Skip to content

Prune the image's dormant docker and nginx from co-processor (or declare keep) #143

Open

Description

@gregoryfoster

The 2026-11-17 patch run (#113, record) left two findings open, both dormant components:

  • named:docker: idle, but deploy/ names it, so the operator reads what that says, then either proposes the prune or declares keep:docker.
  • dormant:nginx: idle, nothing names it. The skill's verdict is to prune it or declare keep:nginx.

They are new findings, not new state: the 11-10 run reported none. The probe has reported them since the vendored skill was bumped (217364d, 8a1b79f). Until each is pruned or excepted, every patch run's before- and after-probe lists both.

What the host shows (read 2026-11-17, non-root unless noted)

Both came with the image. One apt-get install -y … nginx … docker.io docker-buildx docker-compose-v2 … at 2026-09-29 17:12Z, the exeuntu build. The skill's profile says the same of nginx: it "ships with the image, disabled, and was in the security set on six hosts: patched for nothing."

docker:

  • Installed: docker.io 29.1.3, containerd 2.2.1, runc 1.3.4, docker-buildx 0.30.1, docker-compose-v2 2.40.3.
  • docker.service, docker.socket and containerd.service are all disabled. docker.socket was disabled 2026-10-05 (Contain the extraction child: it runs as the service user and can reach the service's secrets #2 gate 1, docs/PROVISIONING.md), since /run/docker.sock is root to the docker group.
  • The probe read it as inactive for 49 days, never started since boot, 0 containers, 0 volumes.
  • exedev is still in group docker (gid 107). That is root-equivalent the moment Docker comes back. A purge deletes the group (pruning.md, group; cf. co-watcher leftovers: docker group (exedev member) and the retired notifier backup key watcher#336).
  • What the repo says about it is never a use on this host:
    • deploy/processor.service: processor is not in group docker;
    • deploy/apt/apt.conf.d/processor-periodic: the name sorts after the image's docker-disable-periodic-update. That file is owned by no package (dpkg -S finds none), so a purge leaves it, and tests/test_apt_automatic.py keeps reading it;
    • deploy/disk-gc/disk-gc.conf: no docker tier, because the daemon is inactive;
    • tests/test_child.py: the child refuses socket() before any connect, so /run/docker.sock need not exist;
    • CI's redis:7.0.15 service runs on GitHub's runners, not here. Tests here use the apt redis-server.

nginx:

  • Installed: nginx and nginx-common 1.24.0-2ubuntu7.19.
  • nginx.service is disabled. The probe read it as inactive 48 days; /var/log/nginx is all empty files, the newest 2026-09-29.
  • sites-enabled/default is the image's link. Nothing in this repo names nginx.

Proposed route (the vendored patching-hosts skill, references/pruning.md)

For each component, one stage at a time. Each stage gets its own approval in this repo's session, and every prune-plan.sh/prune.sh run escalates through sudo -n, so each is a sudo step.

  1. Plan:
    • prune-plan.sh --component docker --out <file> and --component nginx --out <file>. They change nothing. Attach the JSON or summarise it here: units, packages (remove/purge sets, held), autoremove_would_take, reverse_depends, purge_scripts, paths, conffiles, group.
    • Read autoremove_would_take closely for docker. On noble, docker.io with its recommends orphans about 47 packages, iptables, git and openssh-client among them. Here git and openssh-client were named in the same install line, so they should be manual and stay. The plan confirms it.
  2. Decide docker: prune, or exception keep:docker <review-by> <reason>. Nothing found here needs it.
  3. Disable:
    • Command: prune.sh --stage disable --plan <fresh plan> --approve. It stops, disables and masks every unit the packages ship.
    • Knob line: exception disabled:<name> <review-by> <reason>. Its review-by date is the soak: 30 days by default.
    • The knob is a RUNTIME_FILES path, so the knob change is a PR and a deploy.
  4. Savepoint (optional, preferred), taken during the soak: prune.sh --stage savepoint --plan <fresh plan> --approve. It tarballs the config and conffiles at mode 600. With it, the remove stage is skipped, and no rc residue is left holding config.
  5. Purge, once the soak ends:
    • Command: prune.sh --stage purge --plan <fresh plan> --approve --savepoint <dir>.
    • nginx-common's purge script deletes /etc/nginx and /var/log/nginx, so its purge needs --purge-data /var/log/nginx.
    • docker.io's purge leaves /var/lib/docker, which holds 0 containers and 0 volumes here. Name it with --purge-data only if the plan lists it and the operator wants it gone.
    • Knob line: exception purged:<name> <review-by> <reason>, which stays until the image stops shipping it.
  6. Docs:
    • docs/PROVISIONING.md: the docker.socket row, and the docker group note.
    • docs/MONITORING.md and deploy/disk-gc/disk-gc.conf: the "daemon is inactive" reason for no docker tier.
    • docs/PATCHING.md: the knob table, for the new exception lines.

