "cmdeploy.lxc.cli.get_version_string",
+            lambda: "abc123def456",
+        )
+        assert "IN-SYNC" in cli._deploy_status(ct, "abc123def456", ix)
+        assert "STALE" in cli._deploy_status(ct, "other_hash_here", ix)
+
+        # Hash matches but local has uncommitted changes
+        monkeypatch.setattr(
+            "cmdeploy.lxc.cli.get_version_string",
+            lambda: "abc123def456\ndiff --git a/foo",
+        )
+        assert "DIRTY" in cli._deploy_status(ct, "abc123def456", ix)
+
+        monkeypatch.setattr(
+            "cmdeploy.lxc.incus.RelayContainer.deployed_version",
+            lambda _self: None,
+        )
+        assert "NOT DEPLOYED" in cli._deploy_status(ct, "abc123", ix)

cmdeploy/src/cmdeploy/tests/test_rshell.py

@@ -0,0 +1,68 @@
+from unittest.mock import patch
+
+from cmdeploy.remote.rshell import dovecot_recalc_quota
+
+
+def test_dovecot_recalc_quota_normal_output():
+    """Normal doveadm output returns parsed dict."""
+    normal_output = (
+        "Quota name Type    Value  Limit  %\n"
+        "User quota STORAGE     5 102400  0\n"
+        "User quota MESSAGE     2      -  0\n"
+    )
+
+    with patch("cmdeploy.remote.rshell.shell", return_value=normal_output):
+        result = dovecot_recalc_quota("user@example.org")
+
+    # shell is called twice (recalc + get), patch returns same for both
+    assert result == {"value": 5, "limit": 102400, "percent": 0}
+
+
+def test_dovecot_recalc_quota_empty_output():
+    """Empty doveadm output (trailing newline) must not IndexError."""
+    call_count = [0]
+
+    def mock_shell(cmd):
+        call_count[0] += 1
+        if "recalc" in cmd:
+            return ""
+        # quota get returns only empty lines
+        return "\n\n"
+
+    with patch("cmdeploy.remote.rshell.shell", side_effect=mock_shell):
+        result = dovecot_recalc_quota("user@example.org")
+
+    assert result is None
+
+
+def test_dovecot_recalc_quota_malformed_output():
+    """Malformed output with too few columns must not crash."""
+    call_count = [0]
+
+    def mock_shell(cmd):
+        call_count[0] += 1
+        if "recalc" in cmd:
+            return ""
+        # partial line, fewer than 6 parts
+        return "Quota name\nUser quota STORAGE\n"
+
+    with patch("cmdeploy.remote.rshell.shell", side_effect=mock_shell):
+        result = dovecot_recalc_quota("user@example.org")
+
+    assert result is None
+
+
+def test_dovecot_recalc_quota_header_only():
+    """Only header line, no data rows."""
+    call_count = [0]
+
+    def mock_shell(cmd):
+        call_count[0] += 1
+        if "recalc" in cmd:
+            return ""
+        return "Quota name Type    Value  Limit  %\n"
+
+    with patch("cmdeploy.remote.rshell.shell", side_effect=mock_shell):
+        result = dovecot_recalc_quota("user@example.org")
+
+    assert result is None

cmdeploy/src/cmdeploy/tests/test_util.py

@@ -0,0 +1,97 @@
+import pytest
+
+from cmdeploy.util import (
+    build_chatmaild_sdist,
+    collapse,
+    get_chatmaild_sdist,
+    get_git_hash,
+    get_version_string,
+    shell,
+)
+
+
+def test_collapse():
+    text = """
+        line 1
+        line 2
+    """
+    assert collapse(text) == "line 1 line 2"
+    assert collapse("  single line  ") == "single line"
+
+
+def test_git_helpers_no_git(tmp_path):
+    # Not a git repo
+    assert get_git_hash(root=tmp_path) is None
+    assert get_version_string(root=tmp_path) == "unknown"
+
+
+def test_git_helpers_empty_repo(tmp_path):
+    shell("git init", cwd=tmp_path, check=True)
+    # No commits yet
+    assert get_git_hash(root=tmp_path) is None
+    assert get_version_string(root=tmp_path) == "unknown"
+
+
+def test_git_helpers_with_commits_and_diffs(tmp_path):
+    shell("git init", cwd=tmp_path, check=True)
+    shell("git config user.email you@example.com", cwd=tmp_path, check=True)
+    shell("git config user.name 'Your Name'", cwd=tmp_path, check=True)
+
+    # First commit
+    path = tmp_path / "file.txt"
+    path.write_text("content")
+    shell("git add file.txt", cwd=tmp_path, check=True)
+    shell("git commit -m initial", cwd=tmp_path, check=True)
+
+    git_hash = get_git_hash(root=tmp_path)
+    assert len(git_hash) >= 7  # usually 40, but git is git
+    assert get_version_string(root=tmp_path) == git_hash
+
+    # Create a diff
+    path.write_text("new content")
+    v = get_version_string(root=tmp_path)
+    assert v.startswith(git_hash + "\n")
+    assert "new content" in v
+    assert not v.endswith("\n")
+
+    # Commit again -> no diff
+    shell("git add file.txt", cwd=tmp_path, check=True)
+    shell("git commit -m second", cwd=tmp_path, check=True)
+    new_hash = get_git_hash(root=tmp_path)
+    assert new_hash != git_hash
+    assert get_version_string(root=tmp_path) == new_hash
+
+    # Diffs inside excluded test dirs are invisible to the version string
+    test_dir = tmp_path / "cmdeploy" / "src" / "cmdeploy" / "tests"
+    test_dir.mkdir(parents=True)
+    test_file = test_dir / "test_foo.py"
+    test_file.write_text("pass")
+    shell("git add .", cwd=tmp_path, check=True)
+    shell("git commit -m 'add test file'", cwd=tmp_path, check=True)
+    test_file.write_text("assert True")
+    assert get_version_string(root=tmp_path) == get_git_hash(root=tmp_path)
+
+
+def test_build_chatmaild_sdist(tmp_path):
+    dist_dir = tmp_path / "dist"
+
+    # First call builds the sdist
+    result = build_chatmaild_sdist(dist_dir)
+    assert result.name.endswith(".tar.gz")
+    assert result.stat().st_size > 0
+
+    # Second call is idempotent - returns the same file, no rebuild
+    mtime = result.stat().st_mtime
+    result2 = build_chatmaild_sdist(dist_dir)
+    assert result2 == result
+    assert result2.stat().st_mtime == mtime
+
+
+def test_get_chatmaild_sdist_errors(tmp_path):
+    with pytest.raises(FileNotFoundError):
+        get_chatmaild_sdist(tmp_path / "nonexistent")
+
+    empty = tmp_path / "empty"
+    empty.mkdir()
+    with pytest.raises(FileNotFoundError):
+        get_chatmaild_sdist(empty)

