If a filer is in a cluster that is not configured for immediate takeover on panic and undergoes a system failure but recovers and reboots, the filer behaves exactly like a nonclustered filer.
If a filer is in a cluster that is configured for immediate takeover on panic and undergoes a system failure, or if that filer undergoes a system failure and cannot reboot, the partner filer in the cluster performs an automatic takeover of the failed filer functions and disk drives by creating an emulated filer that
The partner filer maintains its own identity and its own primary functions but also handles the added functionality of the failed filer through the emulated filer.
When a takeover occurs, you view the surviving partner as having two identities, its own and its partner's, that exist simultaneously on the same filer. Each identity can access only the appropriate volumes and networks. You can send commands or log in to either filer by using the rsh command, allowing remote scripts that invoke filer commands through a Remote Shell connection to continue to operate normally.
Commands sent to the failed filer through a Remote Shell connection are serviced by the partner filer, as are rsh command login requests.
If you log in to a failed filer through a Telnet session, you see a message alerting you that your filer failed and to log in to the partner filer instead. If you are logged in to the partner filer, you can access the failed filer or its resources from the partner filer by using the partner command.
When the failed filer is functioning again,
Copyright © 1994-2003, Network Appliance, Inc. All rights
reserved.
Part No. 210-00472
File: clust3.htm
Last updated: 11/17/03 16:57:48