Control: tag -1 + moreinfo Control: severity: -1 normal
Replying to an old bug report which received no attention at the time.. On Mon, 05 Aug 2019 00:38:10 -0400 Michael Stroucken <[email protected]> wrote:
Package: qemu-system-x86 Version: 1:2.8+dfsg-6+deb9u7 Severity: important I have VM servers that are running Jessie. On testing a migration of one to stretch, I noticed that one of the VMs was getting stuck on IO. I have noticed that this happens when guest is performing IO, and the system issues a "stop" command via the monitor followed by a "cont" command. The system does this to allow for guest filesystem snapshotting. Only the virtual disk that was being accessed will remain stuck. The kernel of the system is linux-image-4.9.0-9-amd64 4.9.168-1+deb9u4 Linux HOST 4.9.0-9-amd64 #1 SMP Debian 4.9.168-1+deb9u4 (2019-07-19) x86_64 GNU/Linux The problem does not occur with the jessie version of qemu running on the stretch kernel (qemu-system-x86 1:2.1+dfsg-12+deb8u11). Conversely, the problem does occur with the stretch version of qemu running on the jessie kernel (linux-image-3.16.0-10-amd64 3.16.70-1).
[...] This was an excellent bug report at the time, with a lot of details. Unfortunately it got no attention whatsoever. I remember some I/O being stuck sometimes after stop/cont or after suspend/resume of the guest. It's been really a while since I've seen it for the last time. It looks like this particular bug has been fixed long time ago. Can you please check current version of qemu (bookworm or trixie) to see if the problem is still there? If not, let's close this bug with bookworm version of qemu (sort of arbitrary). Thanks, /mjt
