Bug 1815877
Summary: | authselect fails to select sssd profile in OpenQA tests | ||
---|---|---|---|
Product: | [Fedora] Fedora | Reporter: | Alexander Bokovoy <abokovoy> |
Component: | authselect | Assignee: | Pavel Březina <pbrezina> |
Status: | CLOSED CURRENTRELEASE | QA Contact: | Fedora Extras Quality Assurance <extras-qa> |
Severity: | unspecified | Docs Contact: | |
Priority: | unspecified | ||
Version: | 32 | CC: | awilliam, cheimes, jhrozek, pbrezina |
Target Milestone: | --- | ||
Target Release: | --- | ||
Hardware: | Unspecified | ||
OS: | Unspecified | ||
Whiteboard: | |||
Fixed In Version: | Doc Type: | If docs needed, set a value | |
Doc Text: | Story Points: | --- | |
Clone Of: | Environment: | ||
Last Closed: | 2020-04-22 15:24:49 UTC | Type: | Bug |
Regression: | --- | Mount Type: | --- |
Documentation: | --- | CRM: | |
Verified Versions: | Category: | --- | |
oVirt Team: | --- | RHEL 7.3 requirements from Atomic Host: | |
Cloudforms Team: | --- | Target Upstream Version: | |
Embargoed: |
Description
Alexander Bokovoy
2020-03-22 11:22:05 UTC
The test is triggered by a new FreeIPA update (https://bodhi.fedoraproject.org/updates/FEDORA-2020-e3a79248dc) which introduces SELinux policy in freeipa-selinux package. However, I was not able to find any authselect-related SELinux policy definition anywhere. The test is only failing on that update, not updates before or after it. So it seems like introducing the FreeIPA selinux policy really triggers this somehow. I'd guess it's something like, with the policy in place, something that used to be unconfined is now confined, and that results in these denials? The first error message "Unable to create selabel context" is coming from authselect [1]. The function call selabel_open(SELABEL_CTX_FILE, NULL, 0) [2] is failing. This could also be a problem with SELinux userspace library. [1] https://github.com/authselect/authselect/blob/478ec8c356d6f0162f8a954b426b1eaeee29f3e0/src/lib/util/selinux.c#L35-L48 [2] https://linux.die.net/man/3/selabel_open I cannot reproduce the problem either. realm join works fine for me on a recently updated F32 machine. Authselect does not define any selinux policy, it is just trying to set the right default context for newly created files - they are first written as temporary files and then moved to their correct location so authselect needs to make sure they are created with right context. Is there any way for FreeIPA selinux policy to interfere with selabel_open()? Christian: did you ensure the packages from FEDORA-2020-e3a79248dc were used in your test? That is where the problem exists. Are there any news on thig bug? Is it still happening? from the update comments it looks like this was fixed somehow in freeipa 4.8.6. FreeIPA tests are passing in F32 and Rawhide ATM so I think we can close this. |