Found by the #192 research sweep, in released code.
Two separate defects, both in the scheduled drift workflow, both distinct from #192's own crash:
- The
docs-assets leg fails identically to the price-drift leg — no AWS region is set in the scheduled run's environment, so importing the product's settings calls sys.exit(1) at collection time.
- Its tracking issue is never filed. The auto-file step posts the whole run log as the issue body and GitHub rejects it:
GraphQL: Body is too long (maximum is 65536 characters). So the leg fails silently — no issue, no signal, and the only evidence is in the workflow run itself.
The second is the more serious of the two: a failure-reporting path that swallows its own failure means every future drift finding on this leg is lost, not just this one. The fix needs to bound what goes in the body (tail the log, attach the rest as a run link) rather than assume it fits.
Related: #192 (the price-drift leg's crash, same root cause for part 1).
Found by the #192 research sweep, in released code.
Two separate defects, both in the scheduled drift workflow, both distinct from #192's own crash:
docs-assetsleg fails identically to the price-drift leg — no AWS region is set in the scheduled run's environment, so importing the product's settings callssys.exit(1)at collection time.GraphQL: Body is too long (maximum is 65536 characters). So the leg fails silently — no issue, no signal, and the only evidence is in the workflow run itself.The second is the more serious of the two: a failure-reporting path that swallows its own failure means every future drift finding on this leg is lost, not just this one. The fix needs to bound what goes in the body (tail the log, attach the rest as a run link) rather than assume it fits.
Related: #192 (the price-drift leg's crash, same root cause for part 1).