try make opendkim reload on test1 faster

Remove the staging VPS-based test-and-deploy workflows and
replace them with LXC-local CI. Refactor image caching
to use per-container aliases instead of a single relay image.
Ensure postfix/dovecot/opendkim restarts after unbound restarts

.github/workflows/ci.yaml

@@ -15,28 +15,98 @@ jobs:
         with:
           ref: ${{ github.event.pull_request.head.sha }}
       - name: download filtermail
-        run: curl -L https://github.com/chatmail/filtermail/releases/download/v0.6.0/filtermail-x86_64 -o /usr/local/bin/filtermail && chmod +x /usr/local/bin/filtermail
+        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 
+      - 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 
+      - name: run formatting checks
-        run: cmdeploy fmt -v 
+        run: cmdeploy fmt -v
 
-      - name: run deploy-chatmail offline tests 
+      - name: run deploy-chatmail offline tests
-        run: pytest --pyargs cmdeploy 
+        run: pytest --pyargs cmdeploy
 
-      # all other cmdeploy commands require a staging server 
+  lxc-test:
-      # see https://github.com/deltachat/chatmail/issues/100
+    name: LXC deploy and test
+    runs-on: ubuntu-24.04
+    timeout-minutes: 15
+    steps:
+      - uses: actions/checkout@v4
+        with:
+          ref: ${{ github.event.pull_request.head.sha }}
+
+      - name: install incus
+        run: sudo apt-get update && sudo apt-get install -y incus
+
+      - name: initialise incus
+        run: |
+          sudo systemctl stop docker.socket docker || true
+          sudo iptables -P FORWARD ACCEPT
+          sudo sysctl -w fs.inotify.max_user_instances=65535
+          sudo sysctl -w fs.inotify.max_user_watches=65535
+          sudo incus admin init --minimal --quiet
+          sudo usermod -aG incus-admin "$USER"
+
+      - name: initenv
+        run: scripts/initenv.sh
+
+      - name: append venv/bin to PATH
+        run: echo venv/bin >>$GITHUB_PATH
+
+      - name: restore cached images
+        id: cache-images
+        uses: actions/cache@v4
+        with:
+          path: |
+            /tmp/localchat-base.tar.gz
+            /tmp/localchat-ns.tar.gz
+            /tmp/localchat-test0.tar.gz
+            /tmp/localchat-test1.tar.gz
+            lxconfigs/id_localchat*
+          key: incus-contain-v2-${{ runner.os }}-${{ github.ref_name }}
+          restore-keys: |
+            incus-contain-v2-${{ runner.os }}-${{ github.ref_name }}-
+            incus-contain-v2-${{ runner.os }}-main-
+            incus-contain-v2-${{ runner.os }}-
+
+      - name: import cached images
+        run: |
+          for alias in localchat-base localchat-ns localchat-test0 localchat-test1; do
+            if [ -f /tmp/$alias.tar.gz ]; then
+              sg incus-admin -c "incus --quiet image import /tmp/$alias.tar.gz --alias $alias" || true
+            fi
+          done
+
+      - name: cmdeploy lxc-test
+        run: sg incus-admin -c 'cmdeploy lxc-test -vv'
+
+      - name: show container logs on failure
+        if: failure() || cancelled()
+        run: |
+          for c in test0-localchat test1-localchat; do
+            echo "::group::$c journal (warnings+errors)"
+            sg incus-admin -c "incus exec $c -- journalctl -p warning --no-pager -n200" 2>/dev/null || echo "no log"
+            echo "::endgroup::"
+            echo "::group::$c failed services"
+            sg incus-admin -c "incus exec $c -- systemctl --no-pager --failed" || true
+            echo "::endgroup::"
+          done
+
+      - name: export images for cache
+        if: always()
+        run: |
+          for alias in localchat-base localchat-ns localchat-test0 localchat-test1; do
+            sg incus-admin -c "incus --quiet image export $alias /tmp/$alias" || true
+          done

