Using the same SQL Server clustering example (two-node cluster of COLTST1 and COLTST3), if any failure is caused by a hardware problem on one of the nodes (for example, if COLTST1 has a bad SCSI card) then you will want to use the following simple procedures to resolve this issue:
1. | After COLTST1 (the active node) fails, SQL Server 2000 should automatically fail-over to COLTST3 (the passive node). |
2. | Once you have determined that there is a hardware problem on COLTST1, you should run SQL Server Setup and remove COLTST1 from the SQL Cluster. |
3. | From Cluster Administrator, “evict” COLTST1 from MSCS (the cluster group). |
4. | Now, you can install new hardware to replace the failed hardware on COLTST1 and bring it back up. |
5. | Run MSCS and join (rejoin) the existing cluster. |
6. | Run the SQL Server Setup on COLTST3 and add COLTST1 back to the SQL cluster (fail-over cluster). |
You have done all of these tasks before when you did the initial installation of SQL clustering and MSCS (that we stepped through in Chapter 4 and earlier in this chapter), except for evicting a node from the cluster. There is nothing difficult about evicting a node, though. It simply removes this node from the owner's list of the cluster allowing you to do whatever you need to fix the problem without affecting the availability of your application.
3.137.163.208