Andrew, thanks for this information. Is there any way to influence the "weight" of the resource rather than just by pure number? Some resources are more intensive than others. I don't think increasing the score for the backup will work in a N+1 type backup scheme. As nodes fail I don't want all the resources to keep going to the backup. I'd like them to be spread evenly across the available nodes. It seems like this scenario will work correctly if I could give the different resources some type of weighting factor.
thanks, kevin On May 8, 2009, at 9:10 AM, Andrew Beekhof wrote: > On Tue, May 5, 2009 at 11:32 PM, Kevin Harms <[email protected]> > wrote: >> >> In the following documentation: >> http://clusterlabs.org/mediawiki/images/7/7d/Configuration_Explained_0.6.pdf >> there is a sub-heading titled "What if Two Nodes Have the Same >> Score" that contains the statement "It would likely have placed >> Webserver based on the node’s uname and Database based on the desire >> to spread the resource load evenly across the cluster." >> >> I would like to know how the "desire to spread the resource load >> evenly" is determined? > > If two nodes have the same score, the one with the least number of > resources on it gets the next one. > If both have the same amount, its effectively random. > >> I'm trying to setup my cluster so resource will >> end up on my "backup" node when a primary node fails. > > Give it a lower score (with a rsc_location constraint) than the > primary and a higher score than the other nodes in the cluster. > _______________________________________________ > Linux-HA mailing list > [email protected] > http://lists.linux-ha.org/mailman/listinfo/linux-ha > See also: http://linux-ha.org/ReportingProblems _______________________________________________ Linux-HA mailing list [email protected] http://lists.linux-ha.org/mailman/listinfo/linux-ha See also: http://linux-ha.org/ReportingProblems