.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

@@ -1,104 +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 
-
-      - name: setup dependencies
-        run: |
-          ssh root@staging-ipv4.testrun.org apt update
-          ssh root@staging-ipv4.testrun.org apt install -y git python3.11-venv python3-dev gcc
-          ssh root@staging-ipv4.testrun.org git clone https://github.com/chatmail/relay
-          ssh root@staging-ipv4.testrun.org "cd relay && git checkout " ${{ github.head_ref }}
-          ssh root@staging-ipv4.testrun.org "cd relay && scripts/initenv.sh"
-
-      - name: initialize config
-        run: |
-          ssh root@staging-ipv4.testrun.org "cd relay && scripts/cmdeploy init staging-ipv4.testrun.org"
-          ssh root@staging-ipv4.testrun.org "sed -i 's#disable_ipv6 = False#disable_ipv6 = True#' relay/chatmail.ini"
-          ssh root@staging-ipv4.testrun.org "sed -i 's/#\s*mtail_address/mtail_address/' relay/chatmail.ini"
-
-      - run: ssh root@staging-ipv4.testrun.org "cd relay && scripts/cmdeploy run --verbose --skip-dns-check --ssh-host localhost"
-
-      - name: set DNS entries
-        run: |
-          ssh root@staging-ipv4.testrun.org "cd relay && scripts/cmdeploy dns --zonefile staging-generated.zone --ssh-host localhost"
-          ssh root@staging-ipv4.testrun.org cat relay/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: ssh root@staging-ipv4.testrun.org "cd relay && CHATMAIL_DOMAIN2=ci-chatmail.testrun.org scripts/cmdeploy test --slow --ssh-host localhost"
-
-      - name: cmdeploy dns
-        run: ssh root@staging-ipv4.testrun.org "cd relay && scripts/cmdeploy dns -v --ssh-host localhost"
-

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

@@ -1,97 +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: |
-          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
-

cmdeploy/src/cmdeploy/cmdeploy.py

@@ -260,10 +260,10 @@ def test_cmd(args, out):
     pytest_args = [
         pytest_path,
         "cmdeploy/src/",
-        "-n4",
         "-rs",
         "-x",
-        "-v",
+        "-vv" if args.verbose > 1 else "-v",
+        "-s",
         "--durations=5",
     ]
     if args.slow:

cmdeploy/src/cmdeploy/dovecot/deployer.py

@@ -145,7 +145,7 @@ def _configure_dovecot(config: Config, debug: bool = False) -> (bool, bool):
         if not can_modify:
             print(
                 "\n!!!! refusing to attempt sysctl setting in shared-kernel containers\n"
-                f"!!!! dovecot: sysctl {key!r}={value}, should be >65534 for production setups\n"
+                f"!!!! dovecot: sysctl {key!r}={value}, should be >65535 for production setups\n"
                 "!!!!"
             )
             continue

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/v0.6.0/filtermail-{arch}"
+        url = f"https://github.com/chatmail/filtermail/releases/download/v0.5.2/filtermail-{arch}"
         sha256sum = {
-            "x86_64": "3fd8b18282252c75a5bbfa603d8c1b65f6563e5e920bddf3e64e451b7cdb43ce",
+            "x86_64": "ce24ca0075aa445510291d775fb3aea8f4411818c7b885ae51a0fe18c5f789ce",
-            "aarch64": "2bd191de205f7fd60158dd8e3516ab7e3efb14627696f3d7dc186bdcd9e10a43",
+            "aarch64": "c5d783eefa5332db3d97a0e6a23917d72849e3eb45da3d16ce908a9b4e5a797d",
         }[arch]
         self.need_restart |= files.download(
             name="Download filtermail",

cmdeploy/src/cmdeploy/lxc/cli.py

@@ -4,7 +4,7 @@ import os
 import time
 
 from ..util import get_git_hash, get_version_string, shell
-from .incus import RELAY_IMAGE_ALIAS, Incus, RelayContainer
+from .incus import Incus, RelayContainer
 
 RELAY_NAMES = ("test0", "test1")
 
@@ -47,6 +47,7 @@ def _lxc_start_cmd(args, out):
     out.green("Ensuring DNS container (ns-localchat) ...")
     dns_ct = ix.get_dns_container()
     dns_ct.ensure()
+    dns_ct.ensure_cached_as_image()
     sub.print(f"DNS container IP: {dns_ct.ipv4}")
 
     names = args.names if args.names else RELAY_NAMES
@@ -116,17 +117,39 @@ def _lxc_start_cmd(args, out):
 
