Today I encountered a very strange issue on a test server. The server is Windows 2000 Advanced server SP3 with SQL 2000 Enterprise SP2. The system is configured as a 1 node cluster, with a fibre SAN backend.
Earlier today, I created a new drive mapping on my workstations, the I: drive. Later, I made a remote desktop connection to this server. I attempted to restore a database using QA, but received a message that the I: drive was not available. I was attempting to restore to I:, as it is our main data volume. I logged off, deleted the mapping on my workstation, and logged back in. I was then able to access the I: drive and performed the restore.
Approximately 20 minutes later, I received a call from a user that all databases were inaccessible. I verified that all databases with files located on the I: drive were inaccessible. I then brought up cluster administrator, which showed the entire SQL group online! I had 30 databases inaccessible, but none were marked suspect and the cluster thought that the I: drive was online.
To fix the situation, I took the SQL Server resource offline and brought it back up. Thankfully, all databases were recovered successfully.
Any thoughts? I can't imagine that a mapped drive on a terminal services session should be able to bring down a cluster.