Run it outside a patch window, never during one, and pick the order with the operator. nginx is the simpler first: no group, no reverse dependants expected.

The base image

pruning.md: "Where the component shipped with the image on every host, the prune belongs in the base image instead … Until it gets there, a prune on one host is drift from that image, and the knob declares it." Both ship in exeuntu on every cohort host. A note to the image's owner is a cross-repo post, so it is drafted here first and posted only with the operator's approval (AGENTS.md § Cross-repo Policy).

Done when

  • The probe on co-processor reports neither named:docker nor dormant:nginx as a finding, either because each is purged (knob purged:) or because each is declared keep: with a reason.
  • exedev is out of group docker, or keep:docker states why it stays.
  • The docs above match the host.

Activity

  1. gregoryfoster commented on Nov 17, 2026

    @gregoryfoster
    MemberAuthor

    Brief: #143, prune the dormant docker and nginx (coordinator, epic #84, 2026-11-17)

    Authorities: this issue's body is the spec. This brief adds a pre-flight against origin/main 8a1b79f and the host. The vendored skill at 063ba5db (skills/patching-hosts/references/pruning.md) governs the mechanics. Where the body and the skill disagree on mechanics, follow the skill. Where they disagree on scope, stop and ask. Labels: [measured] read from the host or a registry in this pre-flight; [traced] followed through the repo or a sibling's; [stated] an issue or a person says so.

    What the host shows now [measured, 2026-11-17 ~20:00Z, non-root]

    The body's inventory holds:

    • Packages, all installed: docker.io 29.1.3, containerd 2.2.1, runc 1.3.4, docker-buildx 0.30.1, docker-compose-v2 2.40.3, nginx and nginx-common 1.24.0-2ubuntu7.19.
    • Units: docker.service, docker.socket, containerd.service and nginx.service are each disabled and inactive. None is masked yet. /run/docker.sock is absent.
    • Group: docker:x:107:exedev.

    Two corrections and one refinement:

    1. The docker group adds no access for exedev here. exedev is in group sudo, and sudo -n works, so it already has root. processor (uid 997) is in no group but its own. So the group is hygiene, not exposure: deleting it at the purge is right, and Tier 3 is the right tier. The body's "root-equivalent the moment Docker comes back" is true in general, but grants exedev nothing it lacks.

    2. Upstream's "about 47 orphans" for docker.io doesn't apply here. apt-get -s purge --autoremove on the five docker packages takes 9 packages:

      • the five themselves;
      • pigz, the only new orphan;
      • bridge-utils, dns-root-data and dnsmasq-base.

      git, openssh-client, docker.io, docker-buildx, docker-compose-v2 and nginx are marked manual. iptables stays: tailscale and ubuntu-standard depend on it.

    3. bridge-utils, dns-root-data and dnsmasq-base are orphans already. apt-get -s autoremove takes them today, before any prune. That's why the nginx purge simulation also lists them. prune.sh never autoremoves (pruning.md: "No autoremove, ever"), so they stay unless the plan takes them by name with --take-orphans. The weighed choice is below.

    prune-plan.sh is the reading of record. Compare its autoremove_would_take and packages against the above, and say on the issue where they differ.