     # Optionally run cmdeploy run + dns on each relay
     if args.run:
+        local_hash = get_git_hash()
         for ct in relays:
+            status = _deploy_status(ct, local_hash, ix)
             with out.section(f"cmdeploy run: {ct.sname} ({ct.domain})"):
-                ret = _run_cmdeploy("run", ct, ix, out, extra=["--skip-dns-check"])
+                if "IN-SYNC" in status:
-                if ret:
+                    out.print(f"{ct.sname} is {status}, skipping")
-                    out.red(f"Deploy to {ct.sname} failed (exit {ret})")
+                else:
-                    return ret
+                    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
+                    # Cache a per-relay image after each successful deploy
+                    # so the next run can launch directly from the deployed state.
+                    with out.section(f"lxc-test: caching {ct.sname} image"):
+                        ct.ensure_cached_as_image()
 
-        with out.section("loading DNS zones"):
+        # Restart mail services to flush stale DNS state.
-            for ct in relays:
+        # Cached container images boot with a resolv.conf
+        # pointing to the previous run's DNS IP;
+        # configure_dns() already restarted unbound,
+        # but postfix/dovecot may hold stale results
+        # from the window between boot and DNS fix.
+        for ct in relays:
+            out.print(f"Restarting mail services on {ct.name} ...")
+            ct.bash("systemctl restart postfix dovecot opendkim")
+
+        for ct in relays:
+            with out.section(f"cmdeploy dns: {ct.sname} ({ct.domain})"):
                 ret = _run_cmdeploy(
-                    "dns", ct, ix, out,
+                    "dns",
+                    ct,
+                    ix,
+                    out,
                     extra=["--zonefile", str(ct.zone)],
                 )
                 if ret:
@@ -134,7 +157,10 @@ def _lxc_start_cmd(args, out):
                     return ret
                 if ct.zone.exists():
                     dns_ct.set_dns_records(ct.zone.read_text())
-                out.print(f"Restarting filtermail-incoming on {ct.name}")
+                # Restart filtermail so its in-process DNS cache
+                # does not hold stale negative DKIM responses
+                # from before the zones were loaded.
+                out.print(f"Restarting filtermail-incoming on {ct.name} ...")
                 ct.bash("systemctl restart filtermail-incoming")
 
 
@@ -209,71 +235,26 @@ def lxc_test_cmd(args, out):
     """
     ix = Incus(out)
     t_total = time.time()
-    relay_names = list(RELAY_NAMES)
+    v_flag = " -" + "v" * out.verbosity if out.verbosity > 0 else ""
-    if args.one:
-        relay_names = relay_names[:1]
 
-    local_hash = get_git_hash()
+    ret = out.shell(f"cmdeploy lxc-start{v_flag} --run test0", cwd=str(ix.project_root))
+    if ret:
+        return ret
 
-    # Per-relay: start, deploy, then snapshot the first relay as a
+    if not args.one:
-    # reusable image so the second relay launches pre-deployed.
+        ret = out.shell(
-    ipv4_only_flags = {RELAY_NAMES[0]: False, RELAY_NAMES[1]: True}
+            f"cmdeploy lxc-start{v_flag} --run test1 --ipv4-only",
-
+            cwd=str(ix.project_root),
-    for ct in map(ix.get_container, relay_names):
+        )
-        name = ct.sname
+        if ret:
-        ipv4_only = ipv4_only_flags.get(name, False)
+            return ret
-        v_flag = " -" + "v" * out.verbosity if out.verbosity > 0 else ""
-        start_cmd = f"cmdeploy lxc-start{v_flag} {name}"
-        if ipv4_only:
-            start_cmd += " --ipv4-only"
-        with out.section(f"cmdeploy lxc-start: {name}"):
-            ret = out.shell(start_cmd, cwd=str(ix.project_root))
-            if ret:
-                return ret
-
-        status = _deploy_status(ct, local_hash, ix)
-        with out.section(f"cmdeploy run: {name}"):
-            if "IN-SYNC" in status:
-                out.print(f"{name} is {status}, skipping")
-            else:
-                ret = _run_cmdeploy("run", ct, ix, out, extra=["--skip-dns-check"])
-                if ret:
-                    out.red(f"Deploy to {name} failed (exit {ret})")
-                    return ret
-
-        # Snapshot the first relay so subsequent ones launch pre-deployed
-        if not ix.find_image([RELAY_IMAGE_ALIAS]):
-            with out.section("lxc-test: caching relay image"):
-                ct.publish_as_relay_image()
-
-    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(f"lxc-test: loading DNS zones {' & '.join(relay_names)}"):
-        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)
-
-        # Restart filtermail so its in-process DNS cache
-        # does not hold stale negative DKIM responses
-        # from before the zones were loaded.
-        for ct in map(ix.get_container, relay_names):
-            out.print(f"Restarting filtermail-incoming on {ct.name} ...")
-            ct.bash("systemctl restart filtermail-incoming")
 
