To chime in, I'd be curious to see what results you get when you set the group tag's ordered attribute to false. When I did that on my system (heartbeat 2.0.8, with an upgrade planned), I not only lost my orderedness (a good thing) I also lost my colocatedness (in my case a bad thing). The first process in the group went where it belonged, the second to the other node, then back and forth until the last two (there are about 9 resources in the group), which went to the second node. My current solution is to do ordered=false, collocate=true, and individual location constraints for each resource in the group.
Doug On Tue, 2008-03-25 at 16:02 +0100, Andreas Kurz wrote: > On Tue, Mar 25, 2008 at 3:36 PM, Nicole Hähnel <[EMAIL PROTECTED]> wrote: > > Hi, > > > > I set up a two node cluster Centos 5 with heartbeat 2.1.3 and kmod drbd > > 8.2.5. > > > > I configured httpd, named and ntpd in a resource group. > > For testing I killed httpd. > > I thought crm only restarts httpd, but all resources in the group were > > restartet. > > > > Is this a normal behavior? > > Yes, if you configured "httpd" as the first resource in your group. > > > Or how can I prevent crm from restarting all resources? > > .... set "ordered=false" for the group so only the implied colocation > constraints of the group remain. > > Regards, > Andreas > > > > > Thanks! > > > > Nicole > > _______________________________________________ > > 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 > _______________________________________________ Linux-HA mailing list [email protected] http://lists.linux-ha.org/mailman/listinfo/linux-ha See also: http://linux-ha.org/ReportingProblems
