CL380 CR3500 problems

[Brett Tesdall]

I tend to agree that the symptoms point to a SCSI ID conflict in the nodes, however, I can't find a conflict. The document you pointed me to says that as long as the SCSI ID's are set in one node to 6 and the other to 7, then SCSI conflicts are impossible, yet it appears to be happening.

I went into the System Setup of each node in the CL380 and looked at the settings for the Ultra2 SCSI controller that attaches to the CR3500. I have transcribed what is on each:

Node 1:

Slot 2 - Compaq 64-Bit Dual Channel Wide Ultra2 SCSI Controller (Port 1)
SCSI Resources: ENABLED
Controller Order: Third
Host Adapter Number: 1
SCSI ID: #6
Bus Speed: Auto-detection

Attached Devices:
SCSI ID 2.....Compaq CR3500 Reserved Device
SCSI ID 3.....Compaq CR3500 Reserved Device

Slot 2 - Compaq 64-Bit Dual Channel Wide Ultra2 SCSI Controller (Port 2)
SCSI Resources: ENABLED
Controller Order: Fourth
Host Adapter Number: 2
SCSI ID: #6
Bus Speed: Auto-detection


Node 2:

Slot 2 - Compaq 64-Bit Dual Channel Wide Ultra2 SCSI Controller (Port 1)
SCSI Resources: ENABLED
Controller Order: Third
Host Adapter Number: 0
SCSI ID: #7
Bus Speed: Auto-detection

Attached devices
SCSI ID 2.....Compaq CR3500 Reserved Device
SCSI ID 3.....Compaq CR3500 Reserved Device

Slot 2 - Compaq 64-Bit Dual Channel Wide Ultra2 SCSI Controller (Port 1)
SCSI Resources: ENABLED
Controller Order: Fourth
Host Adapter Number: 1
SCSI ID: #7
Bus Speed: Auto-detection




If there is a problem somewhere, it would seem that it must be here somewhere, but I can't see where the problem lies.

Could the CR3500 be defective or faulty somehow?

[Brett Tesdall]

I may have found the problem.

I reinstalled the driver for the Ultra2 SCSI card using the driver on the CL380 software page. Up to this point it has been using the MIcrosoft-supplied driver that was automatically installed. Once I replaced the driver with the Compaq one, the conflicts disappeared and I was able to access the CR3500 with the Array Config Utility from both nodes at the same time, which up to this point was not possible--one of the nodes would be knocked off by the other.

The issue I have now is some strange drive mappings on the shared storage array. However, now that I have potentially found the problem, I'm going to blow away the shared storage logical drives and re-install with Windows Server 2003 Standard, without the clustering, and see if I can achieve the desired configuration we want with this server.

I am definitely a step closer! Will assign points if/when I can get this working properly--probably tomorrow or Monday. Otherwise, I'll be back...

[Brett Tesdall]

I tried the Compaq drivers and that did allow both machines to access the shared storage array. However, I could not get the two machines to not fight over the logical drives I created. If I wrote a file to the same logical drive from both nodes, each node would only be able to see the files it created, and when a node was rebooted, the machine would think the drive was corrupt and rebuild the file tables and only show the files the other node wrote. I tried to install Windows 2003 Standard and Enterprise Edition with Clustering, and could not get it to properly cluster. When I tried it with a fresh install of Enterprise and tried to make the cluster, the Quorum drive on the first node became corrupt and unusable, thus preventing it from forming a proper cluster.

Thanks for all the help, but I don't know what else to try on this thing, except maybe backing back down to Windows 2000 Advanced Server, although we really wanted to use 2003 Server on this. I've assigned points. Thanks.

[Brett Tesdall]

Was able to resolve the SCSI conflicts, but could not get it to properly cluster with Windows 2003 Server, Standard or Enterprise Edition.

[Jeff Vincent]

I just found this thread after recently starting something very similar. We also had a CL380 sitting around and want to use it as a central file repository.

I installed SUSE 10.0 on node 1 and got it working. It can see the shared logical drive. If I power up node 2, it deactivates.

I installed Win2K3 on node 2 with the latest drivers to see if running a supported OS would help. It can also see the logical drive as long as node 1 is off. If I turn on node 1, it hangs loading the SCSI driver with continuous "Bus Reset" messages until I power off node 2.

Can I ask how you solved the SCSI ID conflict? I'm sure I am missing an important piece of the puzzle, I am just not familiar with clustering or this hardware. I did what you tried, by setting the SCSI ID of the Node 1 controllers to #6 and the ID for node 2 controllers to #7, but still have the same problems.

Thanks for any additional information you can provide.