Log in


Tags » ‘Storage’

DataOntap 7.3 upgrade & Windows 2008 disks offline on server reboot

Categories: NetApp Windows 2008
May 6th, 2010 by Scott

A colleague of mine sent me a bug that is occuring to some of our clients when upgrading to DataOntap 7.3 and using Windows 2008.  When the server is rebooted the SAN attached disks do not come back on line.  It has been causing some headaches out there so I thought I would share the fix. 

The process to resolve the issue is a manual process, but the good news is that once you do it it doesn’t appear you have to do it again, or at least that is what I have been told from our client. 

Follow these steps:

  1. On Windows 2008 open System Manager
  2. Expand “Storage” and select “Disk Management”
  3. Right click the “Offline” disk and select “Online” (this will bring the disk back online)

Once the disk has been brought back on line all seems to be well, even after additional reboots.  

The NDU process was used but that did not resolve the issue either as noted in the below KB article. 

There is also a NETAPP article regarding the issue,  KB54672:

Disks show as offline in Windows 2008 after Data ONTAP upgrade

Symptoms

After upgrading Data ONTAP from 7.2.x to 7.3.x, all LUNs attached to Windows 2008 servers appear as ‘offline.’

 Cause of this problem
Data ONTAP upgrades to 7.3.x without using the NDU process will change the LUN revision numbers. Data ONTAP versions prior to 7.3 used a LUN revision number of 0.2, which can be seen using the devcon.exe utility:
MPIO\DISK&VEN_NETAPP&PROD_LUN&REV_0.2_\1&7F6AC24&0&4334677547345438416C4B4F: NETAPP LUN Multi-Path Disk Device 

After 7.2.x, LUN revision numbers took on the version of Data ONTAP and the filer was upgraded to:
MPIO\DISK&VEN_NETAPP&PROD_LUN&REV_7310\1&7F6AC24&0&5672482F2F4A54344E46674C: NETAPP LUN Multi-Path Disk Device

 Solution

 There are several options:

1. Use non-disruptive upgrades of Data ONTAP (with takeover/giveback) to prevent the LUN revisions from changing.|
2. Use SnapDrive to disconnect/reconnect drives, as it will automate the process of bringing drives online.
3. Do nothing and online the drives manually in diskpart or disk management GUI.

To prevent this issue entirely:

 For hosts that fall within the above conditions, we recommend setting the SAN Policy to “Online All” prior to the ONTAP upgrade. This will prevent the virtual disks from dropping offline after a reboot. Once the host is rebooted, the SAN Policy can be changed back to the default setting of “Offline Shared.”

Tags:
No Comments »

Where’s My Single Instance Storage?

Categories: Exchange Server Microsoft Related
February 22nd, 2010 by Scott

The MS Exchange Team has a nice post (http://msexchangeteam.com/archive/2010/02/22/454051.aspx) on their site explaining the history of Single Instance Storage (SIS) and where it went!

Reading the article I couldn’t help think that many clients are still concerned about space on their SANs and fear the idea of placing anything on local disk.  The idea in the article is placing your production mailboxes on Cheap Disk.  The only time a client or anyone should consider this is where you are using Database Availability Groups and have at least three servers in the DAG.  This will allow you to go backupless but also have all bases covered in your environment surrounding performance and recoverability. 

My only other beef though is Archiving.  The biggest question I get from clients is, “can I have my production mailboxes on one database and put the archived mail on cheap disk?”  Currently in Exchange 2010 an archived mailbox is stored on the exact same database as the users’ production mailbox. I have heard rumors that this is being looked into for future releases but nothing concrete.   What this means is that using SAN space which is typically RAID 10 or RAID 5 requires expensive disk for the production mailbox, which by default (if leveraged) archiving is also placed on expensive disk.   Explaining to clients that they can use local disk, have HA, but need at least three mailbox servers isn’t easy.  Not to mention that if you want multiple CAS in your environment with WNLB you need separate machines.  This is due to WNLB and MS Clustering cannot run on the same server.

Otherwise, given the way Exchange is deployed these days it is an enterprise solution.  There are times where departments will be included in a single database and other times where users are stored based on their last names.  I do love what they have done with Exchange 2007 and then what they have also done with Exchange 2010.  I just felt that I had to put in my 2 cents on the SIS discussion since I felt the post was a bit misleading.  While the applications are true, we have to consider real life usage and still address client concerns. 

Tags:
No Comments »