ci: deploy via Komodo stack webhook, tag images with commit sha #3

Merged
trex Bot merged 1 commits from fix/komodo-deploy into main 2026-11-19 10:42:21 -06:00

Problem

The Deploy app step POSTs to http://10.5.0.11:7777/v1/update. Nothing listens there any more (Connection refused, no container on cygnus), so every push builds and pushes a new :latest image and then fails at the deploy step — the site never picks it up.

Change

  • Replace the dead webhook with the Komodo stack webhook for jonbio: POST http://10.5.0.3:9120/listener/github/stack/jonbio/deploy, HMAC-SHA256 signed (X-Hub-Signature-256) with a per-stack secret.
  • Tag each build with the commit sha alongside :latest, so a bad deploy can be rolled back to a known image.

Komodo side (already applied)

  • webhook_secret set on stack jonbio (per-stack scope — it can only deploy this one stack).
  • webhook_force_deploy: true so the webhook always pulls and recreates, even though the linked repo (homelab-ops) did not change.
  • auto_pull / auto_update were already true.

Verified

  • Signed POST to the webhook returns 200 and produces a Komodo DeployStack update with Compose Pull + Compose Up, status Complete.
  • The exact run: block from this workflow was executed in a node:22-alpine container with the real secret: Komodo deploy -> 200.

Required before merge

Add repo secret KOMODO_JONBIO_WEBHOOK_SECRET (value handed over separately). The step fails with a clear error if it is missing.

## Problem The `Deploy app` step POSTs to `http://10.5.0.11:7777/v1/update`. Nothing listens there any more (`Connection refused`, no container on cygnus), so every push builds and pushes a new `:latest` image and then fails at the deploy step — the site never picks it up. ## Change - Replace the dead webhook with the Komodo stack webhook for `jonbio`: `POST http://10.5.0.3:9120/listener/github/stack/jonbio/deploy`, HMAC-SHA256 signed (`X-Hub-Signature-256`) with a per-stack secret. - Tag each build with the commit sha alongside `:latest`, so a bad deploy can be rolled back to a known image. ## Komodo side (already applied) - `webhook_secret` set on stack `jonbio` (per-stack scope — it can only deploy this one stack). - `webhook_force_deploy: true` so the webhook always pulls and recreates, even though the linked repo (`homelab-ops`) did not change. - `auto_pull` / `auto_update` were already `true`. ## Verified - Signed POST to the webhook returns `200` and produces a Komodo `DeployStack` update with `Compose Pull` + `Compose Up`, status `Complete`. - The exact `run:` block from this workflow was executed in a `node:22-alpine` container with the real secret: `Komodo deploy -> 200`. ## Required before merge Add repo secret `KOMODO_JONBIO_WEBHOOK_SECRET` (value handed over separately). The step fails with a clear error if it is missing.
trex Bot added 1 commit 2026-11-18 15:26:04 -06:00
The old deploy step POSTed to http://10.5.0.11:7777/v1/update, which no
longer exists (connection refused), so every push built and pushed a new
image but never reached the site.

Replace it with the Komodo stack webhook for `jonbio`
(/listener/github/stack/jonbio/deploy), HMAC-signed with a per-stack
secret. The stack has auto_pull + webhook_force_deploy enabled, so the
webhook pulls :latest and recreates the container.

Also tag each build with the commit sha alongside :latest so a bad
deploy can be rolled back to a known image.
trex Bot merged commit eb0c0ba77b into main 2026-11-19 10:42:21 -06:00
Sign in to join this conversation.
No Reviewers
No labels
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: jblu/jonbio#3