[ovirt-users] Ovirt GusterFS assistance

[Sahina Bose]

[+ ovirt-users]
On 03/24/2015 01:28 PM, Jonathan Mathews wrote:
> Please note you have not been helpful at all as I have tried many 
> things and have not been able to solve the issue and please do realize 
> I am trying to do it through the ovirt front end.

You can provide the mount options from the ovirt front end, while 
creating the storage domain.
I'm assuming you selected Storage type as "Data/Posix compliant FS" or 
"Data/GlusterFS". There is a "Mount Options" field where you can specify 
"backupvolfile-server=<list of comma separated host names>". These are 
the other nodes on which the volume is replicated. So if the primary 
server is down, it will try to use the other servers to get the volume 
file. This is useful only when the virt server is first trying to mount 
the gluster volume.

Ideally, if your virt hosts are UP,  when one of the gluster servers 
goes down, your storage domain should be UP. (even without the 
backupvolfile-server option)

I'm not very familiar with oVirt Storage Domain ,so someone from the 
oVirt storage team should be able to guide you further.


>
> On Tue, Mar 24, 2015 at 9:50 AM, Sahina Bose <sabose at redhat.com 
> <mailto:sabose at redhat.com>> wrote:
>
>
>     On 03/24/2015 12:13 PM, Jonathan Mathews wrote:
>>     the gluster volume is raid 0 replicated over 3 nodes and yes you
>>     are correct when one server goes down the volume stays up, but
>>     the storage domain goes down and that is where I need to find a
>>     volume identity to mount it by in storage domain.
>
>     Aah..ok. Maybe the/"/////backupvolfile-server" mount option would
>     help you?
>
>     [+ Allon] for storage domain issue.
>
>
>>
>>     On Mon, Mar 23, 2015 at 12:02 PM, Sahina Bose <sabose at redhat.com
>>     <mailto:sabose at redhat.com>> wrote:
>>
>>         What is the type of volume that you've created? Is it a
>>         replicate volume?
>>
>>         # gluster volume info - should give you this information
>>
>>         If you're replicating the volume across 3 nodes, even when
>>         one of the server goes down, your storage domain should still
>>         be UP.
>>
>>         thanks
>>         sahina
>>
>>         On 03/23/2015 02:10 PM, Jonathan Mathews wrote:
>>>         Hi I am trying to setup an Ovirt, Glusterfs, VM servers. I
>>>         have followed examples on setting up Ovirt and they have
>>>         helped me so far, but not the end point that I am looking for.
>>>         The web sites are:
>>>         http://community.redhat.com/blog/2014/10/up-and-running-with-ovirt-3-5/
>>>         http://community.redhat.com/blog/2014/05/ovirt-3-4-glusterized/
>>>         http://www.linuxplumbersconf.org/2012/wp-content/uploads/2012/09/2012-lpc-virt-storage-virt-kvm-rao.pdf
>>>
>>>         I am running 3 HP micro servers and 2 HP DL360 G5
>>>         The 3 micro servers are my glusterfs storage and have been
>>>         provisioned for virt storage.
>>>         The 2 DL360 are my processing machines.
>>>
>>>         Now my 3 gluster hosts are in one cluster, the volume is in
>>>         up status and has been provisioned for Virt Storage. But the
>>>         problem is that my mount point is directed to one server, so
>>>         when that server goes down, the volume storage domain goes
>>>         down. I am not sure whether there is a way of mounting it by
>>>         a "volume identity", so when a server goes down the storage
>>>         domain stays up.
>>>
>>>         With my 2 processing hosts, I have them in one cluster, but
>>>         I have not gotten any where with this, as I want the Virtual
>>>         machines to use the gluster volume as storage but use the
>>>         processing hosts hardware for processing power.
>>>
>>>         I would appreciate any assistance.
>>>
>>>
>>>         _______________________________________________
>>>         Users mailing list
>>>         Users at ovirt.org  <mailto:Users at ovirt.org>
>>>         http://lists.ovirt.org/mailman/listinfo/users
>>
>>
>
>

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.ovirt.org/pipermail/users/attachments/20150324/712196a6/attachment-0001.html>