    Cohort precedents [traced: each repo's .skills/patching-hosts]

    Host Knob line Soak
    CannObserv/archiver#375 (host co-registrar) disabled:nginx and disabled:docker, masked 2026-11-13 review-by 2026-11-20, 7 days
    CannObserv/watcher#440 disabled:nginx, masked 2026-11-17 review-by 2026-11-24, 7 days
    CannObserv/index disabled:nginx review-by 2026-12-13, about 30 days
    • Template: Adopt patching-hosts: commit the knob, then take the first monthly run under the skill (gregoryfoster/skills#313) archiver#375 for the disable stage and the knob test. Finish the nginx and docker prunes: savepoint + purge from 2026-11-20 (from #375) archiver#512 (the savepoint and purge, from 2026-11-20) hasn't started, so the purge stage has no cohort run to copy yet. This one may be the first.
    • The knob test: tests/test_patching_knob.py::test_the_knob_reads_clean asserts findings == []. A disabled: line's review-by passing turns into a finding, and CI goes red by calendar. Archiver's tests/deploy/test_patching_knob.py (_assert_clean, PRUNE_STAGES = ("disabled:", "removed:")) lets an expired exception through only on a prune-stage line, and keeps every other finding red. pruning.md asks for exactly that: "A repo that tests its knob allows soak-ended on a disabled: or removed: line". Port it test-first, before the first disabled: line lands. Whether the reader reports it as soak-ended or expired-exception at this pin is for you to read from read-knob.sh; archiver's filter matches expired-exception.

    Shape: two phases, two deploys

    .skills/patching-hosts is a RUNTIME_FILES path (src/processor/drift.py:65). So each knob change is a PR, then scripts/deploy.sh, which restarts processor for a few seconds. Deploys don't wait for a window (PATCHING.md).

    • Phase 1, now:
      • prune-plan.sh for each component;
      • the operator decisions below;
      • disable for both;
      • the savepoint for each, during the soak;
      • the knob gets exception disabled:<name> <review-by> <reason> plus the ported test, as one PR and one deploy, with the docs left as they are (the host still matches them while the units are only masked);
      • pause. The PR says Refs #143, not Closes.
    • Phase 2, after the review-by date:
      • a fresh plan, then purge with --savepoint;
      • the knob gets exception purged:<name> <review-by> <reason>, since exeuntu keeps shipping both;
      • the docs (list below);
      • one PR, one deploy;
      • the probe reads neither finding;
      • then close.
    • Timing: keep every stage out of the patch window, Tue 2026-11-24 05:00–10:00Z. A 7-day soak from 2026-11-18 ends after that window, so the purge lands clear of it.

    Operator decisions

    • [Operator decision: docker, prune it or keep:docker — recommend prune. Nothing on this host uses it (the body's survey, and processor isn't in the group), and archiver pruned both.]
    • [Operator decision: soak length — recommend 7 days, as archiver and watcher chose. The skill's default is 30; an earlier date is the owner's call (pruning.md § The calendar). Neither component's units have started since the image was built.]
    • [Operator decision: --take-orphans — recommend no. It would name pigz (docker's one new orphan) and pull in the three orphans that predate the prune. They're harmless, and taking them widens the change past the issue.]
    • [Operator decision: --purge-data /var/lib/docker — recommend yes, if the plan lists it. It's root-only, with 0 containers and 0 volumes per the patch run's probe [stated: #113 comment 6956497896]. --purge-data /var/log/nginx is required, per the body.]

    Starting the thread approves the recommendations. Every prune-plan.sh and prune.sh run escalates through sudo -n by itself (this issue's body; PATCHING.md:61 for probe.sh), so each one is a sudo step that needs approval in this repo's session. A plan binds one stage and expires in 24 h: take a fresh one for each stage.

