You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
While investigating #2407 (Claude e2e failure analysis migration), found that oadp-1.4's test-e2e Makefile target is missing the --ginkgo.junit-report flag that oadp-1.5 and oadp-dev have — so it never writes junit_report.xml to $ARTIFACT_DIR.
This is unrelated to the Claude CLI hook removed in #2407 (oadp-1.4 never had that hook); it's simply an older Makefile that predates the flag being added on other branches.
Summary
While investigating #2407 (Claude e2e failure analysis migration), found that
oadp-1.4'stest-e2eMakefile target is missing the--ginkgo.junit-reportflag thatoadp-1.5andoadp-devhave — so it never writesjunit_report.xmlto$ARTIFACT_DIR.This is unrelated to the Claude CLI hook removed in #2407 (oadp-1.4 never had that hook); it's simply an older
Makefilethat predates the flag being added on other branches.Impact
oadp-analyze-e2e-failurestep-registry post-step (from oadp-operator: migrate e2e Claude failure analysis to claude-ai-helpers post-step release#83957) still runs on any branch it's wired to and can read must-gather + per-test logs, but onoadp-1.4it loses JUnit-based pass/fail counts as an analysis input — reducing precision, not blocking analysis.junit_report.xmlin Prow artifacts foroadp-1.4jobs is similarly affected.Fix
Add the same
--ginkgo.junit-report="$(ARTIFACT_DIR)/junit_report.xml"flag tooadp-1.4'stest-e2etarget in theMakefile, matchingoadp-1.5/oadp-dev.Note
Responses generated with Claude