UNSOLVED
Trespassed LUNs
Hey
I have got a customers with two VNX5500. Above those systems there is a datacore storage virtualization solution for HA/LB and so on. So one system 1 there was an SP error. So all LUNs seems to be trespassed from SPB to SPA.
I validated the LUNs on system 2 so the same situation is the case. What I don't know, is this due to the datacore appliance? Or was this by design?
The old SPB is running again but all LUNs are still located on SPA. Questions:
1) To rebelance do i need to run a CLI or can this also be done by Unisphere?
2) Should this be done automatically after the SPB is good again?
3) Could i be that this should be by design in relation to the virtualization layer (Datacore). Could the trespassing have inpacts to the datacore?
I would assume that it is recommanded to split up the LUNs even accross the SPs ;-)?
Thank you for your repsonses
ierikson_8c21b2
1 Rookie
•
34 Posts
2970
0
Posted December 15th, 2014 07:00
From my brief knowledge of datacore - it may be managing the pathing used by vmware datastores. This could be the cause of the pathing issues. You may want to see if there is a datacore+SAN (preferably EMC) white paper/best practices documentation. They may reccomend ALUA or another failover method for the LUNs.
Yes, you would hopefully have your LUNs balanced in SP ownership.
Ian
Hosts should have failover software that will auto-restore the LUNs to the correct SP owner after an SP reboots - PowerPath does this as does Round Robin for ESX.
One word of caution - on the VNX series when you use Pools, it is very important that you do not trespass LUNs as this will cause a performance hit. Nor should you change the SP owner for a Pool LUN, for the same reason.
Pool LUNs have three owner models: Allocated (this never changes), Default SP Owner and Current SP Owner. For best practice you should always have the Current = Default = Allocated for best performance. If you need to change the Default SP owner for a Pool LUN for balance, you must use LUN Migration.
glen
brettesinclair
2 Intern
•
715 Posts
2970
0
Posted December 15th, 2014 04:00
Hi Duker,
1. I find it fastest to do from naviseccli, syntax is
naviseccli -h arraysp trespass mine
this moves the LUNs back to their Default owner for the arraysp you specify in the command, so run it against the restored SP.
getlun -trespass can be used to check trespass status.
Can also be done from Unisphere lun by lun.
2. Not necessarily, depends on the failover mode and the host os. If you have Powerpath, failback is auto.
3. I'm not familiar with Datacore. Was an SP Collect checked for causes ?