    Every mention on origin/main [traced: git grep -i 'docker|nginx', skills-vendor and uv.lock excluded]

    The body names PROVISIONING, MONITORING, disk-gc.conf and PATCHING. The search finds more. Each row is a recommendation to weigh:

    Place Says Phase 2
    docs/PROVISIONING.md:19, :175, :193 the docker.socket row, its rationale, and the is-enabled docker.socket check retire in place: "purged , #143"; the check becomes dpkg -s docker.io absent
    docs/PROVISIONING.md:18, AGENTS.md:40, deploy/processor.service:12 "processor … not in docker" leave: still true. AGENTS.md is at 5,945 of 6,000 tokens, so don't grow it
    deploy/disk-gc/disk-gc.conf:12, docs/MONITORING.md:132 no docker tier "because the daemon is inactive" the reason becomes "docker is purged (#143)". tests/test_disk_gc.py:68 still holds
    docs/PATCHING.md:71 held step docker: "Docker and containerd are disabled and inactive" "purged (#143); nothing to restart". Plus wherever PATCHING.md describes the knob's exceptions. The body's "knob table" has no heading of that name [traced: grep]
    docs/SOCRATICODE.md:144-145, not in the body "No Docker here: docker.socket is disabled (#2)" "docker is purged (#143)"
    tests/test_child.py:380 /run/docker.sock parameter, "each when present" leave. It skips today already (the socket is absent), and stays skipped after the purge. That's expected, not lost coverage: the child refuses socket() before any path
    tests/test_apt_automatic.py:55, deploy/apt/apt.conf.d/processor-periodic:5 the image's docker-disable-periodic-update leave. No package owns /etc/apt/apt.conf.d/docker-* [measured: dpkg -S]; they're the image's apt files, so the purge leaves them
    spec :345, :365; src/processor/_contain.py:14; docs/plans/* rationale and history leave
    .github/workflows/ci.yml, tests/test_ci.py Docker Hub images on GitHub's runners unrelated

    Doc headroom (exact counts from the 2026-11-10 measurement; sizes unchanged or smaller since): PATCHING.md 8,690 / 10,000; PROVISIONING.md 8,299; MONITORING.md 7,629; AGENTS.md 5,945 / 6,000. Phase 2 should be net-neutral.

    The base image

    CannObserv/provisioner's co-image-base is FROM exeuntu with no apt removal (image/Dockerfile) [traced]. So moving co-processor onto it (CannObserv/provisioner#16, group 3) drops neither package, and the purged: lines outlive the move. No CannObserv/provisioner issue asks the image to drop them yet [measured: issue search]. Archiver, watcher and index carry the same drift. Draft the note on this issue and stop. It's a cross-repo post, and the operator approves it (AGENTS.md § Cross-repo Policy). It doesn't block either phase.

    Who closes, and when

    Out of scope

  2. gregoryfoster commented on Nov 17, 2026

    @gregoryfoster
    MemberAuthor

    Phase 1 pre-flight: corrections to the brief (2026-11-17 ~21:10Z)

    Read against origin/main 8a1b79f, the skill at 063ba5db, and the host. The operator approved the brief's four recommendations (prune docker, 7-day soak, no --take-orphans, --purge-data /var/lib/docker if listed) and the two prune-plan.sh runs below.

    The plans [measured: prune-plan.sh, 2026-11-17T21:06Z, both exit 0, not_read empty]

    They match the brief on packages, units, holds and the group:

    • nginx: remove = purge = nginx, nginx-common. No drag-along, nothing held, no reverse dependants. autoremove_would_take is empty: the three pre-existing orphans aren't nginx's, so the plan leaves them out.
    • docker: remove = purge = docker.io, containerd, runc, docker-buildx, docker-compose-v2. autoremove_would_take = pigz only. Group docker, members exedev.
    • Units: nginx.service, docker.socket, docker.service, containerd.service, each disabled/inactive, unit files under /usr/lib/systemd/system, so the disable masks all four.
    • Stage: both declared: none, next: disable.

