This hotfix restores periodic updates for LDXP sources that were left in a permanent blocked state after the v3.7.1 crawler rollback.
- Retry
blockedsources after a 6-hour backoff andchallenge_requiredsources after a 24-hour backoff. - Keep
--retry-blockedas an explicit operator override that ignores the remaining backoff window. - Preserve the existing three-source circuit breaker so a site-wide challenge stops a scheduled batch quickly.
- Verify that blocked scans count as failures and never satisfy the successful-scan publication gate.
- Run the crawler self-test and pytest suite in CI.
- No database migration is required.
- Follow
docs/QUICK_DEPLOY.mdand rebuild the Crawler image from this Tag. - Because this release changes crawler behavior, complete one successful multi-source refresh before restoring all production timers.
- Existing blocked rows with no retry timestamp become immediately eligible in bounded batches; successful scans reset their failure state normally.