cmdeploy/src/cmdeploy/util.py

@@ -0,0 +1,126 @@
+"""Shared utility functions for cmdeploy."""
+
+import fcntl
+import subprocess
+import sys
+import textwrap
+from pathlib import Path
+
+
+def _project_root():
+    """Return the project root directory."""
+    return Path(__file__).resolve().parent.parent.parent.parent
+
+
+def collapse(text):
+    """Dedent, join lines, and strip a (triple-quoted) string.
+
+    Handy for writing shell commands across multiple lines::
+
+        cmd = collapse(f\"""
+            cmdeploy run
+            --config {ct.ini}
+            --ssh-host {ct.domain}
+        \""")
+    """
+    return textwrap.dedent(text).replace("\n", " ").strip()
+
+
+def shell(cmd, check=False, **kwargs):
+    """Run a shell command string with sensible defaults.
+
+    *cmd* is passed through :func:`collapse` first, so callers
+    can use triple-quoted f-strings freely.
+    Captures stdout/stderr by default; pass ``capture_output=False``
+    to stream output to the terminal instead.
+    """
+    if "capture_output" not in kwargs and "stdout" not in kwargs:
+        kwargs["capture_output"] = True
+    return subprocess.run(collapse(cmd), shell=True, text=True, check=check, **kwargs)
+
+
+def get_git_hash(root=None):
+    """Return the local HEAD commit hash, or None."""
+    if root is None:
+        root = _project_root()
+    result = shell(
+        "git rev-parse HEAD",
+        cwd=str(root),
+    )
+    if result.returncode == 0:
+        return result.stdout.strip()
+    return None
+
+
+DIFF_EXCLUDES = (
+    ":(exclude)cmdeploy/src/cmdeploy/tests",
+    ":(exclude)chatmaild/src/chatmaild/tests",
+)
+"""Git pathspecs appended to ``git diff`` so that changes
+limited to test files do not affect the deployed version string."""
+
+
+def get_version_string(root=None):
+    """Return ``git_hash\\ngit_diff`` for the local working tree.
+
+    Used by :class:`~cmdeploy.deployers.GithashDeployer` to write
+    ``/etc/chatmail-version`` and by ``lxc-status`` to compare
+    the deployed state against the local checkout.
+
+    Changes inside directories listed in :data:`DIFF_EXCLUDES`
+    are ignored so that test-only edits do not trigger
+    a redeployment.
+    """
+    if root is None:
+        root = _project_root()
+    git_hash = get_git_hash(root=root) or "unknown"
+    excludes = " ".join(f"'{e}'" for e in DIFF_EXCLUDES)
+    try:
+        git_diff = shell(
+            f"git diff -- . {excludes}",
+            cwd=str(root),
+        ).stdout.strip()
+    except Exception:
+        git_diff = ""
+    if git_diff:
+        return f"{git_hash}\n{git_diff}"
+    return git_hash
+
+
+def _chatmaild_default_dist_dir():
+    return _project_root() / "chatmaild" / "dist"
+
+
+def build_chatmaild_sdist(dist_dir=None):
+    """Build the chatmaild sdist if not already present (idempotent, process-safe)."""
+
+    if dist_dir is None:
+        dist_dir = _chatmaild_default_dist_dir()
+    dist_dir = Path(dist_dir).resolve()
+    dist_dir.mkdir(parents=True, exist_ok=True)
+
+    lockfile = dist_dir.parent / ".dist.lock"
+    with open(lockfile, "w") as fh:
+        fcntl.flock(fh, fcntl.LOCK_EX)
+        existing = [p for p in dist_dir.iterdir() if p.suffix == ".gz"]
+        if existing:
+            return existing[0]
+        subprocess.check_output(
+            [sys.executable, "-m", "build", "-n"]
+            + ["--sdist", "chatmaild", "--outdir", str(dist_dir)],
+            cwd=str(_project_root()),
+        )
+        return get_chatmaild_sdist(dist_dir)
+
+
+def get_chatmaild_sdist(dist_dir=None):
+    """Return the path to the pre-built chatmaild sdist."""
+    if dist_dir is None:
+        dist_dir = _chatmaild_default_dist_dir()
+
+    entries = list(Path(dist_dir).iterdir())
+    if len(entries) == 0:
+        raise FileNotFoundError(f"dist directory is empty: {dist_dir}")
+    if len(entries) > 1:
+        raise ValueError(f"expected one file in {dist_dir}, found {len(entries)}")
+    return entries[0]

doc/source/conf.py

@@ -15,7 +15,7 @@ author = 'chatmail collective'
 
 extensions = [
     #'sphinx.ext.autodoc',
-    #'sphinx.ext.viewdoc',
+    #'sphinx.ext.viewcode',
     'sphinxcontrib.mermaid',
 ]
 