    Corrections

    1. The reader reports soak-ended, not expired-exception, at this pin. _knob-lib.sh:657-670 (#636) gives a passed disabled:/removed: line kind soak-ended. Every other passed line, purged: included, stays expired-exception. knob.md:65 asks a repo's test to allow exactly that. Archiver's _assert_clean matches expired-exception, so copied verbatim it would go red on 2026-11-25. The port matches soak-ended on a prune-stage line and keeps everything else red.
    2. Archiver's soak test skips when the knob has no prune line. So it can't fail first. The port drives the filter with synthetic knobs (real knob + one dated line, --today past it): disabled: and removed: pass, purged: and keep: stay red. That's red before the helper exists, so the test comes first, ahead of the disabled: lines.
    3. Data paths the brief's purge-data decision doesn't cover (phase 2's call, not phase 1's):
      • docker: the plan lists /var/lib/containerd (96 KiB) as data, beside /var/lib/docker (184 KiB, only a buildkit/ dir). Both were created 2026-09-30 19:50Z, so the daemon ran once the day after the image build, and never since boot. /etc/docker is empty.
      • nginx: the plan lists /var/www (16 KiB, an empty html/) as data. nginx-common's purge script deletes /var/lib/nginx (empty), /var/log/nginx and /etc/nginx. So --purge-data /var/log/nginx is required, as the body says. /var/www stays unless named.
      • docker has no conffiles outside /etc/docker. nginx has four: /etc/default/nginx, /etc/init.d/nginx, /etc/logrotate.d/nginx and /etc/ufw/applications.d/nginx. The savepoint keeps them.
    4. Review-by date: 2026-11-24, the window day. The disable runs today, 2026-11-17, and a 7-day soak lands there. The reader expires an exception only after its date (today > review-by), so:
      • on 2026-11-24 both lines still hold, and that run's probe reads both components as kept, not as findings (probe.sh:120);
      • soak-ended, and with it the purge, is due from 2026-11-25, clear of the window.
    5. A masked docker.socket reads masked, not disabled. That affects docs/PROVISIONING.md:194's check (systemctl is-enabled docker.socket # disabled) and its row at :19. The brief leaves the docs to phase 2, so phase 1 leaves them, and the check's comment is stale while the unit is masked. Nothing automated runs that check (git grep finds no other caller), so no test goes red. Phase 2 retires it.
    6. src/processor/patching.py parses only window lines (parse_windows), so exception lines can't trip the hourly patching check. The knob is still a RUNTIME_FILES path, so the PR needs a deploy.

    No scope disagreement found, so phase 1 goes ahead as briefed.

  3. gregoryfoster commented on Nov 17, 2026

    @gregoryfoster
    MemberAuthor

    DRAFT, not posted: proposed new issue on CannObserv/provisioner (the base-image note)

    Held for the operator's approval (AGENTS.md § Cross-repo Policy). Nothing has gone to CannObserv/provisioner. Read for accuracy 2026-11-17: image/Dockerfile on its default branch, and issues #16, #41, #141.


    Title: co-image-base: exeuntu's dormant nginx and docker, pruned host by host, are drift from the image

    Body:

    Every exeuntu build installs nginx/nginx-common and docker.io, containerd, runc, docker-buildx and docker-compose-v2. They come in one apt-get install; on co-processor that ran 2026-09-29 17:12Z. co-image-base is FROM exeuntu, and image/Dockerfile carries no apt removal, so moving a host onto it (#16) keeps both.

    Hosts that pruned them, or are pruning them, through patching-hosts' prune.sh (pruning.md: "a prune on one host is drift from that image, and the knob declares it"):

    Host Stage
    co-registrar (CannObserv/archiver#375) disabled:nginx, disabled:docker; purge planned (archiver#512)
    CannObserv/watcher#440 disabled:nginx
    CannObserv/index disabled:nginx
    co-processor (#143) disabled:nginx, disabled:docker, masked 2026-11-17; purge from 2026-11-25

    Each ends with an exception purged:<name> line that stays as long as the image ships the component.

