On Wed, Mar 21, 2018 at 7:01 PM, Hanson Turner <hanson@andrewswireless.net> wrote:

Hi Sahina,

On the fourth node, I've found /var/log/glusterfs/rhev-data-center-mnt-glusterSD-ovirtnode1.core\:_engine.log ... is this the engine.log you're referring to or do you want one from the hosted engine?

I was referring to oVirt's engine.log. Found under /var/log/ovirt-engine/engine.log

I actually do want to go replica 5. Most VM's it runs are small(1 Core,1gb Ram,8gb HDD) and HA is needed. I'd like a bigger critical margin than one node failing.

Ok, in this case - in addition to adding the new nodes to cluster, you will also need to add bricks to the volume and increase the replica count to 5. You can do this using the "Add Bricks" from the Bricks tab on selection of a gluster volume. Ensure that you set the replica count to 5 here. This should be done once you successfully add the new hosts to the cluster.

As far as the repos, it's a straight the ovirtnode iso install, I think it's Node 4.2.0... which is yum updated to 4.2.1.1
When I installed 4.0 I'd installed on top of centos. This round I went straight with the node os because of simplicity in updating.

I can manually restart gluster from cli, the peer and volume status show no peers or volumes.

This is because the nodes have not been added (or peer probed) to existing gluster cluster.

I will need the logs I requested to understand why.

One thing of note, the networking is still as setup from the node install. I cannot change the networking info from the ovirt gui/dashboard. The host goes unresponsive and then another host power cycles it.

Thanks,
Hanson

On 03/21/2018 06:12 AM, Sahina Bose wrote:

On Tue, Mar 20, 2018 at 9:41 PM, Hanson Turner <hanson@andrewswireless.net> wrote:

Hi Guys,

I've a 3 machine pool running gluster with replica 3 and want to add two more machines.

This would change to a replica 5...

Adding 2 more nodes to cluster will not change it to a replica 5. replica 3 is a configuration on the gluster volume. I assume you don't need a replica 5, but just to add more nodes (and possibly new gluster volumes) to the cluster?

In ovirt 4.0, I'd done everything manually. No problem there.

In ovirt 4.2, I'd used the wizard for the hosted-engine. It looks like the fourth node has been added to the pool but will not go active. It complains gluster isn't running (which I've not manually configured /dev/sdb for gluster). Host install+deploy fails. Host can go into maintenance w/o issue. (Meaning the host has been added to the cluster, but isn't operational)

Are the repos configured correctly on the new nodes? Does the oVirt cluster where the nodes are being added have "Enable Gluster Service" enabled?

What do I need to do to get the node up and running proper with gluster syncing properly? Manually restarting gluster, tells me there's no peers and no volumes.

Do we have a wizard for this too? Or do I need to go find the setup scripts and configure hosts 4 + 5 manually and run the deploy again?

The host addition flow should take care of installing gluster.

Can you share the engine log from when the host was added to when it's reported non-operational?

Thanks,

Hanson

_______________________________________________
Users mailing list
Users@ovirt.org
http://lists.ovirt.org/mailman/listinfo/users