-
Bug
-
Resolution: Done-Errata
-
Normal
-
4.14
-
None
-
No
-
2
-
Sprint 241, Sprint 242, Sprint 243
-
3
-
Rejected
-
False
-
-
N/A
-
Release Note Not Required
Description of problem:
When the TestMTLSWithCRLs e2e test fails on a curl, it checks the stdout but the stdout could be empty, so it panics: --- FAIL: TestAll/parallel/TestMTLSWithCRLs (97.09s) --- FAIL: TestAll/parallel/TestMTLSWithCRLs/certificate-distributes-its-own-crl (97.09s) panic: runtime error: slice bounds out of range [-3:] [recovered] panic: runtime error: slice bounds out of range [-3:]
Version-Release number of selected component (if applicable):
4.14
How reproducible:
Experience a failure on the MTLS testing, such as seen in https://redhat-internal.slack.com/archives/CBWMXQJKD/p1688596054069399?thread_ts=1688596036.042119&cid=CBWMXQJKD Search.ci shows two failures in the past two weeks: https://search.ci.openshift.org/?search=FAIL%3A+TestAll%2Fparallel%2FTestMTLSWithCRLs&maxAge=336h&context=1&type=bug%2Bissue%2Bjunit&name=cluster-ingress-operator&excludeName=&maxMatches=5&maxBytes=20971520&groupBy=job
Steps to Reproduce:
1. N/A 2. 3.
Actual results:
Test panics when trying to report an error.
Expected results:
Test reports whatever error it can without panics.
Additional info:
stdout was empty, but https://github.com/openshift/cluster-ingress-operator/blob/4c92a6d1ee80b6b120dd750855a40145a530153c/test/e2e/client_tls_test.go#L1587 doesn't check that the value is empty before it tries to index it.