URL:
https://github.com/SSSD/sssd/pull/5766
Title: #5766: Sources cleanup - part 3.
alexey-tikhonov commented:
"""
I didn't see this debian failure on other pull requests so I
would like to wait until we understand what is happening there.
I still don't know what is happening exactly, but I discovered that the same fail is
observed on master branch if SSSD is configured `--without-secrets`.
With patch
https://github.com/alexey-tikhonov/sssd/commit/1018378bbeae2e6869bdb49e1a... on
the top of master branch on demand run result:
https://ci-jenkins-csb-sssd.apps.ocp4.prod.psi.redhat.com/job/sssd-ci-on-...
So:
- master branch with default config - **success**
- master branch `--without-secrets`**entire** suite - **fail** on Debian (both 10 and 11,
but logs are a bit different)
- master branch `--without-secrets` "-k test_krb5_auth" - **success**
This makes me thinking that some of tests (maybe kcm) leave a leftovers in the working dir
that trigger a fail, but `secrets` tests clean this up, so `test_krb5_auth` succeeds. And
when we disable `secrets` (either via config option or via code removal) clean up
doesn't happen and `test_krb5_auth` fails.
This is the only idea I have at the moment and I must admit it doesn't explain why
fail is only seen on Debian... IIRC, Debian was somewhat different with regards to KCM,
but I don't remember details.
As to what exactly is failing: `krb5_child` fails to get a ticket - KDC replies 'user
not found` - probably `kadmin.local addprinc` fails.
IIUC, test is expected to work with 'FILE:` ccaches, but on Debian11 I saw something
like 'Failed to connect to KCM server' on stderr (I'm not sure but looks like
this is from `kadmin.local`).
@sumit-bose , @pbrezina , do you have any insights based on the above summary?
"""
See the full comment at
https://github.com/SSSD/sssd/pull/5766#issuecomment-927107834