EVA / Continous Access / Storage Server Problem

[Matthias Rettl]

Hy there!

My customer has two EVA 4000s (XCS v5.100) with Continous Access and two HP Storage Servers (HP DL380-3.4G HPM Storage Server SAN 371227-B21) with Windows 2003 Server R2 Clustered.
Latest Command View EVA and RSM Versions are installed on one of the nodes. MPIO Full Featured v2 is installed.

Everything is working fine so far.

There are two DR Groups. One mirrors the LUNs from EVA-1 to EVA-2 and the other group from EVA-2 to EVA-1. All those LUNs are presented to both Nodes on both EVAs.

When I initiate a FAILOVER in RSM (to change the primary/secondary EVA) on one of the DR Groups the following phenomenon happens:

A LUN in the DR Group can be accessed by its host after a few seconds IF THE LUN IS _NOT_ CONFIGURED AS A CLUSTER RESSOURCE. This works fine.

BUT: If the LUN is a Cluster Ressource, i cannot access the LUN after the Failover! Neither a Hardware Rescan nor a Reboot helps.


Any ideas for my problem?

Kind regards,
Matthias

[Roland Kastner_2]

Hi Matthias,

which HBA driver, Storeport or Scsiport, do you use ?
Do you boot from SAN ?

Best regards
Roland

[Matthias Rettl]

First i tried the SCSIPort, then updated to Storport. Same result.
No, I do not boot from SAN.

Regards,
Matthias

[Nigel Poulton]

This could well be a cluster related problem. Do you know if the SCSI reserve has been released from the LUN that is causing you problems?

I have seen "similar" problems where the SCSI reserve was not released from the LUN and had to be manually released.

Have you tried issuing a SCSI Reset to the LUN and then seeing if it can be accessed.

[Matthias Rettl]

I have just heard from an HP Storage guy that it is a RSM issue. It *should* be solved in a patch available soon, but there is no workaround of this problem. RSM & CA is CRAP!

But to your questions: i rebootet the clusternodes... this should solve all, but it did not.

I'll keep you informed in this thread if there are some new informations.

Regards,
Matt

[chris barnett_2]

Have you tried initiating a failover through Command View. If this works OK you could then script it using SSSU.

[Matthias Rettl]

Hy!

Just go a call from a hp guy:

The FAILOVER with CA in a Cluster does only work with CLX (Cluster-Extension-EVA) installed!!!
(But he is not sure werther CLX is already available for W2003 Storageserver).


Can anyone agree to this statement???

regards,
matthias

[Antonio Gonzalez_4]

Matthias, any final conclussion regarding clusters ??
I though that CLX was optional for automatic failover, not a requirement for standar functionality.

thanks

[Cavailhe]

Hi Matthias

About CLX compatible with Win2003, YES it is
for Entreprise Edition with service pack 1 and MSCS V5.2.

Best regards.
Olivier Cavailhe