> > [Damon Estep] > > > > Sounded good, there is a problem though, when the dummy mysql_02 > > migrates back to its preferred node it stops the service on node1. > > Hmmm ... No, why do you think so? It's an independent init script that > does not affect the other mysql instance. > > > > > I assume you were dealing with a V1 or crm=no cluster? > > No, crm=yes > > > > > Did you do something special to make the instance really a dummy, > like > > two different init scripts? > > In fact it was a LSB compliant init script for rinetd, configured to > redirect the request for the second mysql instance to the first one > ... and the other way round on the other node. > So it was not exactly the same setup as yours because I had to start > the IP resources prior to mysql because mysql only listened to a > specific IP. Nevertheless it should also work for your setup. > > Regards, > Andreas >
[Damon Estep] Andreas, I missed a key point in your description earlier. I thought you were referring to an "instance" of the RA when I think you were referring to a mysql "instance" as in two mysqld running on the same machine, correct? At any rate, my misunderstanding of your configuration gave me an idea; 1. create a dummy init script so node1 has a mysql1 init that is the real thing and a mysql2 init that is a dummy, and vice versa on node2. 2. collocate ip1 with mysql1 and ip2 with mysql2 3. prefer the location node1 for mysql1 and node2 for mysql2 So if mysql1 fails on node1 it will move to node2 (dummy init script, really does not start mysql but the crm thinks it did) and the IP will move with it. Haven't tried it yet, but sounds good in theory. I guess the real answer to the question is that this setup will required some trickery as there does not appear to be a native heartbeat solution to the problem, is that your belief as well? Damon _______________________________________________ Linux-HA mailing list [email protected] http://lists.linux-ha.org/mailman/listinfo/linux-ha See also: http://linux-ha.org/ReportingProblems
