Article ID: 175275
Article Last Modified on 11/1/2006
APPLIES TO
- Microsoft Windows NT Server 4.0 Enterprise Edition
- Microsoft Cluster Server 1.1
This article was previously published under Q175275
SUMMARY
If one of the controllers on a shared SCSI bus fails while using Microsoft Cluster Server, simply power off that node and replace the controller with one of the same make, model, and firmware version.
MORE INFORMATION
There are many ways a SCSI controller can fail. Typical signs of imminent SCSI controller failure can be seen by monitoring the System Event Log for Event ID: 11 and Event ID: 9.
For additional information on these events, please see the following articles in the Microsoft Knowledge Base:
ARTICLE-ID: 153128
TITLE : Event IDs 9 and 11: SCSI Controller/Device Errors
ARTICLE-ID: 154690
TITLE : How to troubleshoot event ID 9, event ID 11, and event ID 15 error messages
When you replace the failed SCSI controller, make sure the new controller has the same firmware as the working controller in the other node.
Follow these steps to replace the failed or failing SCSI controller:
- Power off the node with failed or failing adapter. If possible, manually move all groups to another node using Cluster Administrator.
- Verify that all groups and resources are online on the functioning node.
- Remove and replace SCSI controller in failed node.
- Power on the node and log on normally.
- Check the event log for any unusual errors.
- Start Cluster Administrator and verify groups and resources are online and have failed back if so configured.
NOTE: After replacing the controller, set the SCSI ID for the new controller to be the same as the old one. Microsoft Cluster Server should start and function normally. If your configuration does not use SCSI Y cables or specialized cables, you may have to shut down both nodes. SCSI Y cables allow termination of the bus regardless of connection to the controller. With standard SCSI cables, termination may be lost when you disconnect the faulty controller.
You may have successfully repaired the problem if:
- There are no unusual event messages in the event viewer.
- Cluster Administrator starts and allows you to view the cluster.
- Groups and resources may be moved successfully to the node with the new controller.
Additional query words: MSCS server cluster
Keywords: kbhowto kbsetup KB175275