[Search for users] [Overall Top Noters] [List of all Conferences] [Download this site]

Conference spezko::cluster

Title:	+ OpenVMS Clusters - The best clusters in the world! +
Notice:	This conference is COMPANY CONFIDENTIAL. See #1.3
Moderator:	PROXY::MOORE

Created:	Fri Aug 26 1988
Last Modified:	Fri Jun 06 1997
Last Successful Update:	Fri Jun 06 1997
Number of topics:	5320
Total number of notes:	23384

5267.0. "Allocation class question" by BIGCHZ::EZZELL (Mike Ezzell) Tue Mar 25 1997 17:26

I was asked to look at a cluster where they have been having problems with
maintaining cluster under certain failover conditions.  It looks to me like
there is a problem with the way the cluster was setup with allocation classes,
but I am a little rusty on how alloctaion classes work.  

The cluster is made up of two VAX 4105A and a number of SCSI disk connected to 
the DSSI bus via 6 HSD05s.  One VAX has an allocation class of "1" and the
other VAX has an allocation class of "2".  Three of the HSD05s have an 
allocation class of "1" and the other three have an allocation class of "2".

Is the VAX with allocation class "2" serving the disk with allocation class "2"
to the VAX with allocation class "1" and vice versa?  The quorum disk is on a 
disk with allocation class "2".  Can the VAX with allocation class "1" see the
quorum disk when the other VAX is not running?

T.R Title User Personal
Name Date Lines

5267.1 Please See 5265.* XDELTA::HOFFMAN Steve, OpenVMS Engineering Wed Mar 26 1997 09:13 15

T.R	Title	User	Personal Name	Date	Lines
5267.1	Please See 5265.*	XDELTA::HOFFMAN	Steve, OpenVMS Engineering	`Wed Mar 26 1997 09:13`	15
	Please see 5265.* for a similar discussion, and for key information around the configuration of unique unit numbers within a particular disk or tape allocation class value. In most situations, I'd say this VMScluster was mis-configured -- nodes that share one or more storage or VMScluster interconnects should be in the same non-zero allocation class. Nodes on the DSSI or CI that have direct access to DSSI ISEs or CI HSx controllers in another allocation class can access these disks via the direct path. They cannot, however, serve these disks through to other nodes. (Only those disks and tapes in the same non-zero allocation class as the OpenVMS host can be served through to other VMScluster members via the OpenVMS MSCP and TMSCP servers.)

   Please see 5265.* for a similar discussion, and for key information
   around the configuration of unique unit numbers within a particular
   disk or tape allocation class value.

   In most situations, I'd say this VMScluster was mis-configured -- nodes
   that share one or more storage or VMScluster interconnects should be in
   the same non-zero allocation class.

   Nodes on the DSSI or CI that have direct access to DSSI ISEs or CI HSx
   controllers in another allocation class can access these disks via the
   direct  path.  They cannot, however, serve these disks through to other
   nodes.  (Only those disks and tapes in the same non-zero allocation
   class as the OpenVMS host can be served through to other VMScluster
   members via the OpenVMS MSCP and TMSCP servers.)