Need to evaluate the RAC in one instance failure, how long can restore services remaining instances,
Suppose A RAC contains A and B two instances, now A suddenly hang up, then instance B after A long time, can provide normal service?
Personal understanding:

1, the first is the heartbeat timeout 30 seconds (default), during this period, the RAC cluster will not have any action, but when accessing session B need to update A cache block, originally A holds will hang live or error;
2, after reaching the timeout time, within the cluster instance of survival (B) started the resource reconfiguration, this process is to freeze the database access;
3, for instance reads fault instance redo log for recovery, recovery, after the completion of the database to thaw;
4, for example the undo log back online, this stage should not affect access to the database,

So, after A failure, B in front of the three phases is also can not be normal access,

I ask, my understanding is correct? If it is correct, the former three stages in general a total of how much time can take up (must be related to actual deployment, just want to know a rough experience value)?

CodePudding user response:

B machine has been providing services, A dead, the will give A request, should be sent to the B

CodePudding user response:

Has been providing services, isn't it, how to restore the