doc/source/docker.rst

@@ -1,260 +0,0 @@
-Docker installation
-===================
-
-This section provides instructions for installing a chatmail relay
-using Docker Compose.
-
-.. note::
-
-    - Docker support is experimental and not yet covered by automated tests, please report bugs.
-    - This preliminary image simply wraps the cmdeploy process detailed in the :doc:`getting_started` instructions in a full Debian-systemd image with r/w access to `/sys/fs`
-    - Currently, the image has only been tested and built on amd64, though arm64 should theoretically work as well.
-
-
-Setup Preparation
------------------
-
-We use ``chat.example.org`` as the chatmail domain in the following
-steps. Please substitute it with your own domain.
-
-1. Install docker and docker compose v2 (check with `docker compose version`), install, e.g., through
-    - Debian 12 through the `official install instructions <https://docs.docker.com/engine/install/debian/#install-using-the-repository>`_
-    - Debian 13+ with `apt install docker docker-compose`
-
-   If you must use v1 (EOL since 2023), use `docker-compose` in the following and modify the `docker-compose.yaml` to use `privileged: true` instead of `cgroup: host`, though that will run give the container all priviledges.
-
-2. Setup the initial DNS records.
-   The following is an example in the familiar BIND zone file format with
-   a TTL of 1 hour (3600 seconds).
-   Please substitute your domain and IP addresses.
-
-   ::
-
-       chat.example.org. 3600 IN A 198.51.100.5
-       chat.example.org. 3600 IN AAAA 2001:db8::5
-       www.chat.example.org. 3600 IN CNAME chat.example.org.
-       mta-sts.chat.example.org. 3600 IN CNAME chat.example.org.
-
-3. Configure kernel parameters on the host, as these can not be set from the container::
-
-       echo "fs.inotify.max_user_instances=65536" | sudo tee -a /etc/sysctl.d/99-inotify.conf
-       echo "fs.inotify.max_user_watches=65536" | sudo tee -a /etc/sysctl.d/99-inotify.conf
-       sudo sysctl --system
-
-
-Docker Compose Setup
---------------------
-
-Pre-built images are available from GitHub Container Registry. The
-``main`` branch and tagged releases are pushed automatically by CI::
-
-    docker pull ghcr.io/chatmail/relay:main      # latest main branch
-    docker pull ghcr.io/chatmail/relay:1.2.3     # tagged release
-
-
-Create service directory
-^^^^^^^^^^^^^^^^^^^^^^^^
-
-Either:
-
-- Create a service directory, e.g., `/srv/chatmail-relay`::
-
-    mkdir -p /srv/chatmail-relay && cd /srv/chatmail-relay
-    wget https://raw.githubusercontent.com/chatmail/relay/refs/heads/main/docker-compose.yaml
-    wget https://raw.githubusercontent.com/chatmail/relay/refs/heads/main/docker-compose.override.yaml.example -O docker-compose.override.yaml
-
-- or clone the chatmail repo ::
-
-   git clone https://github.com/chatmail/relay
-   cd relay
-
-
-Customize and start
-^^^^^^^^^^^^^^^^^^^
-
-1. Set the fully qualified domain name of the relay::
-
-       echo 'MAIL_DOMAIN=chat.example.org' > .env
-
-   The container generates a ``chatmail.ini`` with defaults from
-   ``MAIL_DOMAIN`` on first start. To customize chatmail settings, mount
-   your own ``chatmail.ini`` instead (see `Custom chatmail.ini`_ below).
-
-2. All local customizations (data paths, extra volumes, config mounts) go in
-   ``docker-compose.override.yaml``, which Compose merges automatically with
-   the base file. By default, all data is stored in docker volumes, you will
-   likely want to at least create and configure the mail storage location, but
-   you might also want to configure external TLS certificates there.
-
-3. Start the container::
-
-       docker compose up -d
-       docker compose logs -f chatmail   # view logs, Ctrl+C to exit
-
-4. After installation is complete, open ``https://chat.example.org`` in
-   your browser.
-
-Finish install and test
------------------------
-
-You can test the installation with::
-
-        pip install cmping chat.example.org # or
-        uvx cmping chat.example.org # if you use https://docs.astral.sh/uv/
-
-You should check and extend your DNS records for better interoperability::
-
-    # Show required DNS records
-    docker exec chatmail /opt/cmdeploy/bin/cmdeploy dns --ssh-host @local
-
-You can check server status with::
-
-    docker exec chatmail /opt/cmdeploy/bin/cmdeploy status --ssh-host @local
-
-You can run some benchmarks (can also run from any machine with cmdeploy installed)
-
