Bug 620780 - libvirtd blocks on crashed qemu guests
Summary: libvirtd blocks on crashed qemu guests
Alias: None
Product: Virtualization Tools
Classification: Community
Component: libvirt   
(Show other bugs)
Version: unspecified
Hardware: All Linux
Target Milestone: ---
Assignee: Libvirt Maintainers
QA Contact:
Depends On:
TreeView+ depends on / blocked
Reported: 2010-08-03 13:28 UTC by Sebastian Hetze
Modified: 2015-03-18 09:06 UTC (History)
4 users (show)

Fixed In Version:
Doc Type: Bug Fix
Doc Text:
Story Points: ---
Clone Of:
Last Closed: 2015-03-18 09:06:52 UTC
Type: ---
Regression: ---
Mount Type: ---
Documentation: ---
Verified Versions:
Category: ---
oVirt Team: ---
RHEL 7.3 requirements from Atomic Host:
Cloudforms Team: ---

Attachments (Terms of Use)

Description Sebastian Hetze 2010-08-03 13:28:29 UTC
Description of problem:

Version-Release number of selected component (if applicable):

How reproducible:

Steps to Reproduce:
Actual results:

Expected results:

Additional info:

Comment 1 Sebastian Hetze 2010-08-03 13:37:55 UTC
hit return by accident, so the empty form fired off by accident.
Here we go again:

Description of problem:

we occasionally get libvirtd into a blocked state after the
quemu guest has run into some crash condition.
The quemu process is still running and uses 100% CPU while
the guest is not reachable in any way.

Any attempt to use libvirt is blocked in this situation.
virsh list, virsh stop guest or anything alike does not
give any messages nor does it return.

Version-Release number of selected component (if applicable):

utuntu host and guest
libvirt 0.8.2 and 0.7.2

How reproducible:

Unfortunately, this behaviour is not reproducable at will.
However, playing arround I found that stopping a guest
with SIGSTOP and calling virsh list brings at least one
thread into a blocked state:

kill -19 `pidof -s qemu-system-x86_64`; virsh list

If we are lucky and this is related we might get our
real problem fixed if we get the latter solved.

Note You need to log in before you can comment on or make changes to this bug.