     with out.section("cmdeploy test"):
-        first = ix.get_container(relay_names[0])
+        first = ix.get_container("test0")
         env = None
-        if len(relay_names) > 1:
+        if not args.one:
             env = os.environ.copy()
-            env["CHATMAIL_DOMAIN2"] = ix.get_container(relay_names[1]).domain
+            env["CHATMAIL_DOMAIN2"] = ix.get_container("test1").domain
         ret = _run_cmdeploy("test", first, ix, out, **({"env": env} if env else {}))
         if ret:
             out.red(f"Tests failed (exit {ret})")

cmdeploy/src/cmdeploy/lxc/incus.py

@@ -14,9 +14,9 @@ DOMAIN_SUFFIX = ".localchat"
 UPSTREAM_IMAGE = "images:debian/12"
 BASE_IMAGE_ALIAS = "localchat-base"
 BASE_SETUP_NAME = "localchat-base-setup"
-RELAY_IMAGE_ALIAS = "localchat-relay"
 
 DNS_CONTAINER_NAME = "ns-localchat"
+DNS_IMAGE_ALIAS = "localchat-ns"
 DNS_DOMAIN = "ns.localchat"
 
 
@@ -184,7 +184,16 @@ class Incus:
         )
         if result.returncode != 0:
             return None
-        return json.loads(result.stdout)
+        try:
+            return json.loads(result.stdout)
+        except json.JSONDecodeError as e:
+            msg = f"Incus JSON processing failed for {args!r}: {e!s}"
+            self.out.red(msg)
+            self.out.red(f"Captured stdout: {result.stdout!r}")
+            self.out.red(f"Captured stderr: {result.stderr!r}")
+            if check:
+                raise
+            return None
 
     def run_output(self, args, check=True):
         """Run an incus command and return its stripped stdout.
@@ -207,8 +216,13 @@ class Incus:
         return None
 
     def delete_images(self):
-        """Delete the cached base and relay images."""
+        """Delete localchat-base and per-container images."""
-        for alias in (RELAY_IMAGE_ALIAS, BASE_IMAGE_ALIAS):
+        for alias in [
+            BASE_IMAGE_ALIAS,
+            DNS_IMAGE_ALIAS,
+            "localchat-test0",
+            "localchat-test1",
+        ]:
             self.run(["image", "delete", alias], check=False)  # ok if absent
 
     def list_managed(self):
@@ -238,7 +252,7 @@ class Incus:
     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
+        The image is cached 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.
@@ -338,7 +352,7 @@ class Container:
 
     def launch(self):
         """Launch from the best available image, return the alias used."""
-        image = self.incus.find_image([RELAY_IMAGE_ALIAS, BASE_IMAGE_ALIAS])
+        image = self.incus.find_image([BASE_IMAGE_ALIAS])
         if not image:
             raise RuntimeError(
                 f"No base image '{BASE_IMAGE_ALIAS}' found. "
@@ -419,6 +433,18 @@ class Container:
                     parts = line.split()
                     return int(parts[2]), int(parts[1])
 
+    def ensure_cached_as_image(self):
+        """Cache this container as a respective image."""
+        alias = self.image_alias
+        if self.incus.find_image([alias]):
+            return
+        self.out.print("  Cleaning apt cache before caching image ...")
+        self.bash("apt-get clean")
+        self.out.print(f"  Caching {self.name!r} as '{alias}' ...")
+        self.incus.run(["publish", self.name, f"--alias={alias}", "--force"])
+        self.out.print(f"  Image '{alias}' cached.")
+        self.wait_ready()
+
 
