On 05/18/2016 01:57 PM, H Yavari wrote: > Hi, > > Thank you for reply. > I tested the first method "multi-site cluster". it was ok with ticket > manually assignment. But I had issues with running Booth. > I tested second method "constraints and attributes". I make a cluster > with 4 nodes and define some constraints for nodes. But I have > problems with node relations now. > I was searching in the docs then I found > "http://clusterlabs.org/doc/en-US/Pacemaker/1.1/html/Pacemaker_Remote/ch05.html". > I thinks this is very close to my answer. Do you offer this solution?
Going with pacemaker-remote would be a 3rd option. Could imagine to have the 2 nodes with the master/slave-resource be the full-fledged pacemaker-nodes and the 2 other nodes would then be remote-nodes. On the other hand this would end up in a 2-node-cluster which is bad for quorum and thus should be avoided whenever possible - especially if you have enough nodes anyway. Thinking in the direction of making the remote-node-resources both tied to one of the full nodes each and collocated with the master role. And your service would then be tied to the remote nodes. Sounds like something funny to play with although I've never set up anything like this ;-) But as pacemaker-remote is not broadly used this way it is probably not advisable to use that in a production environment. Or how did you have in mind to leverage pacemaker-remote for your scenario? > > Regards, > H.Yavari > > > ------------------------------------------------------------------------ > *From:* Kristoffer Grönlund <kgronl...@suse.com> > *To:* H Yavari <hyav...@rocketmail.com>; Cluster Labs - All topics > related to open-source clustering welcomed <users@clusterlabs.org> > *Sent:* Wednesday, 18 May 2016, 10:36:39 > *Subject:* Re: [ClusterLabs] Two related Cluster > > H Yavari <hyav...@rocketmail.com <mailto:hyav...@rocketmail.com>> > writes: > > > Hi, > > So you think for this solution Booth is better or attribute method? > I'm not familiar with them so can you share your experiences with > them?Many thanks. > > > > I think a single cluster using node attributes should be a lot easier to > understand and maintain, so I'd recommend that solution if it works out > for you. > > > Cheers, > Kristoffer > > -- > // Kristoffer Grönlund > // kgronl...@suse.com <mailto:kgronl...@suse.com> > > > > > _______________________________________________ > Users mailing list: Users@clusterlabs.org > http://clusterlabs.org/mailman/listinfo/users > > Project Home: http://www.clusterlabs.org > Getting started: http://www.clusterlabs.org/doc/Cluster_from_Scratch.pdf > Bugs: http://bugs.clusterlabs.org _______________________________________________ Users mailing list: Users@clusterlabs.org http://clusterlabs.org/mailman/listinfo/users Project Home: http://www.clusterlabs.org Getting started: http://www.clusterlabs.org/doc/Cluster_from_Scratch.pdf Bugs: http://bugs.clusterlabs.org