Frederico,
Thank you for responding. I checked the qemu.conf file on all of the nodes and it was
correct. What's even more strange is that last night I put all of the nodes into
maintenance mode and restarted them. No particular reason for this but I was trying
anything I could think of. Now the ownership remains correct. I suppose I'll never
know what I did wrong, but at least it is working.
I found only 2 other anomalies. One, the live migration did not work because TLS was
disabled on the nodes. The "listen_tls = 0" flag was set in
/etc/libvirt/libvirtd.conf. I commented that out and migration works.
The second thing I noticed is that the iptables service is not starting. This doesn't
hinder functionality, but it's a security issue. I'm still checking into why.
I doubt either of these issues caused my ownership problem. I'm sure it's some
mistake I made. Hopefully I won't make it again next time.
Thank you again,
-Jacob
________________________________________
From: Federico Simoncelli [fsimonce(a)redhat.com]
Sent: Thursday, May 10, 2012 3:38 AM
To: Jacob Wyatt
Cc: users(a)ovirt.org
Subject: Re: [Users] image ownership
----- Original Message -----
From: "Jacob Wyatt" <jwyatt(a)ggc.edu>
To: users(a)ovirt.org
Sent: Wednesday, May 9, 2012 8:08:54 PM
Subject: [Users] image ownership
Greetings all,
I've set up a new oVirt installation and it's behaving strangely with
regard to virtual machine image files on the NFS storage. Whenever
I shut down a machine it's changing the owner of the image to
root:root (0:0) instead of vdsm:kvm (36:36). After that it can't
start or do anything with that image again until I manually change
the ownership back. Everything works fine again until I shut the
machine down. I assume this is some mistake I've made in
installation. I did not have this problem in the test environment,
but I'm stumped as to what went wrong.
-Jacob
Hi Jacob,
could you check the dynamic_ownership in /etc/libvirt/qemu.conf:
# grep dynamic_ownership /etc/libvirt/qemu.conf
#dynamic_ownership = 1
dynamic_ownership=0 # by vdsm
Thanks,
--
Federico