-
Notifications
You must be signed in to change notification settings - Fork 1.8k
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
blocked for more than 120 seconds on 0.6.5.2 in KVM VM #4065
Comments
@poiuty please update to 0.6.5.3
or disable dynamic taskq's manually, like you already mentioned in #3955 (comment) |
@kernelOfTruth - dont help for me (already set 0)
And I cant update to 0.6.5.3 |
@poiuty I suspect that this could be a bug in QEMU's virtio code / the guest's virtio code unless there is a hang on the host. 0.6.5.y changed how zvol processing is done to return faster by calling directly into the DMU instead of passing the work to a worker thread, which reduces latencies. It is conceivable that the virtio code cannot reliably handle the faster return. Another possibility is that there is a bug in Linux 3.16 that causes it to sometimes forget to acknowledge IOs when the attempt to send them comes back with completion immediately rather than later. I will try to look into this some more time permitting. In the meantime, I suggest switching QEMU from virtio to AHCI. If my suspicion is correct, the problem is in the virtio code and will go away by doing this. AHCI is fairly efficient, so I would not expect you to lose much performance by switching to it. There is an example of how to do this with the raw QEMU command here: https://wiki.gentoo.org/wiki/QEMU/Options#Hard_drive You likely can translate that into libvirt's XML format, although I am not familiar enough to provide an example. |
@ryao, thx, move test VM to AHCI
|
Debian 8, 3.16.0-4-amd64, qemu-kvm, zfs v0.6.5.2-2, VM in zvol with lz4
zfs => zvol kvm vm => lvm => ext4
Today VM freeze again. Get screenshot from vnc. Check logs (node+vm) - no call trace.

Before (about ~3 months) I use zfs 0.6.4 and dont have this problem.
So. I think problem mb in zfs | or in virtio | or in kernel (vm)
Now I try update vm kernel (3.2 to 3.16)
Any ideas?
also #3955
The text was updated successfully, but these errors were encountered: