4.17 Slow Failovers

Failovers might be slow because the resource is slow to go offline on the source cluster. This can happen if the failover occurs during a time when file I/O is taking place on the cluster.

To resolve this problem, edit the resource unload script and change the NUDP DEL name ip line to NUDP ODEL name ip. Unlike the DEL command, the ODEL command does not wait for all NCP connections to close. This makes it much faster.

NOTE:The cluster resource must be brought offline and then back online for changes to the unload script to take effect. Client data might be lost if clients are accessing the resource when it is brought offline.