We are having disk issues with IBM VIO servers using CX4 disks(SATA). The symptom are that when we do veritas commands (specifically vxdisk list) it take for ever.... like 30-45 minutes for the response. One time, unknowingly we issued too many of the commands and it eventually panicked the box. We are digging deeper into the problem and I am wondering about the initiator type and the Failover mode.
We have several VIO server running on both CLARiiON (CX3 & CX4) and Symmetrix (DMX3) using PowerPath with no problems. We configure VIO identical to AIX and everything just 'works'.
vxdisk list communicates with the kernel devices to get the list of disk drives/luns available to the server. If that takes a long time to finish the list, then I'd look in to the following things first
->Check the output of powerpath & see if there ->Check for filesystems issues ->Check the device messages, dmesg, log messages ->Check for Vertias log messges
We were told to use Failover Mode 3 in our VIO environment. Told it handles NDU Upgrades better. We however don't have veritas in our environment and are using Powerpath as well. So not sure if that mode takes those things into account or not.
Hey hersh. Thanx for the reply. Can you tell me who you mean by "we were told"? Was it EMC support, was the suggested Failover Mode in any form of a document you could share?
Allen Ward
6 Operator
•
2124 Posts
350
0
Posted October 15th, 2009 11:00
We have several VIO server running on both CLARiiON (CX3 & CX4) and Symmetrix (DMX3) using PowerPath with no problems. We configure VIO identical to AIX and everything just 'works'.