[Pacemaker] Resource Agents in OpenVZ containers

Tomasz Kontusz tomasz.kontusz at gmail.com
Sat Feb 15 20:57:50 EST 2014



emmanuel segura <emi2fast at gmail.com> napisał:
>i think, if you use pacemaker_remote inside the container, the
>container
>will be a normal node of you cluster, so you can run pgsql + vip in it
Right. I didn't want to do it like this, as the containers are not accessible from outside the cluster (they are in separate subnet), and i wanted to avoid nodes acting as routers.

A related question is how to set anti-colocation by HW node (so I'll have database master and apps on different nodes if possible)?


>2014-02-15 19:40 GMT+01:00 Tomasz Kontusz <tomasz.kontusz at gmail.com>:
>
>> Hi
>> I'm setting up a cluster which will use OpenVZ containers for
>separating
>> resource's environments.
>> So far I see it like this:
>>  * each node runs Pacemaker
>>  * each container runs pacemaker_remote, and one kind of resource
>(but
>> there might be multiple containers providing the same resource)
>>  * containers are started with VirtualDomain agent (I had to patch it
>a
>> bit to work around libvirt/OpenVZ issue),
>>    each container resource is node-specific (and constrained to only
>run
>> on the right node)
>>
>> The problem I have is with running pgsql database with virtual IP in
>such
>> setup.
>> I want to have IPaddr2 resource started on the node that holds
>container
>> with current pgsql master.
>> How can I go about achieving something like that?
>> Is the idea of using pacemaker_remote in such setup sensible?
>>
>> --
>> Tomasz Kontusz
>>
>> _______________________________________________
>> Pacemaker mailing list: Pacemaker at oss.clusterlabs.org
>> http://oss.clusterlabs.org/mailman/listinfo/pacemaker
>>
>> Project Home: http://www.clusterlabs.org
>> Getting started:
>http://www.clusterlabs.org/doc/Cluster_from_Scratch.pdf
>> Bugs: http://bugs.clusterlabs.org
>>

-- 
Wysłane za pomocą K-9 Mail.




More information about the Pacemaker mailing list