New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Qemu errors when virtiofsd is killed (race condition) #6757
Comments
Show kata-collect-data.sh details
Meta detailsRunning Runtime
Runtime is
|
nydusd is using
Yet another option could be to just SIGKILL the QEMU process instead instead of doing QMP quit. |
There is a race condition when virtiofsd is killed without finishing all the clients. Because of that, when a pod is stopped, QEMU detects virtiofsd is gone, which is legitimate. Sending a SIGTERM first before killing could introduce some latency during the shutdown. Fixes kata-containers#6757. Signed-off-by: Beraldo Leal <bleal@redhat.com>
There is a race condition when virtiofsd is killed without finishing all the clients. Because of that, when a pod is stopped, QEMU detects virtiofsd is gone, which is legitimate. Sending a SIGTERM first before killing could introduce some latency during the shutdown. Fixes kata-containers#6757. Signed-off-by: Beraldo Leal <bleal@redhat.com>
There is a race condition when virtiofsd is killed without finishing all the clients. Because of that, when a pod is stopped, QEMU detects virtiofsd is gone, which is legitimate. Sending a SIGTERM first before killing could introduce some latency during the shutdown. Fixes kata-containers#6757. Signed-off-by: Beraldo Leal <bleal@redhat.com>
There is a race condition when virtiofsd is killed without finishing all the clients. Because of that, when a pod is stopped, QEMU detects virtiofsd is gone, which is legitimate. Sending a SIGTERM first before killing could introduce some latency during the shutdown. Fixes kata-containers#6757. Backport of kata-containers#6959. Signed-off-by: Beraldo Leal <bleal@redhat.com>
There is a race condition when virtiofsd is killed without finishing all the clients. Because of that, when a pod is stopped, QEMU detects virtiofsd is gone, which is legitimate. Sending a SIGTERM first before killing could introduce some latency during the shutdown. Fixes kata-containers#6757. Backport of kata-containers#6959. Signed-off-by: Beraldo Leal <bleal@redhat.com> (cherry picked from commit 0e47cfc)
Description of problem
Occasionally, when halting a pod, systemctl displays multiple error messages originating from QEMU. Although these errors are legitimate, they clutter the log files and may create a negative impression for users.
Expected result
Upon stopping a pod, the systemctl should display only relevant and essential messages without cluttering the log files. Ideally, the QEMU-related error messages should be filtered, minimized, or we should investigate the root cause (maybe give more time to virtiofsd stop). This will result in a cleaner and more user-friendly log presentation, thereby enhancing the user experience.
Actual result
Further information
From @gkurz:
An option could be to send a SIGTERM instead of a SIGKILL to virtiofsd, the same way nydusd is doing.
Another option could be to simply turn QEMU logging off before terminating virtiofsd.
The text was updated successfully, but these errors were encountered: