[systemd-devel] Interesting bug in cgroup limits management

[Lennart Poettering]

On Thu, 27.10.11 16:28, David Strauss (david at davidstrauss.net) wrote:

> I discovered what appears to be a quite unusual bug in application of
> cgroup limits to a daemonized service:
> 
> https://bugzilla.redhat.com/show_bug.cgi?id=749687
> 
> Because rules *sometimes* apply properly on the second service restart
> following a daemon-reload and *almost never* apply properly on the
> third, I'm wondering whether this is a race condition between the cgroup
> setup and the writing of the limits.

Fixed now in git. When restarting a service we endeded up deleting the
original cgroup but not recreating the new cgroup.

Lennart

-- 
Lennart Poettering - Red Hat, Inc.