Skip to content

Pipeline resilience: prevent artifact loss and cascading failures#653

Merged
MaxGhenis merged 2 commits into
mainfrom
fix/pipeline-resilience
Mar 29, 2026
Merged

Pipeline resilience: prevent artifact loss and cascading failures#653
MaxGhenis merged 2 commits into
mainfrom
fix/pipeline-resilience

Conversation

@baogorek
Copy link
Copy Markdown
Collaborator

Summary

  • Never rmtree version directories containing H5 files on fingerprint mismatch — update fingerprint and resume instead
  • Pin fingerprint in RunMetadata so resumed runs aren't invalidated by branch drift
  • Add validation pre-flight to catch DB schema mismatches before spawning workers
  • Harden error dicts with stderr fallback/truncation and traceback capture
  • Add CI tests for query-schema compatibility (test_validation_queries.py)

Closes #652

Test plan

  • pytest policyengine_us_data/tests/test_validation_queries.py — 4/4 pass
  • Existing test_schema_views_and_lookups.py — 17/17 pass
  • Manual: verify fingerprint mismatch with existing H5s preserves files
  • Manual: --build-only then resume with different branch confirms H5 preservation

🤖 Generated with Claude Code

baogorek and others added 2 commits March 27, 2026 16:00
Never rmtree version directories containing H5 files on fingerprint
mismatch — update fingerprint and resume instead. Pin fingerprint in
RunMetadata so resumed runs aren't invalidated by branch drift. Add
validation pre-flight to catch schema mismatches before spawning
workers, and harden error dicts with stderr fallback/truncation and
traceback capture. Add CI tests for query-schema compatibility.

Closes #652

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
@MaxGhenis MaxGhenis merged commit 22f922e into main Mar 29, 2026
7 checks passed
@MaxGhenis MaxGhenis deleted the fix/pipeline-resilience branch March 29, 2026 22:31
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

Pipeline resilience: prevent artifact loss on fingerprint mismatch and cascading failures

2 participants