| name | docker-smoke-container-hygiene |
| description | After local Data Boar Docker smoke tests, keep at most one or two named containers, reuse one local image tag (e.g. data_boar:lab) instead of many smoke-* tags, prune old images, and document cleanup. Use when creating containers, running lab validation, or closing a session that used docker run/compose/build for data_boar. |
Docker smoke container hygiene (Data Boar)
Stable release publish (Hub) — non-optional pieces
When the operator (or release-ritual) is cutting a published X.Y.Z image for Docker Hub:
- Disk: Confirm free space on the volume Docker uses on the primary dev PC before
docker-lab-build.ps1 (full disk ⇒ broken builds).
- Smoke before push: After
.\scripts\docker-lab-build.ps1, run at least one docker run --rm against data_boar:lab — python main.py --version or /health per docs/DOCKER_SETUP.md §7. Do not advise docker push to fabioleitao/data_boar if smoke failed (unless CI-only waiver is explicit).
- After push:
.\scripts\docker-prune-local.ps1 -WhatIf then without -WhatIf on the same machine — see docs/ops/DOCKER_IMAGE_RELEASE_ORDER.md.
Authoritative order: .cursor/rules/release-publish-sequencing.mdc (situational — release-ritual / @release-publish-sequencing.mdc); this skill and docker-local-smoke-cleanup.mdc stay always-on for hygiene.
When to use
- You (or the operator) ran
docker build, docker run, or docker compose for Data Boar (data_boar image, data-boar-* names).
- The session is ending or you are about to open a PR after smoke verification.
- Docker Desktop is accumulating stopped containers from repeated tests, or many local
data_boar:* / fabioleitao/data_boar:* image tags after smoke runs.
Instructions — containers
- Inventory local Data Boar-related containers (name filter is enough for most cases):
docker ps -a and look for data-boar, data_boar, or compose project service names from deploy/.
- Retention policy
- Normal: keep one container (or one compose stack) for day-to-day dev.
- A/B or regression: keep at most two intentionally named containers (e.g. different tags or configs). Anything else is throwaway.
- Cleanup suggestion (confirm with the operator before
rm -f):
- Stop:
docker stop <name>
- Remove:
docker rm <name>
- Compose:
docker compose -f deploy/docker-compose.yml … down (add override file if used).
- Document in the session or PR description if smoke was run and which container name/tag was kept, so the next run does not spawn duplicates.
- Agents: do not delete containers autonomously; propose commands and remind the operator to keep only 1–2 instances unless A/B is required.
Instructions — images and disk (homelab)
Smoke tests do not require a new Docker tag every time. Tags like data_boar:smoke-post93, data_boar:smoke-foo accumulate and waste disk (many digests/layers stay referenced until pruned).
- Default build command for local smoke:
docker build -t data_boar:lab -f Dockerfile . — reuse data_boar:lab and overwrite it on each build (same idea as docs/ops/HOMELAB_VALIDATION.md step 1.3).
- Second tag only when needed: e.g. true A/B (
data_boar:lab-a vs data_boar:lab-b) or compare pulled fabioleitao/data_boar:latest vs local data_boar:lab.
- Retention: keep roughly two useful lines locally (e.g. Hub
latest + data_boar:lab, or latest + one older semver). Remove stale tags: docker rmi data_boar:OLD_TAG, then docker image prune -f and optionally docker builder prune -f — see docs/ops/BRANCH_AND_DOCKER_CLEANUP.md §4.
- Scripts (Windows, repo root): scripts/docker-hub-pull.ps1 (pull Hub tags), scripts/docker-lab-build.ps1 (build
data_boar:lab + optional lab-prev/smoke), scripts/docker-prune-local.ps1 -WhatIf first — see scripts/docker/README.md.
- Agents: propose list/prune commands; do not run destructive
docker rmi / prune without operator confirmation.
Rationale
Stale containers confuse port mappings (8088), volume mounts, and Scout/quickview workflows; too many image tags waste disk on homelab hosts. Policy matches docs/DOCKER_SETUP.md §7, docs/ops/BRANCH_AND_DOCKER_CLEANUP.md §4, and docs/ops/HOMELAB_VALIDATION.md.
