ONTAP Discussions

Recovering RDM's

ttrulis01
5,938 Views

I am interested to know the process of restoring a server with single vmdk for OS and two physical RDMs for exchange DB and logs.  I have backups of this on SMVI (OS only) and SME (exchange and log snaps).  Can someone point me in the right direction on how to go about restoring a server in this situation?

thanks for any help!

1 ACCEPTED SOLUTION

radek_kubka
5,938 Views

OK, worry not, because:

1) If you SnapMirror a volume to another location, then break the mirror & delete the source, you can create reverse SnapMirror relationship & after baseline transfer & breaking the mirror again you are back to square 1. And BTW - if it's VSM then all shapshots within the volume are preserved.

2) You can actually recover Exchange at the remote location, should you wish to do so. The detailed steps are described in TR-3683 (link above) on page 27 - "10.8 TESTING REMOTE REPLICATION AND DISASTER RECOVERY WITH SNAPMIRROR"

Regards,
Radek

View solution in original post

8 REPLIES 8

radek_kubka
5,938 Views

Hi,

I don't see any issues here & would say the recovery process will be straight forward.

What scenario are we talking about though? Fail-over to a DR site? Or say simultaneous data corruption & OS corruption without failing-over to another filer?

With both SMVI & SME in place all angles are covered, so you can first recover the OS in question utilising SMVI & then use SME to roll back to a consistent state of your Exchange.

These two TR documents offer more detailed recommendations & recovery steps:

http://www.netapp.com/us/library/technical-reports/tr-3683.html

http://www.netapp.com/us/library/technical-reports/tr-3785.html

Regards,
Radek

ttrulis01
5,938 Views

Thanks for the response.

In this case I have just finished decommissioning a 2003 Exchange server.  Before I delete the volume on the filer I want to make sure that if for any reason I need to restore it from SMVI + SME that it is possible.   It doesn't hurt to restore, just to make sure I know how to either. We are also on FC which I forgot to mention.

So off the top of my head I would have to recreate both Volume and LUN for Exchange - then restore OS via SMVI then through Snapdrive reconnect the physical RDMs? The SMVI portion is relatively straightforward but the RDMs have me a little puzzled.  Or would I be just using an SME snapshot to restore logs and exchange?

radek_kubka
5,938 Views
 Before I delete the volume on the filer I want to make sure that if for any reason I need to restore it from SMVI + SME that it is possible.

Hold your horses!

Both SMVI & SME rely on snapshots to do a backup and restore, so if you delete the volume in question, snapshots are gone & so are your backups - unless you have some form of a secondary copy, be it on another filer, or NDMP-generated tape.

Regards,
Radek

ttrulis01
5,938 Views

Well that certainly is a good point.  I guess I'll have to move the snapshots off the volume before doing that.  Good catch.

Going to attempt moving snapshots off volume and restoring to a new volume.

amiller_1
5,938 Views

Hmm.....you can't really move snapshots -- you can do an entire volume copy (with or without snapshots depending on the method).

Basically if you want to retain restore ability you have to keep the data around (statement of the obvious I know....but seems like you're thinking you can remove the volume/data and still restore?).

ttrulis01
5,938 Views

Yes I was looking into lun clone but I really don't have the space on that filer to do that.  I snapmirrored the (physical) RDM's to our DR site.  If I deleted the RDM volumes on the source side, is it possible to snapmirror the volumes back and bring them up again?  My concern is essentially how snapshotting works on RDM's both physical and virtual, although physical in this case.  I was under the impression that physical RDMs are essentially pointers to a specific location on filer and that the pointer would lose its function if the filer it points to is unavailable. This would effectively make snapmirroring physical RDM's useless.  I am worried I may have lost flexibility in restoral processes by making these RDM's physical for SME and SMBR.

Thanks for the responses.

radek_kubka
5,939 Views

OK, worry not, because:

1) If you SnapMirror a volume to another location, then break the mirror & delete the source, you can create reverse SnapMirror relationship & after baseline transfer & breaking the mirror again you are back to square 1. And BTW - if it's VSM then all shapshots within the volume are preserved.

2) You can actually recover Exchange at the remote location, should you wish to do so. The detailed steps are described in TR-3683 (link above) on page 27 - "10.8 TESTING REMOTE REPLICATION AND DISASTER RECOVERY WITH SNAPMIRROR"

Regards,
Radek

ttrulis01
5,938 Views

Thanks! That seems to be the best thing for me to do.  I'll test it out.

Public