[ovirt-users] Trouble Adding Gluster Host in oVirt Manager

[Langley, Robert]

Hello oVirt Community,

This may be my first time writing in. Although, I've been reviewing and playing with oVirt for the past 3 years. My name is Robert Langley and I work for a local Government in Southern California. My agency (with me as the primary/lead IT person) has been using VMWare vSphere to host our server environment for 6 years. I have a few different reasons I want to jump the VMWare ship and I've been pleased with the progression oVirt has been making. I do not care to jump from VMWare to Microsoft's Hyper-V, so this had lead me to look into Xen for a while, then dropped anchor with oVirt. I'm determined to see this implementation through.

I have moved over from the test (workstation class) environment, where I was really just getting familiar with the commands and UI, and now I am working on implementation with server class hardware.
I have the engine up and running. I'll get into more specifics below.
I'm running into an issue where I have not been able to add my Gluster hosts or add the Storage Domain (I had tried adding the storage domain first).
For my new oVirt and GlusterFS hosts I have the following:
*       I want to first mention that I do have a DNS server for the separated/private networks and those are configured and working properly (forward & reverse).
Dell PE R830 and I would like this to be a failover server housed in a separate building where we have an Emergency Operations Center and Datacenter (server room).
-       Has bonded gigabit (2 port) network that is connected to our main network & internet, configured to be used for ovirtmgnt. Name = gsaov08.ent...
o       As far as all the various names, oVirt Hosted Engine Deployment complained that the Main network name was seen on more than one interface, or something to that effect.
o       I did originally have DNS for the separated/private networks configured for forwarding DNS from our main network DNS servers. Which has been removed and the Hosted Engine Deployment stopped complaining at that point. I was going to use the forwarding DNS for another private network I use for some management traffic, but that is not as important or an absolute need either.
-       Has 2nd bonded gigabit (4 port) network (separate/private) for engine storage. Name = gsaov08e.engine.local
-       Has 3rd bonded 10G (2 port) network (separate/private) for VM Guest storage. Name = gsaov08s.stor.local
-       Will have a 4th network (separate/private) configured for security related devices that are managed by a couple of our VM guests.
    As far as storage in this same server.
-       I have split up the local storage to be hosted by GlusterFS and this host is a peer with the other two dedicated GlusterFS storage servers. That is for the failover piece.
o       Engine volume (currently being used for the Hosted Engine) and setup as a replica 3 with the two dedicated GlusterFS storage servers.
o       VM Guest "Data" volume. The volume is created, configured and started. Setup as a replica 3 with the two dedicated GlusterFS storage servers.
So far, this is the only host I have hosting the engine. Now, onto the other host I am about to run Hosted-Engine -Deploy on.
Dell PE R730 - It has the same network setup as the R830. Storage is a bit different. I would like to run another VM Guest GlusterFS volume from here, along with 2 other hosts I have. They are currently serving as the vSphere ESX hosts and have the same total size of storage to create another replica 3.

The Dedicated GlusterFS Hosts are custom AMD Opteron G2 builds with ASUS server motherboards. Strictly designed to be dedicated storage servers.
-       I'm using the onboard NICs to connect to my management network only for the purpose of getting updates. So, internet is served to these hosts via a NAT router. Since I figured they don't otherwise need access to our main network. The primary names = gsa-stor1 & gsa-stor2
-       They both have a bonded (4 port) gigabit network for the Engine storage network. Names = gsa-stor1e.engine.local & gsa-stor2e.engine.local
-       They both have a bonded (2 port) 10G network for the VM Guest storage network. Names = gsa-stor2s.stor.local & gsa-stor2s.stor.local
    As far as storage in these servers.
-       It is the same as the R830, as these servers are the peers to the R830.

First, I added an ISO Storage Domain (NFS). That was successful. Next, I was trying to add the GlusterFS VM Guest Data Storage Domain. I receive an Operation Canceled: Error while executing action AddGlusterFsStorageDomain: Error creating a storage domain.
When I try to add the Gluster hosts as hosts in the oVirt Manager, I get a message stating that SSH has failed, verify that the host is reachable (IP address, routable address etc.) You may refer to the engine.log file for further details.
I can otherwise communicate with the hosts and the engine.log does not appear to have specifics to this particular failure. But, there are other clues that I don't understand. So, I can follow up with the engine log. It contains "WARN" entries. They seem to be because I do not have the hosts, for the engine volume, added to the cluster.
Oh, one thing I should mention, I did select the option to enable the Gluster service in the default cluster. So, both virt and Gluster are enabled. After I did so, I wasn't so sure that was a good thing or not. It seems that documentation is outdated on whether that is now supported. I noticed from screenshots that the selection used to be radio buttons and now they are check boxes. I did read the release notes for 4.0.1, which seem to talk as though support for both in a cluster has been added.

Thank you for any help that can be extended.
Robert

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.ovirt.org/pipermail/users/attachments/20170428/78ee59de/attachment-0001.html>