Comment # 22 on bug 1220119 from Petr Vorel

(In reply to Jiri Wiesner from comment #21)
> Petr, I need to ask:
> Is the log in comment 13 a composite, meaning collated from logs coming from
> different boots of the VM?

No, single boot.

FYI These serial0.txt files are created as QEMU output from SUT, I believe by
these QEMU parameters: "-chardev
ringbuf,id=serial0,logfile=serial0,logappend=on -serial chardev:serial0" (from
[1], this file will not last, but any new systemd-networkd@64bit job from
Tumbleweed group [2] should have it in the future).
I mean, QEMU machine boots writing dmesg to this file, tests runs and machine
shutdown.

And we have serial0.txt when we test on bare metal machines, again, it's always
from single boot.

> As for the association between the test failure and the occurrences of the
> clocksource watchdog error, does the clocksource watchdog error (or
> "clocksource: Long readout interval...") always happen when the
> ltp_net_tcp_cmds test fails?

FYI this log is from different test (pure openQA test, I don't maintain), from
job  opensuse-Tumbleweed-DVD-x86_64-Build20240216-systemd-networkd@64bit
running on https://openqa.opensuse.org/tests/3945256 (e.g. on o3 worker).

The one I was able to reproduce on ltp_net_tcp_cmds (e.g. job using LTP for
testing) was also on Tumbleweed, but on my private worker.

> It seems to me that, since causation has not been established, there could

It's also unclear for me why this happens randomly. Maybe we could find more
from dmesg on actual KVM host. openQA KVM hosts are quite often overloaded,
maybe that influence clock clocksource on SUT.

> be a hidden root cause causing both the test to fail as well as the
> (possibly occasional) clocksource errors. Regarding kernel options, it is
> also possible to use tsc=reliable, which is meant for virtualized
> environments, or tsc=unstable to disable the watchdog checks.

Do you think we should start using tsc kernel param? BTW these machines are
configured: -m 1536 -cpu host (sometimes have more CPU or RAM for particular
tests), e.g. if it happens to us, it can happen to anybody using VMs on cloud,
right?

[1] https://openqa.opensuse.org/tests/3945256/file/autoinst-log.txt
[2] https://openqa.opensuse.org/group_overview/1