 class RelayContainer(Container):
     """Container handle for a chatmail relay.
@@ -434,12 +460,22 @@ class RelayContainer(Container):
             domain=f"_{name}{DOMAIN_SUFFIX}",
         )
         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 potentially cached image) and clear inherited chatmail-version."""
+        """Launch from localchat-{sname} if cached, else localchat-base."""
-        image = super().launch()
+
+        candidates = [self.image_alias]
+        candidates.append(BASE_IMAGE_ALIAS)
+        image = self.incus.find_image(candidates)
+        assert image, f"No deployment base, candidates: {','.join(candidates)}"
+        self.out.print(f"  Launching from '{image}' image ...")
+        cfg = []
+        cfg += ("-c", f"{LABEL_KEY}=true")
+        cfg += ("-c", f"user.localchat-domain={self.domain}")
+        self.incus.run(["launch", image, self.name, *cfg])
         self.bash("rm -f /etc/chatmail-version")
         return image
 
@@ -474,22 +510,6 @@ class RelayContainer(Container):
             echo '{ip} {self.name} {self.domain}' >> /etc/hosts
         """)
 
-    def publish_as_relay_image(self):
-        """Publish this container as a reusable relay image.
-
-        Stops the container, 'publishes' it as 'localchat-relay', then restarts it.
-        """
-        if self.incus.find_image([RELAY_IMAGE_ALIAS]):
-            return
-        self.out.print(
-            f"  Locally caching {self.name!r} as '{RELAY_IMAGE_ALIAS}' image ..."
-        )
-        self.incus.run(
-            ["publish", self.name, f"--alias={RELAY_IMAGE_ALIAS}", "--force"]
-        )
-        self.wait_ready()
-        self.out.print(f"  Relay image '{RELAY_IMAGE_ALIAS}' ready.")
-
     def deployed_version(self):
         """Read /etc/chatmail-version, or None if absent."""
         return self.bash("cat /etc/chatmail-version", check=False)
@@ -572,6 +592,21 @@ class DNSContainer(Container):
 
     def __init__(self, incus):
         super().__init__(incus, DNS_CONTAINER_NAME, domain=DNS_DOMAIN)
+        self.image_alias = DNS_IMAGE_ALIAS
+
+    def launch(self):
+        """Launch from localchat-ns if cached, else localchat-base."""
+        image = self.incus.find_image([DNS_IMAGE_ALIAS, BASE_IMAGE_ALIAS])
+        if not image:
+            raise RuntimeError(
+                f"No base image '{BASE_IMAGE_ALIAS}' found. "
+                "Call ensure_base_image() before launching containers."
+            )
+        self.out.print(f"  Launching from '{image}' image ...")
+        cfg = []
+        cfg += ("-c", f"{LABEL_KEY}=true")
+        cfg += ("-c", f"user.localchat-domain={self.domain}")
+        self.incus.run(["launch", image, self.name, *cfg])
 
     def pdnsutil(self, *args, check=True):
         """Run ``pdnsutil <args>`` inside the DNS container."""

cmdeploy/src/cmdeploy/opendkim/opendkim.conf

@@ -60,6 +60,10 @@ PidFile			/run/opendkim/opendkim.pid
 # by the package dns-root-data.
 TrustAnchorFile		/usr/share/dns/root.key
 
+# Use the local unbound resolver rather than querying root servers directly.
+# on IPv4-only hosts opendkim may otherwise try ipv6 requests and time out. 
+Nameservers		127.0.0.1
+
 # Sign messages when `-o milter_macro_daemon_name=ORIGINATING` is set.
 MTA ORIGINATING
 

cmdeploy/src/cmdeploy/tests/online/test_0_login.py

@@ -89,9 +89,7 @@ def test_concurrent_logins_same_account(
         assert login_results.get()
 
 
-def test_no_vrfy(cmfactory, chatmail_config):
+def test_no_vrfy(chatmail_config):
-    ac = cmfactory.get_online_account()
-    addr = ac.get_config("addr")
     domain = chatmail_config.mail_domain
 
     s = smtplib.SMTP(domain)
@@ -100,7 +98,7 @@ def test_no_vrfy(cmfactory, chatmail_config):
     s.putcmd("vrfy", f"wrongaddress@{chatmail_config.mail_domain}")
     result = s.getreply()
     print(result)
-    s.putcmd("vrfy", addr)
+    s.putcmd("vrfy", f"echo@{chatmail_config.mail_domain}")
     result2 = s.getreply()
     print(result2)
     assert result[0] == result2[0] == 252