[ovirt-users] GlusterFS Distributed Replicate HA with KeepAlived

[Humble Devassy Chirammal]

The 'referred' mount point is the 'gluster server IP' and the VM behaviour
depends on the Gluster Volume availability as discussed before.

Reg#localhost instead of server IP, I havent tried this, but logically it
should work. You may test it. Also there is an option called
'backup-vol-file-servers' [1] , which should serve(if I am not wrong in
interpreting the mentioned mail thread )the same purpose.

--Humble


On Wed, Aug 20, 2014 at 9:23 AM, Punit Dambiwal <hypunit at gmail.com> wrote:

> Hi Humble,
>
> Thanks for the update...the mount point is the ip address to mount the
> gluster volume,so if this server goes down...all the VM's will pause :-
>
> [image: Inline image 1]
>
> Some one from community suggest me to use the localhost instead of the ip
> address for HA....but i have doubt if i use localhost...then whenever any
> node goes down the VM's of the failed node will not migrate to another
> node... ??
>
> http://lists.ovirt.org/pipermail/users/2014-July/025728.html
>
> Thanks,
> Punit Dambiwal
>
>
>
>
> On Tue, Aug 19, 2014 at 8:23 PM, Humble Devassy Chirammal <
> humble.devassy at gmail.com> wrote:
>
>> Hi,
>>
>>
>> >>
>> ....but i want the way from the ovirt portal...if possible ??
>> >>
>>
>>
>> Yes, its possible.
>>
>> >>
>> Also in my case if the mount point HV node goes down...all the VM's will
>> pause...and once the node come up..i need to manually poweroff the VM's and
>> manually start again...
>> >>
>>
>> I failed to interpret  'mount point' here. If you are talking about
>> 'gluster' volume server , the VM behaviour depends on the gluster
>> configuration which indirectly depend on the quorum configuration and
>> volume configuration. If there is an error from underlying storage ( here
>> gluster volume) VM can pause , however latest versions of ovirt support
>> auto resume.
>>
>> >>
>> Someone from community suggest me to use the localhost as mount point if
>> you are using compute & storage on the same node...but i doubt about the
>> failover...that means if any node goes down....how the VM migration on
>> another node will happen ??
>> >>
>>
>> If you are using local data center the VM migration can't happen. You
>> have to configure a storage domain which can be accessed from your Ovirt
>> Cluster  Hosts .
>>
>> Please have a look at
>> http://www.ovirt.org/Gluster_Storage_Domain_Reference which talks about
>> the recommended way of configuring it.
>>
>> --Humble
>>
>>
>> On Tue, Aug 19, 2014 at 7:26 AM, Punit Dambiwal <hypunit at gmail.com>
>> wrote:
>>
>>> Hi Humble,
>>>
>>> Thanks for the updates....but i want the way from the ovirt portal...if
>>> possible ??
>>>
>>> Also in my case if the mount point HV node goes down...all the VM's will
>>> pause...and once the node come up..i need to manually poweroff the VM's and
>>> manually start again...
>>>
>>> Someone from community suggest me to use the localhost as mount point if
>>> you are using compute & storage on the same node...but i doubt about the
>>> failover...that means if any node goes down....how the VM migration on
>>> another node will happen ??
>>>
>>> Thanks,
>>> Punit
>>>
>>>
>>> On Mon, Aug 18, 2014 at 9:37 PM, Humble Devassy Chirammal <
>>> humble.devassy at gmail.com> wrote:
>>>
>>>> Hi Punit,
>>>>
>>>>
>>>> If you are using gluster volume for virt store, it is always recommend
>>>> to enable /considering virt store use case mentioned here :
>>>> http://gluster.org/documentation/use_cases/Virt-store-usecase/
>>>>
>>>> Regarding how to add bricks to the volume , you can refer #
>>>> https://access.redhat.com/documentation/en-US/Red_Hat_Storage/2.1/html/Administration_Guide/sect-User_Guide-Managing_Volumes-Expanding.html
>>>>
>>>> The virt store use case enables quorum which ideally resolves the
>>>> possibilities of split brain.
>>>>
>>>> Please feel free to revert if you come across any issues.
>>>>
>>>> --Humble
>>>>
>>>>
>>>> On Fri, Aug 15, 2014 at 12:05 PM, Punit Dambiwal <hypunit at gmail.com>
>>>> wrote:
>>>>
>>>>> Hi,
>>>>>
>>>>> I have 4 node GlusterFS Distributed Replicate volume...the same 4 host
>>>>> node i am using for compute purpose....now i want to make it HA....so if
>>>>> any host goes down .....VM will not pause and it will migrate to another
>>>>> available node...
>>>>>
>>>>> 1. Can any one have any document or reference to do this with
>>>>> keepalived...
>>>>> 2. I have one more node as spare...so if any host goes down and can
>>>>> not come up again because of any HW failure...i can add it...but i didn't
>>>>> find any way to add these bricks to volume...??
>>>>>
>>>>> Thanks,
>>>>> Punit
>>>>>
>>>>> _______________________________________________
>>>>> Users mailing list
>>>>> 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/20140822/ae403e64/attachment-0001.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: image.png
Type: image/png
Size: 27695 bytes
Desc: not available
URL: <http://lists.ovirt.org/pipermail/users/attachments/20140822/ae403e64/attachment-0001.png>