fix: stop build-publish-deploy red on report-status (#81) #15
Loading…
Add table
Add a link
Reference in a new issue
No description provided.
Delete branch "%!s()"
Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?
Scope
Implementation ticket for the bug reported in coilysiren/backend#81. Every push since 2026-05-21 leaves the run red on the final
report status to datastorestep, even though the deploy and rollout succeed. Latest red: run 26430605589 on5c4cc59. Goal here is to stop the red without changing the success criteria for the deploy itself.Pick one
http://api/documentwith the api node's tailnet IP, fetched from SSM (/coilysiren/api/tailnet-ipor equivalent, stash if not there yet per AGENTS.md "cache ids on first lookup"). Removes the MagicDNS dependency entirely.curlin a short retry loop (for i in 1 2 3 4 5; do curl -fsS ... && break || sleep 5; done). Cheapest fix; survives MagicDNS slow-start without touching topology.continue-on-error: trueon the step. Surfaces the failure in the run summary but keeps the run green. Acceptable if the datastore record is best-effort, not the canonical deploy signal.Fix idea 1 is the most durable, fix idea 2 is the smallest diff, fix idea 3 is the laziest. Recommendation: try retry first, fall back to tailnet-IP if the resolve never lands within the retry budget.
Acceptance
Out of scope
Resolves coilysiren/backend#81.
Merged into #3 in the 2026-05-29 backlog burn-down. Impl ticket for same report-status bug as #3 Reopen if it should stand alone.