Evpd and Veritas load balancing

[Richard Pereira_1]

Hi,

I am currently investigating some performance issues when I noticed a repeated message from EVPd. My system has been up 194 days but the issue occurred over the past 3.

11.31 server is rx6600. 2 internal drives, everything else is on a Hitachi SAN.

Sep 24 10:03:42 whmh9952 vmunix: Evpd inquiry page 83h/80h failed or the current page 83h/80h data do not match the previous known page 83h/80h data on LUN id 0x0 probed beneath the target path (class = tgtpath, instance = 6) The lun path is (class = lunpath, instance 150).Run 'scsimgr replace_wwid' command to validate the change
Sep 24 10:03:42 whmh9952 vmunix: Evpd inquiry page 83h/80h failed or the current page 83h/80h data do not match the previous known page 83h/80h data on LUN id 0x0 probed beneath the target path (class = tgtpath, instance = 7) The lun path is (class = lunpath, instance 151).Run 'scsimgr replace_wwid' command to validate the change

Its always the same 2 instances (6 & 7). Since I'm using veritas to manage my disks, should I just stop and disable EVP ?

[Richard Pereira_1]

Bueller? Bueller?

Should EVP and Veritas co-exist or just I just shut off EVP since Veritas already does load balancing?

[Viktor Balogh]

>Should EVP and Veritas co-exist or just I just shut off EVP since Veritas already does load balancing?

I don't know what EVP is, but I am pretty sure that Veritas does the loadbalancing. It calls Dynamic MultiPathing, or shortly DMP. To check it, try the (man) vxdmpadm command.

[chris huys_4]

Hi,

evpd is an integral part of HP-UX 11.31, so disabling is a nono. (besides the fact that also vxvm now uses default "diskxxx" device files when issuing a vxdisk -o alldgs list output, so DMP is fully aware and "using" them.. )

What you have here, is that the storage guys 3 days ago, changed something on the storage side. Something like, removing a lun at a certain lunID and then presenting a lun back at the same lunID, without checking in the meantime, if on the host side, something also should be done, between those 2 steps.

And as the wwid of the removed lun, had a different wwid, but the same lunid#, as the
wwid of the newly presented lun, the OS is asking if that new lun, can be acknowledged to get the "same" "diskxxx" device file as the original lun, through the scsimgr replace_wwid command..

Issuing ioscan -fn/ioscan -fnN/ioscan -m hwpath/ioscan -P health/ioscan -m dsf/ioscan -m lun.. should give more details.. about the luns involved.. and asking the storage guys offcourse..

Greetz,
Chris