[Users] Dedicated Bonding interface for Gluster

[Sahina Bose]

On 02/12/2014 11:21 PM, Gianluca Cecchi wrote:
> On Wed, Feb 12, 2014 at 6:18 PM, ml ml wrote:
>
>> I guess the brick details are stored in the postgres database and everything
>> else after will fail?!

Yes, we fixed the issue with resolving brick's host while syncing with 
gluster CLI in oVirt 3.4. However, when you use multiple addresses, you 
will need to use the workaround below.

>>
>> Am i the only one with dedicated migration/storage interfaces? :)
>>
>> Thanks,
>> Mario
>>
> One of the workarounds I found and that works for me as I'm not using
> dns is this:
>
> - for engine host node1 and node two have ip on mgmt
> - for node1 and node2 their own ip addresses are on dedicated gluster network
>
> so for example
>
> 10.4.4.x = mgmt
> 192.168.3.x = gluster dedicated
>
> before:
>
> on engine
> /etc/hosts
> 10.4.4.58 node01
> 10.4.4.59 node02
> 10.4.4.60 engine
>
> on node01
> 10.4.4.58 node01
> 10.4.4.59 node02
> 10.4.4.60 engine
>
>
> after:
>
> on engine (the same as before)
> /etc/hosts
> 10.4.4.58 node01
> 10.4.4.59 node02
> 10.4.4.60 engine
>
> on node01
> #10.4.4.58 node01
> #10.4.4.59 node02
> 192.168.3.1 node01
> 192.168.3.3 node02
> 10.4.4.60 engine
>
> No operations on RDBMS.

Thanks, Gianluca!

I will update the wiki page so that this workaround is clear.

>
> HIH,
> Gianluca