Hi Eric,
> I just noticed this message. Is this still a problem in 0.9.12? There
> have been several patches in the meantime. Unfortunately, the patch as
> is will not work:
Sorry for the late reply: I just tried and it is still a problem in
0.9.12. I hit this bug every 5 minutes in my setup s
This fixes a race condition when VIR_EVENT_HANDLE_HANGUP is triggered
during lxcDomainDestroyFlags: lxcMonitorEvent tries to acquire the
driver lock held by lxcDomainDestroyFlags and blocks. Meanwhile
lxcDomainDestroyFlags will free the "vm" structure and release the driver
lock. lxcMonitorEvent un
Hi,
last email hopefully. My last patch was broken because it double
locked vm. I discovered that virDomainFindBy* locks vm before
returning it. The new patch does not double-lock.
best,
Tom
From f2fd4cb72c8f2e01567320a643fe2d665308119e Mon Sep 17 00:00:00 2001
From: Thomas Hunger
Date: Thu, 15
Hey again,
attached is a patch against master that fixes the race condition for
me. Not sure if this is the best solution. My system is quite stable
now under load instead of dying every so often.
best,
Tom
From db8d9d3e674328b81fabc3dcb7981b4727a8e369 Mon Sep 17 00:00:00 2001
From: Thomas
Hi,
I've got a theory on what's happening now, but I am not sure how to
fix this. The following seems to be the order of events that kills
libvirtd:
[Thread A] lxcDomainDestroyFlags requested
[Thread A] acquires driver lock
[Thread A] cgroup killing triggers VIR_EVENT_HANDLE_HANGUP
[Thread B] lxc
Hi,
I'm running version 0.9.10 on debian [1]. When starting and stopping
containers I see deadlocks every so often. I attached a full backtrace
(see at the end of the log). This seems to happen in conditions
similar to the segfault I wrote about recently. I.e. usually when
destroying domains.
I'm
Hi,
I am running libvirt 0.10 on debian [1] and I am getting the following
segfault [2] about once a day when using the lxc driver. I don't know
how to reproduce the circumstances unfortunately. It seems to happen
when destroying domains. I attached the debug log leading up to the
segfault. Is thi