-    docker exec chatmail /opt/cmdeploy/bin/cmdeploy bench chat.example.org
-
-You can run the test suite with
-
-    docker exec chatmail /opt/cmdeploy/bin/cmdeploy test chat.example.org --ssh-host localhost
-
-
-Customization
--------------
-
-Website
-^^^^^^^^^^^^^^
-
-You can customize the chatmail landing page by mounting a directory with
-your own website source files.
-
-1. Create a directory with your custom website source::
-
-       mkdir -p ./custom/www/src
-       nano ./custom/www/src/index.md
-
-2. Add the volume mount in ``docker-compose.override.yaml``::
-
-       services:
-         chatmail:
-           volumes:
-             - ./custom/www:/opt/chatmail-www
-
-3. Restart the service::
-
-       docker compose down
-       docker compose up -d
-
-
-Custom chatmail.ini
-^^^^^^^^^^^^^^^^^^^
-
-If you want to go beyond simply setting the ``MAIL_DOMAIN`` in ``.env``, you
-can use a regular `chatmail.ini` to give you full control.
-
-1. Extract the generated config from a running container::
-
-       docker cp chatmail:/etc/chatmail/chatmail.ini ./chatmail.ini
-
-2. Edit ``chatmail.ini`` as needed.
-
-3. Add the volume mount in ``docker-compose.override.yaml`` ::
-
-       services:
-         chatmail:
-           volumes:
-             - ./chatmail.ini:/etc/chatmail/chatmail.ini
-
-4. Restart the container, the container skips generating a new one: ::
-
-       docker compose down && docker compose up -d
-
-
-External TLS certificates
-^^^^^^^^^^^^^^^^^^^^^^^^^
-
-If TLS certificates are managed outside the container (e.g. by certbot,
-acmetool, or Traefik on the host), mount them into the container and set
-``TLS_EXTERNAL_CERT_AND_KEY`` in ``docker-compose.override.yaml``.
-Changed certificates are picked up automatically via inotify.
-See the examples in the example override and :ref:`external-tls` in the getting started guide for details.
-
-
-Migrating from a bare-metal install
-------------------------------------
-
-If you have an existing bare-metal chatmail installation and want to
-switch to Docker:
-
-1. Stop all existing services::
-
-       systemctl stop postfix dovecot doveauth nginx opendkim unbound \
-         acmetool-redirector filtermail filtermail-incoming chatmail-turn \
-         iroh-relay chatmail-metadata lastlogin mtail
-       systemctl disable postfix dovecot doveauth nginx opendkim unbound \
-         acmetool-redirector filtermail filtermail-incoming chatmail-turn \
-         iroh-relay chatmail-metadata lastlogin mtail
-
-2. Copy your existing ``chatmail.ini`` and mount it into the container
-   (see `Custom chatmail.ini`_ above)::
-
-       cp /usr/local/lib/chatmaild/chatmail.ini ./chatmail.ini
-
-3. Copy persistent data into the ``./data/`` subdirectories (for example, as configured in `Customize and start`_) ::
-
-       mkdir -p data/dkim data/certs data/mail
-
-       # DKIM keys
-       cp -a /etc/dkimkeys/* data/dkim/
-
-       # TLS certificates
-       rsync -a /var/lib/acme/ data/certs/
-
-   Note that ownership of dkim and acme is adjusted on container start.
-
-   For the mail directory::
-
-       rsync -a /home/vmail/ data/mail/
-
-   Alternatively, mount ``/home/vmail`` directly by changing the volume
-   in ``docker-compose-override.yaml``::
-
-       - /home/vmail:/home/vmail
-
-   The three ``./data/`` subdirectories cover all persistent state.
-   Everything else is regenerated by the ``configure`` and ``activate``
-   stages on container start.
-
-Building the image
-------------------
-
-Clone the repository and build the Docker image::
-
-    git clone https://github.com/chatmail/relay
-    cd relay
-    docker compose build chatmail
-
-The build bakes all binaries, Python packages, and the install stage
-into the image. After building, only ``docker-compose.yaml`` and a ``.env`` with
-``MAIL_DOMAIN`` are needed to run the container.
-
-You can transfer a locally built image to your server directly (pigz is parallel `gzip` which can be used instead as well) ::
-
-    docker save chatmail-relay:latest | pigz | ssh chat.example.org 'pigz -d | docker load'
-
-
-Forcing a full reinstall
-------------------------
-
-On container start, only the ``configure`` and ``activate`` stages run by default.
-
-To force a full reinstall (e.g. after updating the source), either
-rebuild the image::
-
-    docker compose build chatmail
-    docker compose up -d
-
-Or override the stages at runtime without rebuilding::
-
-    CMDEPLOY_STAGES="install,configure,activate" docker compose up -d

