Strange one this, but simple to solve.

During a scripted installation (part PowerShell, part standard SQL configuration file) of SQL Server 2012 in a cluster configuration (initial cluster install in this case) the installation would halt (annoyingly silently) before after the pre-flight checks and cluster validation checks.

A review of the summary.txt (as always found in the bootstrap folder) revealed an error:

“The volume that contains SQL Server data directory L:dblogs does not belong to the cluster group.”

A careful check of the cluster storage (including a running of the cluster validation tool) and Windows presentation told me that the storage was presented, available and correctly mounted and assigned a drive letter. All appeared to be well.

A second attempt was made, alas. No joy. In the time honoured fashion, a reboot of all the cluster nodes was tired. No dice.

Frustrated, I was thinking about tearing down the cluster (I have seen oddities where clusters just don’t come up quite right and cluster resources incorrectly report when queried in certain ways) when I noticed something slightly out of the ordinary.

The name of the cluster resource (in this case “logs”) had a capital letter – unusual, as I am known for my disregard of capitals(!), a quick change in CLUADMIN and, PRESTO! the install ran fine.

I’m not sure if this is a definative, known requirement, but it caught me unawares.

more to follow…