On 06/25/2012 09:26 PM, Nathan Stratton wrote:
On Mon, 25 Jun 2012, Robert Middleswarth wrote:
Make sure vdsm-gluster is installed on all nodes? It wasn't installed on nodes that were added before I added a check box to support gluster on the cluster.
Yes, it was out of 8 boxes I was looking at the vdsm logs on box 1 and ovirt-engine tried to start gluster on a different box so I did not see the vdsm log. Once I did, it was clear that gluster we not happy about the directory being used for gluster in the past. To fix that I just ran the following on all 8 of my hosts:
cd /export/ ;for i in `attr -lq .`; do setfattr -x trusted.$i .; done
I now have a happy gluster volume called share. Now I am lost as to how to use that volume. Do I have to change the Default NFS storage type? How do I get this bad boy mounted? I spent 2 weeks tiring to get glusterfs working with no luck. However I was able to get gluster nfs working. Here are the steps I needed to do.
1) I added the following option to the volume. Might only need the 1st one but I have all 3 of them set on my working system. a) nfs.nlm off b) nfs.register-with-portmap on c) nfs.addr-namelookup off 2) Mount the volume outside of ovirt. mount -t nfs ipofonenode:/volumename /temp/share/name 3) chown -R 36.36 /temp/share/name. 4) umount /temp/share/name 5) Added a new NFS domain with the NFS Export Path: localhost:/volumename I was then able to active the NFS based Data Center and add working VM's Thanks Robert
<> Nathan Stratton nathan at robotics.net http://www.robotics.net