If you are using Service Guard, you either have your own scripts embedded into the package or you are using the Extension set from HP that has all of that for you. Typically, it's under /etc/cmcluster/<package name> and you can look at the log from the package to see where it is failing.
Splitting up your CI from DB adds a bit of complexity, so you just have to be careful that each knows where the other is running and that both servers in each cluster are identically configured. Your core issue can't really be resolved until you look at those logs from your failover attempt to see what actually failed.
A common problem I've seen is with the sequence of tasks in the package definition. You have to be very thorough that tasks are done in proper sequence. For example, the DB instance has to be started before the CI instance can be started. If they are on different servers or a different cluster, coordinating all of that is critical.
I know a lot of Basis folks don't get access to the Service Guard Logs but you'll never solve the issue until you do. Those logs have a lot of stuff like starting the volume group and checking the heartbeat, but there will be one section that will be very familiar with the startup of the Oracle Listener and the startup of the DB and the startup of the CI. Work with your Unix Admin to get those logs.
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.