Hello Suresh, The FC clients will experience an outage. Suppose in your scenario that node1 is the filer you have taken down. node1's FC clients will not be able to perform I/O to node1's LUNs through node2, because the "cf disable" command prevented node2 from placing a SCSI reserve on the node1 disks as part of the takeover process. With an active/active pair of nodes, only the node that owns the SCSI reserve for a disk can perform I/O to that disk, So, the only way prevent an outage to the FC clients is to perform the cluster takeovers and givebacks that are part of the NDU. Hope this helps.
Neal Jones IBM N Series Product Field Engineering
From: "Suresh Rajagopalan" SRajagopalan@williamoneil.com To: "NetApp Toasters List" toasters@mathworks.com Date: 02/19/2010 05:52 PM Subject: Single image cfmode and mpio failover Sent by: owner-toasters@mathworks.com
Scenario:
Filers running 7.2.6.1, cfmode -- single image FC clients, use DSM/MPIO to connect
If a non NDU upgrade is attempted (after setting cf disable), will taking down 1 of the filers make the FC clients use the passive path for failover, or will the FC clients experience an outage? Note I am not doing “cf takeover” before bringing one of the filers down. I’d just like to know what the client failover logic is in this case.
Thanks Suresh