doc/source/getting_started.rst

@@ -98,12 +98,6 @@ steps. Please substitute it with your own domain.
    configure at your DNS provider (it can take some time until they are
    public).
 
-Docker installation
--------------------
-
-There is experimental support for running chatmail via Docker Compose.
-See :doc:`docker` for full setup instructions.
-
 Other helpful commands
 ----------------------
 
@@ -235,7 +229,11 @@ The deploy will verify that both files exist on the server.
    You are responsible for certificate renewal.
    When the certificate file changes on disk,
    all relay services pick up the new certificate automatically
-   (via a systemd path watcher installed during deploy).
+   via a systemd path watcher installed during deploy.
+   The watcher uses inotify, which does not cross bind-mount boundaries.
+   If you use such a setup, you must trigger the reload explicitly after renewal::
+
+      systemctl start tls-cert-reload.service
 
 
 Migrating to a new build machine

doc/source/index.rst

@@ -13,9 +13,9 @@ Contributions and feedback welcome through the https://github.com/chatmail/relay
     :maxdepth: 5
 
     getting_started
-    docker
     proxy
     migrate
     overview
+    lxc
     related
     faq

doc/source/lxc.rst

@@ -0,0 +1,288 @@
+Local testing with LXC/Incus
+============================
+
+.. warning::
+
+   cmdeploy LXC support is geared towards local testing and CI, only.
+   Do not base production setups on it.
+
+
+The ``cmdeploy`` tool includes support for running
+chatmail relays inside local
+`Incus <https://linuxcontainers.org/incus/>`_ LXC containers.
+This is useful for development, testing, and CI
+without requiring a remote server.
+LXC system containers behave like lightweight virtual machines.
+They share the host's kernel but run their own init system
+(systemd), package manager, and network stack,
+so the cmdeploy deployment scripts work exactly
+as they would on a real Debian server or cloud VPS.
+
+Prerequisites
+-------------
+
+Install `Incus <https://linuxcontainers.org/incus/>`_
+(LXC container manager).
+See the `official installation guide
+<https://linuxcontainers.org/incus/docs/main/installing/>`_
+for full details.
+
+After installing incus, initialise and grant yourself access::
+
+    sudo incus admin init --minimal
+    sudo usermod -aG incus-admin $USER
+
+.. warning::
+
+   You **must now log out and back in** (or run ``newgrp incus-admin``)
+   after adding yourself to the group.
+   Without this, all ``cmdeploy lxc-*`` commands
+   will fail with permission errors.
+
+Verify the installation works by running ``incus list``,
+which should print an empty table without errors.
+
+
+Quick start
+-----------
+
+::
+
+    cd relay
+    scripts/initenv.sh               # bootstrap venv
+    source venv/bin/activate         # activate venv
+    cmdeploy lxc-test                # create containers, deploy, test
+
+
+The ``lxc-test`` command executes each ``cmdeploy`` subprocess command
+so you can copy-paste and run them individually.
+A section timing summary is printed at the end.
+No host DNS delegation or ``~/.ssh/config`` changes are needed
+because lxc-test passes ssh-related CLI options to
+``cmdeploy run`` and ``cmdeploy test`` commands.
+
+
+CLI reference
+--------------
+
+``lxc-start [--ipv4-only] [--run] [NAME ...]``
+    Create and start containers.
+    Without arguments, creates ``test0-localchat`` and ``ns-localchat`` (DNS).
+    Pass one or more ``NAME`` arguments to create user relay containers instead
+    (e.g. ``cmdeploy lxc-start myrelay``).
+    Use ``--ipv4-only`` to set ``disable_ipv6 = True`` in the generated ``chatmail.ini``,
+    producing an IPv4-only relay.
+    Use ``--run`` to automatically run ``cmdeploy run`` on each container after starting it.
+    Generates ``lxconfigs/ssh-config``.
+    It reuses existing containers and resets DNS zones to minimal records.
+
+``lxc-stop [--destroy] [--destroy-all] [NAME ...]``
+    Stop relay containers.
+    Without arguments, stops ``test0-localchat`` and ``test1-localchat``.
+    Pass ``NAME`` to stop specific containers.
+    Use ``--destroy`` to also delete the containers and their config files.
+    Use ``--destroy-all`` to additionally destroy
+    the ``ns-localchat`` DNS container **and** remove all cached
+    images (``localchat-base``, per-relay images),
+    giving a fully clean slate for the next ``lxc-test``.
+    User containers are **never** destroyed unless named explicitly.
+
+``lxc-test [--one]``
+    Idempotent full pipeline:
+
+    1. ``lxc-start``: create ``test0`` + ``test1`` containers,
+       configure DNS with readiness check
+
+    2. ``cmdeploy run``: deploy chatmail services
+       on all relays **in parallel**
+
+    3. publish per-relay cached images (``localchat-test0``,
+       ``localchat-test1``) after first successful deploy
+
+    4. ``cmdeploy dns --zonefile``: generate standard
+       BIND-format zone files, load full DNS records
+
+    5. ``cmdeploy test``: run full test suite
+       with ``-n4 -x``
+
+    By default creates, deploys, and tests both ``test0`` and ``test1``
+    for dual-domain federation testing (sets ``CHATMAIL_DOMAIN2=_test1.localchat``).
+    test0 runs dual-stack (IPv4 + IPv6) while test1 runs IPv4-only (``disable_ipv6 = True``).
+    Pass ``--one`` to only deploy and test against ``test0``
+    (skips ``test1``, does not set ``CHATMAIL_DOMAIN2``).
+
+``lxc-status``
+    Show live status of all LXC containers (including the DNS container),
+    deploy freshness (comparing ``/etc/chatmail-version``
+    against local ``git rev-parse HEAD`` and ``git diff``),
+    SSH config inclusion, and host DNS forwarding for ``.localchat``.
+    Reports **IN-SYNC**, **DIRTY** (hash matches but uncommitted changes exist),
+    **STALE** (different commit), or **NOT DEPLOYED**.
+
+
+Container types
+-----------------
+
+**Test relay containers** (``test0-localchat``, ``test1-localchat``)
+    Created automatically by ``lxc-test``.
+    **test0** has IPv4 and IPv6 configured,
+    **test1** is IPv4-only (``disable_ipv6 = True``).
+
+**User relay containers** (``<name>-localchat``)
+    Created by ``cmdeploy lxc-start <name>``
+    where ``<name>`` does not start with ``test``.
+    These are personal development instances,
+    never touched by ``lxc-stop --destroy`` unless named explicitly.
+
+**DNS container** (``ns-localchat``)
+    Singleton container running PowerDNS.
+    Created automatically when any relay is started.
+
+
+.. _lxc-ssh-config:
+
+SSH configuration
+-----------------
+
+``cmdeploy lxc-start`` generates ``lxconfigs/ssh-config``,
+a standard OpenSSH config file mapping every container name,
+its domain, and a short alias to the container's IP address::
+
+    Host test0-localchat _test0.localchat _test0
+        Hostname 10.204.0.42
+        User root
+        IdentityFile /path/to/relay/lxconfigs/id_localchat
+        IdentitiesOnly yes
+        StrictHostKeyChecking accept-new
+        UserKnownHostsFile /dev/null
+        LogLevel ERROR
+
+All ``cmdeploy`` commands (``run``, ``dns``, ``status``, ``test``)
+accept ``--ssh-config lxconfigs/ssh-config`` to use this file.
+``lxc-test`` passes it automatically.
+
+**Using containers from the host shell:**
+
+To make ``ssh _test0`` work from any terminal, add one line to ``~/.ssh/config``::
+
+    Include /absolute/path/to/relay/lxconfigs/ssh-config
+
+
+.. _lxc-dns-setup:
+.. _localchat-tld:
+
+``.localchat`` DNS and name resolution
+---------------------------------------
+
+All LXC-managed chatmail domains use the ``.localchat`` pseudo-TLD
+(e.g. ``_test0.localchat``, ``_test1.localchat``),
+a non-delegated suffix that exists only within the local PowerDNS infrastructure.
+A dedicated DNS container (``ns-localchat``)
+is created so that local test relays interact
+with DNS similar to a regular public Internet setup.
+On first start, ``cmdeploy lxc-start`` creates this container
+running two `PowerDNS <https://www.powerdns.com/>`_ services:
+
+* **pdns-server** (authoritative) serves ``.localchat``
+  zones from a local SQLite database.
+
+* **pdns-recursor** (recursive) listens on the Incus
+  bridge so all containers can use it.
+  Forwards ``.localchat`` queries to the local
+  authoritative server and everything else to Quad9 (``9.9.9.9``).
+
+After the DNS container is up, ``lxc-start`` configures the Incus bridge
+to advertise its IP via DHCP and disables Incus's own DNS.
+DNS records are then created in two phases matching the "cmdeploy run" deployment flow:
+
+1. **``lxc-start``** resets each relay zone to
+   **SOA, NS, and A** records (plus **AAAA** for dual-stack containers).
+   If host DNS resolution is configured, users can
+   afterwards run ``cmdeploy run --config lxconfigs/chatmail-test0.ini
+   --ssh-config lxconfigs/ssh-config --ssh-host _test0.localchat``.
+   LXC subcommands do not depend on host DNS resolution
+   and resolve addresses via ``lxconfigs/ssh-config``.
+
+2. **``cmdeploy dns --zonefile``** generates a standard
+   BIND-format zone file (MX, TXT/SPF, TXT/DMARC,
+   TXT/MTA-STS, SRV, CNAME, DKIM) and loads it
+   into PowerDNS.
+
+This two-phase approach prevents premature configuration of mail records
+before the relay is actually deployed and running.
+Once ``cmdeploy run`` deploys `Unbound <https://nlnetlabs.nl/projects/unbound/>`_
+inside a relay container, Unbound has a configuration plugin snippet
+that forwards all ``.localchat`` queries to the PowerDNS recursor,
+and lets all other queries go through normal recursive resolution.
+
+
+State outside the repository
+-----------------------------
+
+All generated configuration by lxc subcommands live in ``lxconfigs/``
+(git-ignored), including the SSH key pair (``id_localchat``),
+per-container ``chatmail-*.ini`` files, zone files, and ``ssh-config``.
+
+The only state *outside* the repository is the Incus containers and images themselves
+(managed via the ``incus`` CLI, labelled with ``user.localchat-managed=true``).
+Several cached images are published to the local Incus image store:
+
+* ``localchat-base``: Debian 12 with openssh-server and Python
+  (built on first run)
+
+* ``localchat-test0``, ``localchat-test1``: per-relay snapshots
+  published after the first successful ``cmdeploy run``.
+  Subsequent containers launch from these images
+  so the deploy step is mostly no-ops.
+
+Relay containers are limited to **500 MiB RAM**
+and the DNS container to **256 MiB**.
+
+
+.. _lxc-tls:
+
+TLS handling and underscore domains
+------------------------------------
+
+Container domains start with ``_`` (e.g. ``_test0.localchat``).
+As described in :doc:`getting_started` ("Running a relay with self-signed certificates"),
+underscore domains automatically use self-signed TLS
+and ``smtp_tls_security_level = encrypt``.
+This permits cross-relay federation between LXC containers
+without any external certificate authority.
+Delta Chat clients connecting to these relays
+must be configured with
+``certificateChecks = acceptInvalidCertificates``
+(the test fixtures handle this automatically).
+`PR #7926 on chatmail-core <https://github.com/chatmail/core/pull/7926>`_
+is meant to make this special setting unnecessary for chatmail clients
+that are connecting to underscore domains.
+
+
+Known limitations
+------------------
+
+The LXC environment differs from a production
+deployment in several ways:
+
+**No ACME / Let's Encrypt**:
+  Self-signed TLS only (see :ref:`lxc-tls`);
+  ACME code paths are never exercised locally.
+
+**No inbound connections from the internet**:
+  Containers sit on a private Incus bridge and are not port-forwarded.
+  Only the host and other containers on the same bridge can reach them.
+
+**Local federation only**:
+  Cross-relay mail delivery (e.g. test0 → test1) works between containers on the same host,
+  but these relays are invisible to any external mail server.
+
+**DNS is local only**:
+  The ``.localchat`` pseudo-TLD is not resolvable from the wider internet
+  (see :ref:`lxc-dns-setup`).
+
+**IPv6 is ULA-only**:
+  Containers receive IPv6 addresses from the ``fd42:...`` ULA range on the Incus bridge.
+  These are not globally routable, but are sufficient for testing IPv6 service binding
+  (Postfix, Dovecot, Nginx) and DNS AAAA records inside the local environment.
+  test1 runs with ``disable_ipv6 = True`` to exercise the IPv4-only deployment path.

