Default branch: Release record did not succeed

[github-actions]

The newest completed run of Release record on main concluded failure.

https://github.com/Flowfin/site/actions/runs/33617011281

Read at 2026-09-02T09:59:08Z, by https://github.com/Flowfin/site/actions/runs/33859250623.

This entry is rewritten while that stays true, and closed once the newest completed run of that workflow on main succeeds. Nothing re-runs the job: a re-run that happens to pass turns a defect into an intermittent one.

Renaming this entry produces a second one on the next pass, because the title is what identifies it.

[iderex]

The closing condition, met and read rather than assumed

This entry closes once the newest completed run of Release record on main
succeeds. The merge of #233 closed it a minute before that was true, on the
Closes keyword, so the condition is recorded here rather than left to be
inferred from a merge.

The record was re-taken in #233 and merged as 766fb1b. The run over that
commit is a new run of the workflow on new content, not a re-run of the job this
entry was opened for:

gh run list --repo Flowfin/site --workflow "Release record" --branch main --limit 3 \
  --json databaseId,status,conclusion,createdAt,headSha,event \
  --jq '.[] | "\(.databaseId)\t\(.status)\t\(.conclusion)\t\(.createdAt)\t\(.headSha[0:7])\t\(.event)"'
33866306701	completed	success	2026-09-04T11:05:39Z	766fb1b	workflow_dispatch
33617011281	completed	failure	2026-09-02T09:58:29Z	f51005f	schedule
33543375761	completed	success	2026-09-01T18:23:51Z	6541274	workflow_dispatch

Run 2026-09-04. The failed run 33617011281 is untouched and still reads
failure, which is what keeps the defect from being turned into an intermittent
one.

What had actually moved

The run that opened this entry found one entry adrift. Five had moved by the
time the repair was taken, and three of them change a state word a page prints:
metadata-sync, server-pairing and watchlist have each published a first
finished release. #233 carries the readings.

What this does not settle

The record ages again the moment a repository publishes. The comparison reports
weekly and writes nothing by design, so what keeps data/releases.json current
is a person re-taking it, and nothing here changes that.