Forgejo Actions job logs purge within hours - post-hoc CI diagnosis is impossible #545

New issue

Open

opened 2026-07-12 06:20:25 +00:00 by coilysiren · 0 comments

coilysiren commented

2026-07-12 06:20:25 +00:00

Owner

Check the Forgejo deployment config for the Actions log retention / cleanup cron (e.g. [actions] LOG_RETENTION_DAYS or the cleanup cron aggressiveness) and raise it to at least a week, or ship failing-run logs somewhere durable before purge (GlitchTip shipper pattern).

During the 2026-07-11 stability session, job logs for aos release runs from THE SAME DAY (runs 1046-1055, failed ~03:00-23:15 UTC) were already purged: both the API logs route and the UI log-cursor POST return "task ... resource does not exist". The chronic publish-dev-base failures of 7/9-7/11 are now undiagnosable post-hoc; the only surviving signal is step-level status/duration. Check the Forgejo deployment config for the Actions log retention / cleanup cron (e.g. [actions] LOG_RETENTION_DAYS or the cleanup cron aggressiveness) and raise it to at least a week, or ship failing-run logs somewhere durable before purge (GlitchTip shipper pattern).