home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!stanford.edu!rutgers!cmcl2!acf3!tihor
- From: tihor@acf3.NYU.EDU (Stephen Tihor)
- Newsgroups: comp.os.vms
- Subject: Re: 2 node LAVC "problem"!
- Message-ID: <61050093@acf3.NYU.EDU>
- Date: 23 Nov 92 16:34:00 GMT
- References: <17017@umd5.umd.edu>
- Sender: notes@cmcl2.nyu.edu (Notes Person)
- Organization: New York University
- Lines: 10
- Nntp-Posting-Host: acf3.nyu.edu
-
- The traditional dirty trick in this case =--=which works nly because
- the two systems share NO HARDWARE is to create a quorum disk on each
- opne with different names and 1 vote. When both systems are up the quorum
- disk name difference invalidates the Qdisk and quorum is the 2 CPU votes.
- When you lose connect (the axe murder takes out the ethernet either or both systems
- can manage to survive, they only see tyhe CPU and 1 qdisk so thats 2 votes
- and presto. Note that there is still the issue of where you
- put common queues, etc so that they survive. Absetent a multiheaded SCSI control
- you are still looking at one node losing functionality or needing to
- run a separate deamon to fail out some functions.
-