0

I am trying to setup a new sql server 2008 cluster, on windows 2012R2, and the installer is failing on the cluster shaed disk availability check. I have verified that there are 5 disks assigned to "Available Storage" when viewed in the fail over cluster manager.

Some background, this is my second attempt to install sql server on this cluster. The first time, the cluster object was unable to create the new computer object during the cluster installation. This caused the installer to fail do to lack of permissions. I have since resolved this, and have run "Remove node from cluster" to uninstall sql server from the node. I am now trying again to run the installer.

About the environment, OS: windows 2012R2 SQL Version/Edition: 2008/Enterprise I am running the installer from the current cluster host(node1), and all storage is owned by node1. This includes the quorum, as well as the 5 disks assigned to the available storage group. Both cluster nodes are up and available, and accessible either through the node names, or through the windows cluster name.

There are no cluster validation warnings that I know of, but I have asked the windows admin to rerun the validation tool to confirm that, that is still the case. SQL 2008 is required by the front end application (I pushed for at least 2012, but was told it was a no go)

  • Does the Cluster Validation check come back clean? Does it report all of your storage is in good health, accessible, and configured? Also, you are on a very old version of Sql Server, any reason for not installing a newer version? – rvsc48 Apr 2 '18 at 14:52
  • I reran the validation report and I saw that there are a few warnings, but nothing that failed. The only warnings that worry me, is there are some warnings related to the sql server instance name still in the report, but I don't see any entries related to the cluster role in fail over cluster manager – Patrick Apr 2 '18 at 16:46
  • This is very hard to help troubleshoot without actually seeing your situation. Were you able to make any additional progress with this or are you still stuck? – rvsc48 Apr 3 '18 at 14:26
  • I have not made any additional progress. I did rerun the validation report, the only disk related warnings are "Successfully issued call to Persistent Reservation REGISTER using Invalid RESERVATION KEY 0x10000000c, SERVICE ACTION RESERVATION KEY 0x10000000d, for Test Disk 1 from node XXX" Its almost like some part of the first sql install is still lingering and hasn't completely let go of the storage – Patrick Apr 3 '18 at 14:35
  • I would remove Sql Server from the 2 nodes completely, leaving only the Windows cluster installed at the OS level. Then, re-run the cluster validation again, which will again check the storage, to help narrow down the exact issue with the storage configuration. – rvsc48 Apr 3 '18 at 15:17
0

Well, it looks like when the first sql server install ran, it also configured the file server services, which were still trying to hold the disk and the network name. Odd thing is, the disks where all showing 0 dependencies. This probably could have been fixed if we opened a case with microsoft, maybe cleaned up some registry entries, etc, but the guys from out windows team wanted to just rebuild from bare metal, as it would be quicker and easier. I was on board with that, so that is the solution I am implementing

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.