4Node P4500 constantly restriping

[MattGG]

I have a 4 node P4500 (10.00) with 12 NR5 LUNs and 68% capacity remaining. 

The LUNs are constantly restriping.  It is usually 2 at a time. 

What would cause this?  Where can I look to troubleshoot this issue.

Thanksm

-MattGG

P.S. This thread has been moved from Storage Area Networks (SAN) (Enterprise) to HP StoreVirtual Storage / LeftHand. - HP Forum Moderator

[a_o]

Basically, for an existing LUN, re-striping occurs when a  node goes down for a while and then comes back up.

If it's been down for a while, re-striping will take longer so that the down nodes are 'caught-up'.

Otherwise, re-striping can occur when you've made changes to your LUNs -i.e. NR10 ->NR5, etc

There's no log per-se in SAN i/Q to view that would let you know why the node went down after the fact.

However, in CMC go to one of the nodes, expand to "Diagnostics->Log Files", RMC and save the "eventMan.log". Open in Notepad,  search inside the file for "Event Manager Startup" and make note of the time stamp. This indicates a re-boot.

Also, search for "resyncSaniqObjects", and look for your LUN names in the vicinity.

This will let you know when your re-stripes started.

However, it would help if you give more information about your set-up:

Are they in the same site in a multi-site cluster?

Do they ever complete the re-stripe?

[MattGG]

They are in the same site.

It seems like there are always 2 restripes going on.  They complete and 2 different LUNs begin to restripe.  The network RAID level has not changed and no new LUNs have been added in at least a month.

Thanks,

-MattGG

[a_o]

It would seem that one or more of your nodes has been going down or disconnected without being noticed.

It coould also be a networking issue with the downed nodes.

Again, if the re-stripes are running for long enough that you're noticing them,  the main contributing factor would be loss of a node. Also, note that if the LUNs are large in size you will tend to notice them re-striping, even if the downed node was down for a few seconds.

If you can't find any clues in the above mentioned logs, I would give support a call.