Issue #3689
closedPulp 2 tests are failing on Fedora 27: BIOError: padding check failed
Description
======================================================================
ERROR: test_get_certs_from_string_empty (test_repo_cert_utils.TestCertVerify)
----------------------------------------------------------------------
Traceback (most recent call last):
File "/home/jenkins/workspace/unittest-pulp-pr/node-type/f27-np/pulp/repoauth/test/test_repo_cert_utils.py", line 392, in test_get_certs_from_string_empty
certs = self.utils.get_certs_from_string("")
File "/home/jenkins/workspace/unittest-pulp-pr/node-type/f27-np/pulp/repoauth/pulp/repoauth/repo_cert_utils.py", line 358, in get_certs_from_string
bio = BIO.MemoryBuffer(data)
File "/usr/lib64/python2.7/site-packages/M2Crypto/BIO.py", line 191, in __init__
m2.bio_write(self.bio, data)
BIOError: padding check failed
See this PR: https://github.com/pulp/pulp/pull/3499
Looks like there was a new release on Fedora of m2crypto back in March.
Updated by dalley over 6 years ago
- Priority changed from Normal to High
- Triaged changed from No to Yes
Updated by daviddavis over 6 years ago
Gotten a little bit closer to finding the cause. It appears to be rooted in the test_not_validated test in unit.server.agent.test_auth.TestAuthentication. If you skip this test or run it after test_repo_cert_utils, the problem goes away.
Updated by daviddavis over 6 years ago
I think this is a problem in m2crypto so I've opened an issue with a failing test:
Updated by daviddavis over 6 years ago
Opened up PR to skip failing unit test until m2crypto issue is resolved:
Added by daviddavis over 6 years ago
Updated by daviddavis over 6 years ago
Matej responded to the m2crypto issue but I am not sure what his response means for us. I'm also not clear on what our tests are actually testing.
Updated by dalley over 6 years ago
- Status changed from NEW to ASSIGNED
- Assignee set to dalley
Updated by dalley over 6 years ago
- Priority changed from High to Normal
- Sprint deleted (
Sprint 38)
Looks like Jeremy Cline also ran into this issue 2 years ago: https://gitlab.com/m2crypto/m2crypto/issues/95
Going by that explanation of the issue, I think we should just leave the test disabled until upstream fixes their error handling. The issue is simply that M2Crypto is returning an error generated (correctly) by a previous test, and not clearing out its error stack properly.
Removing from the sprint, moving to normal Prio, and leaving assigned as I will track what is going on upstream with this issue.
Updated by bmbouter almost 6 years ago
- Status changed from ASSIGNED to CLOSED - WONTFIX
Temporarily skip failing test until m2crypto issue is resolved
ref #3689 https://pulp.plan.io/issues/3689