May 24, 2012 at 3:35 pm
I finished swapping out all of my shared drives on two clusters and was able to run cluster validation without error. 🙂
I tested complete network failure as described above and could not produce split-brain.
I did the following to replace all the shared drives without issue:
Add the new "Raw" Drives to Cluster Storage
Add the new SQL "Raw" drives under the SQL Server Cluster Service,
Take just the SQL Server Service off line using the Failover Cluster Manager
xcopy the data (to retain permissions) to the new "Raw" drives.
Write down old and new drive letters since it is necessary to retain the old drive letters
Remove the old SQL shared virtual drives from the SQL Server (using Failover Cluster Manager)
Change the Drive Letter of the old SQL shared virtual drives so that the drive letters can be reused (using Failover Cluster Manager)
Delete the old SQL shared virtual drives from Cluster Available Storage (using Failover Cluster Manager)
Change the Drive Letters of the new "Raw" drives to match the old deleted drives (using Failover Cluster Manager)
Then, using vSphere Client, delete the old virtual disks from each VM
(Cluster Validation will fail all drives even if one unused shared virtual disk is found)
The Quorum Disk can easily be changed to point to a new "Raw" 2GB drive using Failover Cluster Manager
In Failover Cluster Manager,-->Services and Applications -->SQL Server
Under "Other Resources" in lower pane, right click SQL Server --> Properties --> Dependencies tab
Add the SQL Data Drive and the SQL Log Drive back into the list (and'd)
(These were deleted when the old drives were deleted)
Copies (with permissions...xcopy) and Drive Letter Swaps after copy work fine except for DTC
(DTC had to be deleted and re-created)
DTC and SQL Services can be brought back on line using Failover Cluster Manager
May 24, 2012 at 4:06 pm
sbaker-757360 (5/24/2012)
I finished swapping out all of my shared drives on two clusters and was able to run cluster validation without error. 🙂I tested complete network failure as described above and could not produce split-brain.
great, thanks for posting back.
sbaker-757360 (5/24/2012)
The Quorum Disk can easily be changed to point to a new "Raw" 2GB drive using Failover Cluster Manager
Yes it's very easy to change quorum type\disk. The quorum disk only needs to be 200MB, 512MB at most.
Remember you can also present iSCSI storage directly to the VMs via the iSCSI initiator, it's worth investigating and represents a solid storage solution, see my guide for more on this.
-----------------------------------------------------------------------------------------------------------
"Ya can't make an omelette without breaking just a few eggs" 😉
Viewing 2 posts - 16 through 16 (of 16 total)
You must be logged in to reply to this topic. Login to reply