Bug 2092215
| Summary: | [NMCI] dracut_NM_vlan_over_team_no_boot test fails | |||
|---|---|---|---|---|
| Product: | Red Hat Enterprise Linux 9 | Reporter: | Thomas Haller <thaller> | |
| Component: | NetworkManager | Assignee: | Beniamino Galvani <bgalvani> | |
| Status: | CLOSED CURRENTRELEASE | QA Contact: | Filip Pokryvka <fpokryvk> | |
| Severity: | high | Docs Contact: | ||
| Priority: | unspecified | |||
| Version: | 9.0 | CC: | bgalvani, ferferna, fpokryvk, lrintel, pvalena, rkhan, sfaye, sukulkar, till, vbenes, wenliang | |
| Target Milestone: | rc | Keywords: | Triaged | |
| Target Release: | --- | Flags: | pm-rhel:
mirror+
|
|
| Hardware: | Unspecified | |||
| OS: | Unspecified | |||
| Whiteboard: | ||||
| Fixed In Version: | NetworkManager-1.41.4-1.el9 | Doc Type: | No Doc Update | |
| Doc Text: | Story Points: | --- | ||
| Clone Of: | ||||
| : | 2166257 (view as bug list) | Environment: | ||
| Last Closed: | 2023-03-06 16:09:05 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: | ||||
| Bug Depends On: | 2066816, 2166257 | |||
| Bug Blocks: | ||||
|
Description
Thomas Haller
2022-06-01 06:13:05 UTC
On RHEL9 this dracut commit is missing https://github.com/dracutdevs/dracut/commit/a97d2cedcf65a9a2fbff2591171f0163c7d3cb46 , needed to properly support teaming in initrd. Hello, please test, if you can, whether the planned rebase for 9.1 works for you. RPMS: https://github.com/pvalena/rpms/tree/main/dracut/2066816 (In reply to Pavel Valena from comment #2) > Hello, please test, if you can, whether the planned rebase for 9.1 works for > you. > > RPMS: https://github.com/pvalena/rpms/tree/main/dracut/2066816 It's far better but we still are still not 100% stable. Switching back to NetworkManager. When can we expect the 57 version to be in compose? (In reply to Vladimir Benes from comment #3) > (In reply to Pavel Valena from comment #2) > > Hello, please test, if you can, whether the planned rebase for 9.1 works for > > you. > > > > RPMS: https://github.com/pvalena/rpms/tree/main/dracut/2066816 > > It's far better but we still are still not 100% stable. Hi Vladimir, do you have a report for a failure using the new dracut? It seems I can't reproduce the problem. There is still fix in teamd needed, waiting for upstream merge. Moving ITM. The test is re-enabled (together with new team tests, which seems to alive now due to libteam fixes), if no fail occurs, we can verify. The problem happened again, setting FailedQA, moving back to ASSIGNED. It happened only once so far, but we need to be stable. https://desktopqe-jenkins.rhev-ci-vms.eng.rdu2.redhat.com/job/beaker-NetworkManager-main-veth-rhel9-upstream/1954/artifact/artifacts/FAIL_report_NetworkManager-ci_Test0631_dracut_NM_vlan_over_team_no_boot.html (jenkins build kept forever). Comments from Beniamino: > the problem is that after switch-root NM tries to assume the previous > connection, but since it can't connect to teamd it doesn't re-use the > existing connection; instead it creates a new one: > > Jan 11 14:23:38 localhost.localdomain NetworkManager[787]: <debug> [1673465018.6242] device[bbd8a218654cdaa5] (eth0): assume-state: set guess-assume=0, connection="171b737e-afd5-45b7-b929-7c76183a35d6" > [...] > Jan 11 14:23:38 localhost.localdomain NetworkManager[787]: <debug> [1673465018.6278] device[3c034abd74aecce8] (team0): failure to connect to teamdctl with cli_type=dbus, err=-22 > Jan 11 14:23:38 localhost.localdomain NetworkManager[787]: <debug> [1673465018.6280] device[3c034abd74aecce8] (team0): failure to connect to teamdctl, err=-22 > Jan 11 14:23:38 localhost.localdomain NetworkManager[787]: <debug> [1673465018.6280] device[bbd8a218654cdaa5] (eth0): assume-state: set guess-assume=0, connection=(null) > > I don't know the exact reason, this needs to be investigated. Maybe original problem is fixed and this is consequent, but result is the same, test is unstable. Assumed connections should not be there and connection to teamdctl should be stable. In 1.34 and 1.38 test sometimes fails with VLAN connection not activated in 45 seconds, which might be also related to unstable teamdctl. Hi Filip, looking at the failed executions for this test on RHEL 9.1, I see that most are crashes. The libteam version there is 1.31-14.el9. The crash is solved by the RHEL-9.1.0.1 z-stream update (libteam-1.31-16.el9_1). The z-stream package also contains a fix to improve the transition between the initrd and real root, and I believe it will help also with the other failures. Can you please try the new libteam package? It was RHEL 9.1 updates, with fixed libteam. But so far, issue happened only once (excluding crashes of older libteam, which for some reason reappears in newer compose?) @fpokryvk I have linked two MRs, one for NM and one for dracut. With those, I got >100 runs of the test without failures, while normally they would fail after <10. Please check them, thanks! Great news! I will check patched versions. Does it mean, we will need clone of this for dracut, when upstream MR is merged? > Does it mean, we will need clone of this for dracut, when upstream MR is merged?
Yes, that's correct.
I have checked all dracut tests and it is stable again. Thanks a lot! I will make the clone of this then, and add "blocked by" flag to this bug. This test is now passing and stable. |