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
{{ message }}
Repository navigation
e2e: two capture-path comments name one consumer where two exist #3663
Two comments on the e2e capture path describe one consumer where two exist, both pre-existing:
hack/e2e-chainsaw/.chainsaw.yaml (the comment near line 31) says cozytest.sh "only covers the BATS install/openapi tests", while packages/core/testing/Makefile also runs e2e-prepare-cluster.bats through it.
hack/e2e-capture-dataplane.sh (header, line 6) names only cozytest.sh's on-failure hook as its invoker, while the Chainsaw global catch invokes it too; the same file partially corrects itself around lines 299-303.
Why it matters
Both sentences steer a reader sizing budgets or changing behavior toward accounting for one caller out of two. The recent bounding work measured exactly this hazard: the two call sites wrap the same script in different envelopes (600s vs 300s), so a change validated against the caller the comment names can still overrun the caller it omits.
Fix shape
Two one-line comment edits naming both consumers. No behavior change.
Scope update: the .chainsaw.yaml half is being fixed in PR #3662 after all, together with a THIRD site of the same class that the review round found one grep away: hack/e2e-chainsaw/README.md line 5 also lists two BATS suites where three exist. With those two, the tree stops contradicting itself about which suites run through cozytest.sh (docs/agents/e2e-testing.md was always correct). What remains for this issue is the second bullet only: the hack/e2e-capture-dataplane.sh header naming one invoker where two exist.
Both parts are on main. #3662 fixed the .chainsaw.yaml comment, and after #3672 the hack/e2e-capture-dataplane.sh header names both callers with their backstops.
Activity
lexfrei commentedon Aug 8, 2026
Scope update: the
.chainsaw.yamlhalf is being fixed in PR #3662 after all, together with a THIRD site of the same class that the review round found one grep away:hack/e2e-chainsaw/README.mdline 5 also lists two BATS suites where three exist. With those two, the tree stops contradicting itself about which suites run through cozytest.sh (docs/agents/e2e-testing.mdwas always correct). What remains for this issue is the second bullet only: thehack/e2e-capture-dataplane.shheader naming one invoker where two exist.fix(e2e): stop the data-plane capture reporting what it never observed (
lexfrei commentedon Sep 23, 2026
Both parts are on main. #3662 fixed the
.chainsaw.yamlcomment, and after #3672 thehack/e2e-capture-dataplane.shheader names both callers with their backstops.