Gregory, If you haven't already, be sure to look at the appropriate hardware administration manual for the type of storage your cluster has. The Sun Cluster hardware manuals are located here:
http://docs.sun.com/app/docs/coll/1024.2 For example, if you are using SCSI JBOD storage, you can look at the chapter "Maintaining Storage Arrays" for procedures to add, remove, and replace individual disk drives or entire arrays: http://docs.sun.com/app/docs/doc/819-2995/installationtemplate-66?a=view There is also a procedure in the Sun Cluster System Administration Guide, "How to Remove Connectivity Between an Array and a Single Node, in a Cluster With Greater Than Two-Node Connectivity", that might or might not apply to your situation: http://docs.sun.com/app/docs/doc/820-4679/babgecdb?a=view FYI, the ha-clusters-discuss alias is targeted at users of the open-source version of Sun Cluster software. There is another forum for users of the commercial Sun Cluster software at http://forums.sun.com/forum.jspa?forumID=842. The archive of that Clustering forum would be another resource for questions about using Sun Cluster software. Several Sun experts on Sun Cluster software monitor both forums. HTH. Lisa Shepherd Sun Cluster Technical Publications "We're the M in RTFM" On 07/24/09 02:04, G Gabriel wrote: > Hi all, > > I have a 6 node cluster running 5.9 Generic_122300-37 (SPARC) and with Sun > Cluster 3.2. I have to remove over 600 LUNs on this cluster and am somewhat > anxious in how to approach this. > > The recommendations I received were the following: > > - remove the disks completely out of veritas (VxVM 5.0) > - remove SCSI keys from the disks using /usr/cluster/sc/lib/scsi -c scrub -d > /dev/dsk/... > - unmap them on the storage side > - perform cleanup on nodes > > The scrub part is where I get anxious; as far as I know, the cluster puts > SCSI-3 keys on each disk which you can list with scsi -c inkeys. > However, I also noticed that the majority of those disks have a reservation > active (scsi -c inresv) > > My questions here are: > > - Is my understanding of inkeys (list keys) and inresv (list active > reservations) correct? > - what will happen if I issue scsi -c scrub on a disk which lists an active > reservation (scsi -c inresv); as far as I can gather from training and > reading, this might induce a node panic. > - I also stumbled in my reading on this upon an option called scsi -c > disfailfast; will this disactivate the probing if a node can access a disk or > is this something different? > - finally: is there a safer way to remove LUNs from a 2+ node cluster? > > I'd be most grateful if someone with experience and more expertise could shed > some light on this. > > Gregory >
