On Wed, Jul 31, 2013 at 04:45:03AM -0400, Sven Knohsalla wrote:
Hi,
I just set up a vanilla oVirt 3.2.1 environment, containing
oVirt engine 3.2.1 F18 based oVirt Engine Version: 3.2.1-1.fc18
F18 minimal node vdsm-4.10.3-10.fc18 libvirt-0.10.2.6-1.fc18 1.2.2 - 13.fc18 K ernel 3.9.11 - 200.fc18.x86_64
gluster-storage, replicated, EL6 glusterfs 3.4.0
F18 minimal node was installed via oVirt engine, worked fine, expect vdsmd wasn't started (node non-responsive for engine)
Manually starting vdsm works (HV up in engine), autostart of vdsm (during boot) fails:
Jul 31 10:04:47 deovn-a04 systemd-vdsmd[711]: Note: Forwarding request to 'systemctl disable libvirt-guests.service'. Jul 31 10:04:48 deovn-a04 systemd-vdsmd[711]: vdsm: libvirt already configured for vdsm [ OK ] Jul 31 10:04:48 deovn-a04 systemd-vdsmd[711]: Starting ntpd... Jul 31 10:04:48 deovn-a04 systemd-vdsmd[711]: Redirecting to /bin/systemctl start ntpd.service Jul 31 10:04:48 deovn-a04 systemd-vdsmd[711]: Starting wdmd... Jul 31 10:04:48 deovn-a04 systemd-vdsmd[711]: Redirecting to /bin/systemctl start wdmd.service Jul 31 10:04:48 deovn-a04 systemd-vdsmd[711]: Starting sanlock... Jul 31 10:04:48 deovn-a04 systemd-vdsmd[711]: Redirecting to /bin/systemctl start sanlock.service Jul 31 10:04:48 deovn-a04 systemd-vdsmd[711]: Starting iscsid: Jul 31 10:04:48 deovn-a04 systemd-vdsmd[711]: Redirecting to /bin/systemctl start libvirtd.service Jul 31 10:06:17 deovn-a04 systemd[1]: vdsmd.service operation timed out. Terminating. Jul 31 10:06:17 deovn-a04 systemd[1]: Unit vdsmd.service entered failed state.
Did find similar issues: https://bugzilla.redhat.com/show_bug.cgi?id=832199 https://bugzilla.redhat.com/show_bug.cgi?id=720359 http://lists.ovirt.org/pipermail/users/2012-September/009624.html
Unfortunately, there's no hint.. libvirtd starts without any latency when manually triggered.
The strange thing here: F18 node deploy worked with engine 3.2.0 3-4 weeks ago. Did update engine / F18.
Can anybody give me an advice here?
I have only debugg idea. Could you extend TimeoutStartSec=10min in /usr/lib/systemd/system/vdsmd.service to give you more time for debugging? Then you can find the pid of the vdsmd.service script, and pstree -p <pid> to see what it runs. You can edit the first line of /lib/systemd/systemd-vdsmd to #!/bin/sh -xv to vomit more information to systemd's log. Please inform us what you found out. Dan.