    Two things limit what the image can take on:

    1. The scope rule (image/Dockerfile, owner 2026-10-01): "only fixes every host needs", and "COPY and LABEL only, no RUN". An apt-get purge is a RUN, so a removal in the image breaks the rule as it stands.
    2. docker isn't dormant on every host. co-provisioner builds the image with docker buildx (#40 test: deterministic reclaim test; sweep of timing margins in tests/ #41). nginx has no exe.dev cohort user that we know of.

    Asks, in your order of preference:

    • nginx: drop it from the image, by a RUN exception to the rule or upstream in exeuntu (boldsoftware/exe). Or say it stays, and each host keeps its purged:nginx line as declared drift.
    • docker: confirm that it stays in the image, since co-provisioner needs it, and that host-level prunes with purged:docker are the intended shape for every other host.
    • Either way, a line in image/README.md naming the hosts that prune them would let the next image bump re-check the drift.

    Not blocking anything: #143's phase 2 goes ahead either way.

  4. gregoryfoster commented on Nov 17, 2026

    @gregoryfoster
    MemberAuthor

    Phase 1: done on the host, PR open, shipping paused (2026-11-17)

    PR #145 (Refs #143), head 961ab930274edafc4b2950c7cbfe7125187b4b3a (11e3f1c..961ab93, 5 commits). Not merged yet: the operator ships it. Corrections to the brief: the pre-flight comment.

    • Plans (prune-plan.sh, 21:06Z) match the brief. docker's only new orphan is pigz; nginx takes none.
    • Disable: prune.sh --stage disable --approve at 21:57Z masked nginx.service, docker.socket, docker.service and containerd.service. Each was disabled/inactive before. Logs: /var/backups/patching-hosts-prune-{nginx,docker}/disable.log.
    • Savepoints: taken at 23:54Z from fresh 23:02Z plans, each into its run directory (root, 700/600). nginx's config.tar (/etc/nginx plus 4 conffiles) is sha256 120a38d7245c8877cd1ce95ce7d83210eb8fcb51109d1100d5746b14788ad1b8. docker's (/etc/docker, empty) is 85c544c770f5475cce3c6a0058fea0911ab2e56ba1363c3e98070e4c1372e9a6.
    • Knob and test: exception disabled:nginx and disabled:docker, review-by 2026-11-24. The knob test now allows soak-ended on a disabled:/removed: line and keeps every other expiry red (purged: included), test-first. CR 1–3 fixed, 4 stet, 5 held (PR body).
    • Every prune-plan.sh and prune.sh run (12) was approved one by one in the session, and all ran outside the Tue 2026-11-24 05:00–10:00Z window.

    The soak

    Review-by 2026-11-24 for both. The reader expires a line only after its date. So the 11-24 window's probe reads both as kept, and soak-ended (the purge's prompt) is due from 2026-11-25, clear of the window.

    Post-merge host steps (operator, once CI is green on 961ab93)

    1. From /home/exedev/processor: git fetch origin && git merge --ff-only origin/143-prune-docker-nginx && git push origin main (docs/SKILLS.md § Merging under pr; never GitHub's button).
    2. Deploy: .skills/patching-hosts is a RUNTIME_FILES path (src/processor/drift.py:65). Run scripts/deploy.sh in the background and read journalctl -t processor-deploy for the verdict. processor restarts for a few seconds; no window needed. Skip it and the drift check alerts 8 h after the push.
    3. Verify:
      • cat /srv/processor/live/REVISION reads 961ab9302740;
      • systemctl is-active processor is active;
      • bash skills/patching-hosts/scripts/read-knob.sh --host co-processor from the checkout gives findings: [];
      • systemctl is-enabled nginx.service docker.socket docker.service containerd.service reads masked for each.
    4. Optional: a probe.sh dry run (a sudo step) should list docker and nginx under dormant with verdict kept, and neither named:docker nor dormant:nginx as a finding.
    5. Undo, if anything needs either component during the soak: sudo systemctl unmask the unit. Every unit was disabled/inactive before, so nothing to enable or start. Then drop its knob line in a PR.

