Hi all,
We have to support forced failover under certain scenarios. Whilst using 'ForceFailoverAndAllowDataLoss' work perfectly the database is not available for several minutes. The failover command returns very quickly, but the database enters a recovery/restore state and will not accept connections for up to 2 minutes.
Am I missing an additional step that would speed this up?
Many thanks,
Nick
Hi Nick,
When you force failover, the mirror database first redoes whatever outstanding log it has received from the principal and then restarts the database so it goes through normal crash recovery.
You can monitor the duration of the first phase by looking at the "Redo Queue KB" and the "Redo Bytes/sec" performance counters - the DBM redo is running as long as the redo queue is non-zero. Then, for the second phase there should be entries in the errorlog saying when crash recovery started, what it's doing - redo or undo and when it has completed.
Hope this helps.
Cheers,
Kaloian.
|||Also can you see anything such as warning or error on the SQL Server error log.ssql
No comments:
Post a Comment