Closed resulin closed 2 years ago
This only started happening recently (~1 week ago).
The unresponsiveness is likely in part due to #4719. Do you have anything interesting in /var/log/libvirt/libxl/sys-whonix.log
?
@DemiMarie
I can't tell from looking at it but here's my entire sys-whonix.log. https://privatebin.net/?6b82b3a92c3cb250#6gAZvyZ3TbeFH7gkcHSMojPAHu2HvaGe7JW4L5pbVdWX
Reports in the forums suggest that it's related to https://github.com/QubesOS/qubes-issues/issues/7404 : https://forum.qubes-os.org/t/suspend-swap-and-whonix-gateway-performance-issues/11310/3
I will test later today whether applying the patch in #7404 solves the issue.
I don't know if it's related but I also noticed that systemd-socket-proxyd
jumps to 100% CPU usage after a while. I have to keep killing it each time. Should I open a different bug report for this one?
systemd-socket-proxyd
spinning at 100% CPU is a systemd bug. Please report it to systemd if it is one of the two most recent versions, or to your distribution maintainer otherwise. Only report it here if it is in dom0.
Would you mind adding:
log_filters="3:*"
log_level="1"
log_outputs="1:syslog:libvirtd 1:journald"
to /etc/libvirtd/libvirtd.conf
and then restarting libvirtd with systemctl restart libvirtd
? That will cause libvirt to log debug messages from libxl.
@DemiMarie
After adding these lines to /etc/libvirt/libvirtd.conf
:
log_filters="3:*"
log_level="1"
log_outputs="1:syslog:libvirtd 1:journald"
And then running systemctl restart libvirtd
I run into the following error:
Job for libvirtd.service failed because the control process exited with error code.
See "systemctl status libvirtd.service" and "journalctl -xe" for details.
dom0 libvirtd[207503]: libvirt version: 6.6.0, package: 5.fc32 (Unknown, 2021-10-02-04:34:58, )
dom0 libvirtd[207503]: hostname: dom0
dom0 libvirtd[207503]: internal error: unexpected domain Domain-0 already exists
dom0 libvirtd[207503]: Failed to load config for domain 'Domain-0'
@DemiMarie
After adding these lines to
/etc/libvirt/libvirtd.conf
:log_filters="3:*" log_level="1" log_outputs="1:syslog:libvirtd 1:journald"
And then running
systemctl restart libvirtd
I run into the following error:Job for libvirtd.service failed because the control process exited with error code. See "systemctl status libvirtd.service" and "journalctl -xe" for details.
Can you post the full log output of libvirtd?
dom0 libvirtd[207503]: libvirt version: 6.6.0, package: 5.fc32 (Unknown, 2021-10-02-04:34:58, ) dom0 libvirtd[207503]: hostname: dom0 dom0 libvirtd[207503]: internal error: unexpected domain Domain-0 already exists dom0 libvirtd[207503]: Failed to load config for domain 'Domain-0'
This is a libvirtd bug, but should not be causing libvirtd start to fail.
Manually applying the patch from https://github.com/QubesOS/qubes-core-admin/pull/473 and then restarting seems to have solved the issue, suspend works without problem now.
@DemiMarie feel free to re-open this issue if you want to investigate that specific libvirtd problem if you think it's worth it, I can definitely go back and undo the patch above to provide the full logs for libvirtd if needed.
Manually applying this patch seems to fix this issue for me as well.
How to file a helpful issue
Qubes OS release
4.1
Brief summary
Steps to reproduce
Shutting down a whonix based VM (either whonix-ws or whonix-gw).
Expected behavior
No error.
Actual behavior
Error when shutting down whonix templates, also if done using the Qube Manager or the Qube Widget they both lag and become unresponsive.
journalctl log: