The following scenario was seen at a Sun Cluster 3.0 Sun Fire[TM], HDS 99x0 campus cluster installation.

Problems observed: pkgadd VRTSvxvm could fail, vxconfigd could hang, and excessive boot times could occur with volume manager enabled. At the same time, the following errors would appear in /var/adm/messages:

Aug 6 12:23:03 nodename REPORT LUN to D_ID=0x110400 lun=0x0 failed: sense key=Illegal Request, ASC=25, ASCQ=0. Giving up
Aug 6 12:23:13 nodename scsi: [ID 243001 kern.warning] WARNING: /ssm@0,0/pci@18,600000/pci@1/SUNW,qlc@4/fp@0,0 (fcp0): Aug 6 12:23:13 brock0 REPORT LUN to D_ID=0x110600 lun=0x0 failed: sense key=Illegal Request, ASC=25, ASCQ=0. Giving up      

The definition of ASC 25 / ASCQ=0 code is:

	Logical Unit Not Supported      

The addressed logical unit is currently not configured. An Add Logical Unit operation to the Logical Array Mode Page must be executed to define and configure the logical unit before it is accessible.

This error in this configuration was due to the Hitachi HDS 99xx array controller that was connected to the system. The controller had luns that were unmapped and were not present anymore.

Disconnecting the controller that has no LDEVs mapped to it makes the ASC=25 errors and Veritas Volume Manager problems disappear.

Do not hook up HDS 99x0 ports unless there are LDEVs mapped to them. Even though this problem was observed on Sun Cluster 3.0 nodes, it is not a cluster-related issue and can be reproduced in a non-clustered environment.

The root cause of it is that VxVM sees the controller trying to find the devices on it, but they are not present. Sun StorEdge[TM] T3 arrays do not seem to be vulnerable to this.

Keywords: asc=25, asc, failed