Monitoring Db2 HADR Takeover
In a DB2 HADR system, if the Primary database encounters a fatal error or is brought down, then the Standby database has to takeover to become the new Primary. Sometimes the process could take longer than most would expect. This is because during takeover the old Standby must undo all the open transactions before transforming into the new Primary. Recently we have received a case that the user thought the takeover hanged while it was just taking a long time to finish backward phase. Today I will talk about how to monitor the process of takeover in such scenario and our new feature to allow faster takeover. The scenario When takeover command is issued on an active Standby database, the Standby will have to replay all the log records it has received, and then bring up the database consistently. This means at the end of the logs, if there are open transactions, these transactions must be undone as they have not committed at the time of takeover. Normally, this should not take too ...