Under heavy load qemu hits bdrv_error_action: Assertion `error >= 0' failed
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
qemu (Ubuntu) |
Fix Released
|
Medium
|
Dave Chiluk | ||
Trusty |
Fix Released
|
Undecided
|
Unassigned |
Bug Description
[Impact]
* VM running in QEMU with this message in qemu log
qemu-system-x86_64: /build/
* This results in the VM needing to be restarted.
[Test Case]
* This only reproduces under extremely high I/O load and very infrequently.
Once per month on major clouds.
[Regression Potential]
* Commit 3bbf572345c is included in 2.4.0-rc4 and newer, and has not been removed in latest develepment. I have not found any commits referencing it as a regressor.
* Patch adds memory barriers so regressions may show up in the form of
performance issues.
[Other Info]
* Redhat identified this commit as the fix for the issue here
https:/
https:/
https:/
* Only apply to Trusty as Xenial and newer have fix already. Ignoring
Precise because no currently reported cases.
Changed in qemu (Ubuntu): | |
importance: | Undecided → Medium |
Changed in qemu (Ubuntu Trusty): | |
status: | Fix Committed → Fix Released |
Changed in qemu (Ubuntu): | |
status: | Fix Committed → Fix Released |
Here's the proposed debdiff. I'm currently waiting on confirmation from the customer of resolution, but that may never come, as it is so difficult to reproduce this issue.
Also need to double check if this applies to the cloud archives.