doc/source/overview.rst

@@ -109,10 +109,6 @@ short overview of ``chatmaild`` services:
    is contacted by Dovecot when a user logs in and stores the date of
    the login.
 
--  `metrics <https://github.com/chatmail/relay/blob/main/chatmaild/src/chatmaild/metrics.py>`_
-   collects some metrics and displays them at
-   ``https://example.org/metrics``.
-
 ``www/``
 ~~~~~~~~~
 
@@ -142,11 +138,9 @@ Chatmail relay dependency diagram
         nginx-internal --- autoconfig.xml;
         certs-nginx[("`TLS certs
         /var/lib/acme`")] --> nginx-internal;
-        systemd-timer --- chatmail-metrics;
         systemd-timer --- acmetool;
         systemd-timer --- chatmail-expire-daily;
         systemd-timer --- chatmail-fsreport-daily;
-        chatmail-metrics --- website;
         acmetool --> certs[("`TLS certs
         /var/lib/acme`")];
         nginx-external --- |993|dovecot;

docker-compose.override.yaml.example

@@ -1,40 +0,0 @@
-# Local overrides — copy to docker-compose.override.yaml in the repo root.
-# Compose automatically merges this with docker-compose.yaml.
-#
-#   cp docker-compose.override.yaml.example docker-compose.override.yaml
-#
-# Volumes are APPENDED to the base file's volumes list.
-# Environment and other scalar keys are MERGED by key.
-services:
-  chatmail:
-    volumes:
-      ## Data paths — bind-mount to host directories for easy access/backup.
-
-      # - ./data/dkim:/etc/dkimkeys
-      # - ./data/certs:/var/lib/acme
-
-      # - ./data/mail:/home/vmail
-      ## Or mount from an existing bare-metal install.
-      # - /home/vmail:/home/vmail
-
-      ## Mount your own chatmail.ini (skips auto-generation):
-      # - ./chatmail.ini:/etc/chatmail/chatmail.ini
-
-      ## Custom website:
-      # - ./custom/www:/opt/chatmail-www
-
-      ## Debug — mount scripts from the repo for live editing:
-      # - ./docker/files/setup_chatmail_docker.sh:/setup_chatmail_docker.sh
-      # - ./docker/files/entrypoint.sh:/entrypoint.sh
-
-    # environment:
-      ## Mount certs (above) and set TLS_EXTERNAL_CERT_AND_KEY to in-container paths.
-      ## Changed certs are picked up automatically (inotify via tls-cert-reload.path).
-      ##
-      ## Host acmetool (bare-metal migration): create mount above, and
-      ##   rsync -a /var/lib/acme/live data/certs
-      # TLS_EXTERNAL_CERT_AND_KEY: "/var/lib/acme/live/${MAIL_DOMAIN}/fullchain /var/lib/acme/live/${MAIL_DOMAIN}/privkey"
-      ##
-      ## (Untested) Traefik certs-dumper (see docker/docker-compose-traefik.yaml) - also add volume:
-      ##   - traefik-certs:/certs:ro
-      # TLS_EXTERNAL_CERT_AND_KEY: "/certs/${MAIL_DOMAIN}/certificate.crt /certs/${MAIL_DOMAIN}/privatekey.key"

docker-compose.yaml

@@ -1,47 +0,0 @@
-# Base compose file — do not edit. Put customizations (data paths, extra
-# volumes, env overrides) in docker-compose.override.yaml instead.
-# See docker/docker-compose.override.yaml.example for a starting point.
-#
-# Security note: this container uses network_mode:host (chatmail needs many
-# ports: 25, 53, 80, 143, 443, 465, 587, 993, 3340, 8443) and cgroup:host
-# (required for systemd). Together these give the container near-host-level
-# access. This is acceptable for a dedicated mail server, but be aware that
-# the container can bind any port and see all host network traffic.
-services:
-  chatmail:
-    build:
-      context: ./
-      dockerfile: docker/chatmail_relay.dockerfile
-      args:
-        GIT_HASH: ${GIT_HASH:-unknown}
-    image: chatmail-relay:latest
-    restart: unless-stopped
-    container_name: chatmail
-    # Required for systemd — use only one of the following:
-    cgroup: host          # compose v2
-    # privileged: true    # compose v1 (less restricted)
-    tty: true # required for logs
-    tmpfs: # required for systemd
-      - /tmp
-      - /run
-      - /run/lock
-    logging:
-      driver: json-file
-      options:
-        max-size: "10m"
-        max-file: "3"
-    environment:
-      MAIL_DOMAIN: $MAIL_DOMAIN
-    network_mode: "host"
-    volumes:
-      ## system (required)
-      - /sys/fs/cgroup:/sys/fs/cgroup:rw
-      ## data (defaults — override in docker-compose.override.yaml)
-      - mail:/home/vmail
-      - dkim:/etc/dkimkeys
-      - certs:/var/lib/acme
-
-volumes:
-  mail:
-  dkim:
-  certs:

docker/build.sh

@@ -1,9 +0,0 @@
-#!/bin/sh
-# Build the chatmail Docker image with the current git hash baked in.
-# Usage: ./docker/build.sh [extra docker-compose build args...]
-#
-# .git/ is excluded from the build context (.dockerignore) so the hash
-# must be passed as a build arg from the host.
-
-export GIT_HASH=$(git rev-parse --short HEAD)
-exec docker compose build "$@"

docker/chatmail_relay.dockerfile

@@ -1,99 +0,0 @@
-FROM jrei/systemd-debian:12 AS base
-
-ENV LANG=en_US.UTF-8
-
-RUN echo 'APT::Install-Recommends "0";' > /etc/apt/apt.conf.d/01norecommend && \
-    echo 'APT::Install-Suggests "0";' >> /etc/apt/apt.conf.d/01norecommend && \
-    apt-get update && \
-    apt-get install -y \
-        ca-certificates && \
-    DEBIAN_FRONTEND=noninteractive \
-    TZ=UTC \
-    apt-get install -y tzdata && \
-    apt-get install -y locales && \
-    sed -i -e "s/# $LANG.*/$LANG UTF-8/" /etc/locale.gen && \
-    dpkg-reconfigure --frontend=noninteractive locales && \
-    update-locale LANG=$LANG \
-    && rm -rf /var/lib/apt/lists/*
-
-RUN apt-get update && \
-    apt-get install -y \
-        git \
-        python3 \
-        python3-venv \
-        python3-virtualenv \
-        gcc \
-        python3-dev \
-        opendkim \
-        opendkim-tools \
-        curl \
-        rsync \
-        unbound \
-        unbound-anchor \
-        dnsutils \
-        postfix \
-        acl \
-        nginx \
-        libnginx-mod-stream \
-        fcgiwrap \
-        cron \
-    && rm -rf /var/lib/apt/lists/*
-
-# --- Build-time: install cmdeploy venv and run install stage ---
-# Editable install so importlib.resources reads directly from the source tree.
-# On container start only "configure,activate" stages run.
-COPY . /opt/chatmail/
-WORKDIR /opt/chatmail
-
-RUN printf '[params]\nmail_domain = build.local\n' > /tmp/chatmail.ini
-
-# Dummy git repo init: .git/ is excluded from the build context (.dockerignore)
-# but setuptools calls `git ls-files` when building the sdist.
-RUN git init -q && \
-    python3 -m venv /opt/cmdeploy && \
-    /opt/cmdeploy/bin/pip install --no-cache-dir \
-        -e chatmaild/ -e cmdeploy/
-
-RUN CMDEPLOY_STAGES=install \
-    CHATMAIL_INI=/tmp/chatmail.ini \
-    CHATMAIL_NOSYSCTL=True \
-    CHATMAIL_NOPORTCHECK=True \
-    /opt/cmdeploy/bin/pyinfra @local \
-        /opt/chatmail/cmdeploy/src/cmdeploy/run.py -y
-
-RUN cp -a www/ /opt/chatmail-www/
-
-RUN rm -f /tmp/chatmail.ini
-
-# Record image version (used in deploy fingerprint at runtime).
-# GIT_HASH is passed as a build arg (from docker-compose or CI) so that
-# .git/ can be excluded from the build context via .dockerignore.
-ARG GIT_HASH=unknown
-RUN echo "$GIT_HASH" > /etc/chatmail-image-version && \
-    echo "$GIT_HASH" > /etc/chatmail-version
-# --- End build-time install ---
-
-ENV TZ=:/etc/localtime
-ENV PATH="/opt/cmdeploy/bin:${PATH}"
-RUN ln -s /etc/chatmail/chatmail.ini /opt/chatmail/chatmail.ini
-
-ARG SETUP_CHATMAIL_SERVICE_PATH=/lib/systemd/system/setup_chatmail.service
-COPY ./docker/files/setup_chatmail.service "$SETUP_CHATMAIL_SERVICE_PATH"
-RUN ln -sf "$SETUP_CHATMAIL_SERVICE_PATH" "/etc/systemd/system/multi-user.target.wants/setup_chatmail.service"
-
-# Remove default nginx site config at build time (not in entrypoint)
-RUN rm -f /etc/nginx/sites-enabled/default
-
-COPY --chmod=555 ./docker/files/setup_chatmail_docker.sh /setup_chatmail_docker.sh
-COPY --chmod=555 ./docker/files/entrypoint.sh /entrypoint.sh
-
-HEALTHCHECK --interval=60s --timeout=10s --retries=3 \
-  CMD systemctl is-active dovecot postfix nginx unbound opendkim filtermail doveauth chatmail-metadata || exit 1
-
-STOPSIGNAL SIGRTMIN+3
-
-ENTRYPOINT ["/entrypoint.sh"]
-
-CMD [   "--default-standard-output=journal+console", \
-        "--default-standard-error=journal+console" ]
-

docker/files/entrypoint.sh

@@ -1,12 +0,0 @@
-#!/bin/bash
-set -eo pipefail
-
-SETUP_CHATMAIL_SERVICE_PATH="${SETUP_CHATMAIL_SERVICE_PATH:-/lib/systemd/system/setup_chatmail.service}"
-
-# Whitelist only the env vars needed by setup_chatmail_docker.sh.
-# Forwarding all env vars (via printenv) would leak Docker internals,
-# orchestrator secrets, and other unrelated variables into systemd.
-env_vars="MAIL_DOMAIN CMDEPLOY_STAGES CHATMAIL_INI TLS_EXTERNAL_CERT_AND_KEY PATH"
-sed -i "s|<envs_list>|$env_vars|g" "$SETUP_CHATMAIL_SERVICE_PATH"
-
-exec /lib/systemd/systemd "$@"

docker/files/setup_chatmail.service

@@ -1,14 +0,0 @@
-[Unit]
-Description=Run container setup commands
-After=multi-user.target
-ConditionPathExists=/setup_chatmail_docker.sh
-
-[Service]
-Type=oneshot
-ExecStart=/bin/bash /setup_chatmail_docker.sh
-RemainAfterExit=true
-WorkingDirectory=/opt/chatmail
-PassEnvironment=<envs_list>
-
-[Install]
-WantedBy=multi-user.target

docker/files/setup_chatmail_docker.sh

@@ -1,63 +0,0 @@
-#!/bin/bash
-
-set -euo pipefail
-export CHATMAIL_INI="${CHATMAIL_INI:-/etc/chatmail/chatmail.ini}"
-export CHATMAIL_NOSYSCTL=True
-export CHATMAIL_NOPORTCHECK=True
-
-CMDEPLOY=/opt/cmdeploy/bin/cmdeploy
-
-if [ -z "$MAIL_DOMAIN" ]; then
-    echo "ERROR: Environment variable 'MAIL_DOMAIN' must be set!" >&2
-    exit 1
-fi
-
-### MAIN
-
-if [ ! -f /etc/dkimkeys/opendkim.private ]; then
-    /usr/sbin/opendkim-genkey -D /etc/dkimkeys -d "$MAIL_DOMAIN" -s opendkim
-fi
-# Fix ownership for bind-mounted keys (host opendkim UID may differ from container)
-chown -R opendkim:opendkim /etc/dkimkeys
-
-# Journald: forward to console for docker logs
-grep -q '^ForwardToConsole=yes' /etc/systemd/journald.conf \
-    || echo "ForwardToConsole=yes" >> /etc/systemd/journald.conf
-systemctl restart systemd-journald
-
-# Create chatmail.ini (skips if file already exists, e.g. volume-mounted)
-mkdir -p "$(dirname "$CHATMAIL_INI")"
-if [ ! -f "$CHATMAIL_INI" ]; then
-    $CMDEPLOY init --config "$CHATMAIL_INI" "$MAIL_DOMAIN"
-fi
-
-# Inject external TLS paths from env var (unless user mounted their own ini)
-if [ -n "${TLS_EXTERNAL_CERT_AND_KEY:-}" ]; then
-    if ! grep -q '^tls_external_cert_and_key' "$CHATMAIL_INI"; then
-        echo "tls_external_cert_and_key = $TLS_EXTERNAL_CERT_AND_KEY" >> "$CHATMAIL_INI"
-    fi
-fi
-
-# --- Deploy fingerprint: skip cmdeploy run if nothing changed ---
-# On restart with identical image+config, systemd already brings up all
-# enabled services — the full cmdeploy run is redundant (~30s saved).
-# The install stage runs at image build time (Dockerfile), so only
-# configure+activate are needed here.
-IMAGE_VERSION_FILE="/etc/chatmail-image-version"
-FINGERPRINT_FILE="/etc/chatmail/.deploy-fingerprint"
-image_ver="none"
-[ -f "$IMAGE_VERSION_FILE" ] && image_ver=$(cat "$IMAGE_VERSION_FILE")
-config_hash=$(sha256sum "$CHATMAIL_INI" | cut -c1-16)
-current_fp="${image_ver}:${config_hash}"
-
-# CMDEPLOY_STAGES non-empty in env = operator override → always run.
-# Otherwise, if fingerprint matches the last successful deploy, skip.
-if [ -z "${CMDEPLOY_STAGES:-}" ] \
-    && [ -f "$FINGERPRINT_FILE" ] \
-    && [ "$(cat "$FINGERPRINT_FILE")" = "$current_fp" ]; then
-    echo "[INFO] No changes detected ($current_fp), skipping deploy."
-else
-    export CMDEPLOY_STAGES="${CMDEPLOY_STAGES:-configure,activate}"
-    $CMDEPLOY run --config "$CHATMAIL_INI" --ssh-host @local
-    echo "$current_fp" > "$FINGERPRINT_FILE"
-fi

env.example

@@ -1 +0,0 @@
-MAIL_DOMAIN=chat.example.com