[Users] Running Gluster on hypervisor nodes

[Vincent Van der Kussen]

>> On 10/09/2013 12:01 AM, Vincent Van der Kussen wrote:
>>> Hi,
>>>
>>> We are currently exeprimenting with running Gluster on the hypervisor
>>> nodes. To test
>>> this we have 2 machines running as hypervisor nodes with each a
>>> replicated
>>> Gluster Volume.
>>>
>>> We created a VM and stored the disks on the Gluster volume and this
>>> works
>>> quite well.
>>> However, there's one thing that doesn't seem to work or might not be
>>> supported.
>>>
>>> When we put one of the 2 nodes that holds the SPM role in maintenance,
>>> the
>>> VM is
>>> successfully migrated, but the SPM role is not migrated to the
>>> remaining
>>> active hypervisor.
>>> As a result, the Data Center goes into "Down" state.
>>>
>>> Other than that the VM keeps working.
>>>
>>> Our question is. Can this be done?
>>>
>>>
>>
>> you're using the webadmin to place the host in maint mode?
>> SPM role should migrate. anything bad in the logs?
>>
>
> Hi,
>
> I added 2 files with excerpts from the logfiles.
>
> our setup
> - cloud-001.test.btr.local -> engine
> - cloud-002.test.btr.local -> hypervisor1 (had SPM role)
> - cloud-003.test.btr.local -> hypervisor2
>
> Regards,
> Vincent
>

Hi,

Apparently it is working. We just didn't wait long enough. It takes around
1-2 min to transfer the role.

Nice!

-- 
Vincent Van der Kussen
@vincentvdk