Hi,

we are seeing intermittent boot hangs in Fedora 44 during openQA testing.
The system appears to stall shortly after initrd-switch-root.service. On
the console we typically see:

Starting initrd-switch-root.service

However, the journal shows that the boot progressed somewhat further. The
log then simply ends without any obvious error. An example openQA job:

https://openqa.stg.fedoraproject.org/tests/5920121

In the affected boots, the last messages are around rpc_pipefs.target being
reached, after which nothing else is logged. openQA waits ~15 minutes and
the system never reaches login.

This happens most frequently on aarch64, but we have occasionally seen it
on x86_64 as well. It is not trivially reproducible: local repeated reboot
testing (10–15 consecutive boots) did not trigger it. openQA likely hits it
due to the volume of runs.

We can confirm occurrences at least as far back as early January. One
hypothesis was a possible interaction with systemd-259-1.fc44, but that is
not yet confirmed.

Full journals from both a failed and a successful boot are attached in the
Bugzilla:

https://bugzilla.redhat.com/show_bug.cgi?id=2439826

A recent observation suggests the hang may occur around
plymouth-read-write.service, as in bad boots we see it start but not
finish, while in good boots it completes successfully.

Interestingly, enabling systemd.log_level=debug has so far made the issue
disappear in testing, which may indicate a timing-related race.

If anyone has seen similar behavior, has ideas about debugging early
userspace hangs after switch-root, or suspects recent changes in systemd,
plymouth, or related early-boot components, any pointers would be
appreciated. Suggestions for additional instrumentation that would not
significantly perturb timing would also help.

Given Beta freeze timing, we are trying to understand root cause and
potential fix urgency.

Thanks.


-- 

Lukáš Růžička

FEDORA QE, RHCE

Red Hat

<https://www.redhat.com>

Purkyňova 115

612 45 Brno - Královo Pole

[email protected]
TRIED AND PERSONALLY TESTED, ERGO TRUSTED. <https://redhat.com/trusted>
-- 
_______________________________________________
devel mailing list -- [email protected]
To unsubscribe send an email to [email protected]
Fedora Code of Conduct: 
https://docs.fedoraproject.org/en-US/project/code-of-conduct/
List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines
List Archives: 
https://lists.fedoraproject.org/archives/list/[email protected]
Do not reply to spam, report it: 
https://forge.fedoraproject.org/infra/tickets/issues/new

Reply via email to