    Follow-up:

    • Phase 2, from 2026-11-25: a fresh plan for each, then prune.sh --stage purge --approve --savepoint /var/backups/patching-hosts-prune-<name> with --purge-data; the knob's purged: lines; the docs (the brief's table); one PR and one deploy; then close Prune the image's dormant docker and nginx from co-processor (or declare keep) #143.
    • The CannObserv/provisioner base-image note is drafted here, waiting for the operator's approval to post.

    Known limits:

    • No probe ran after the disable. That the dormant findings read kept comes from probe.sh:120 and read-knob.sh, not from a probe reading.
    • docs/PROVISIONING.md:194's check (is-enabled docker.socket # disabled) and its row at :19 now mismatch the host (masked) until phase 2 retires them (CR 5, held to the brief's scope).
    • The filter tests' red is a mutation shown in-session, not a red commit (CR 4, stet).
    • Each savepoint holds config only. prune.sh says to dump the data off the node before the purge: /var/www (an empty html/) and /var/log/nginx (empty logs), /var/lib/docker (buildkit/ only) and /var/lib/containerd. Measured near-empty; phase 2 decides whether a dump is worth it.
    • exedev stays in group docker until the purge's groupdel. It already has sudo, so the group adds nothing (brief correction 1).

    For #143:

    • Phase 2's --purge-data:
      • /var/log/nginx is required: nginx-common's purge script deletes it, and the purge refuses until it's named.
      • The brief approved /var/lib/docker.
      • The plan also lists /var/lib/containerd (96 KiB) and /var/www (16 KiB), which no decision covers yet. Ask the operator.
    • The 24 h plans: the savepoint plans expire 2026-11-18T23:02Z. The purge needs a fresh plan anyway.
    • The declaration test: test_the_dormant_components_are_pruned accepts purged: lines and a staggered order (CR 1). test_the_knobs_own_soaks_end_green skips once no disabled: line is left.

    For #84:

  5. gregoryfoster commented on Nov 18, 2026

    @gregoryfoster
    MemberAuthor

    Phase 1 shipped (2026-11-18): PR #145 fast-forwarded main to 961ab930274edafc4b2950c7cbfe7125187b4b3a; CI passed on it (run). scripts/deploy.sh exited 0: /srv/processor/live/REVISION = 961ab930274e, containment required, the smoke run passed, processor is active with 0 warnings since. The live knob declares disabled:nginx and disabled:docker, and read-knob.sh gives findings: []. All four units read masked. Post-merge steps 1–3 are done; step 4 (a probe) didn't run. #143 stays open for phase 2, from 2026-11-25.

  6. gregoryfoster commented on Nov 18, 2026

    @gregoryfoster
    MemberAuthor

    The base-image note drafted above (comment 6964631233) is filed as CannObserv/provisioner#165, with the operator's approval (2026-11-18). One change from the draft: the archiver purge ref is written in full (CannObserv/archiver#512) so it links from the provisioner repo. It blocks nothing here; phase 2 goes ahead from 2026-11-25 either way.

  7. gregoryfoster commented on Nov 18, 2026

    @gregoryfoster
    MemberAuthor

    Operator decision (2026-11-18), for phase 2's --purge-data: the purge also deletes /var/lib/containerd and /var/www.

    So phase 2's paths are:

    • nginx: --purge-data /var/log/nginx (required: nginx-common's purge script deletes it) and --purge-data /var/www.
    • docker: --purge-data /var/lib/docker (decided 2026-11-17) and --purge-data /var/lib/containerd.

    As measured on 2026-11-17 by phase 1's plans [stated: the pre-flight comment above]:

    • /var/lib/containerd: 96 KiB;
    • /var/lib/docker: 184 KiB, a buildkit/ directory only;
    • /var/www: 16 KiB, an empty html/;
    • /var/log/nginx: empty logs.

    Phase 2 takes fresh plans, so re-read each path's contents from them before the purge. If any path holds more than the above (a container, a volume, a non-empty site, a non-empty log), stop and ask before deleting it. With these contents, no off-node dump is planned; the savepoints keep the config.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

No one assigned

    Labels

    enhancementNew feature or request

    Type

    No type

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions