From:	SMTP%"RELAY-INFO-VAX@CRVAX.SRI.COM" 19-DEC-1993 19:40:35.02
To:	EVERHART
CC:	
Subj:	Re: flaky cluster hangs

X-Newsgroups: comp.os.vms
Subject: Re: flaky cluster hangs
Message-ID: <1993Dec19.113102.18621@avatar.montagar.com>
From: davidc@avatar.montagar.com (David L. Cathey)
Date: 19 Dec 93 11:31:02 CDT
Organization: Montagar Software Concepts, Plano TX
Lines: 19
To: Info-VAX@CRVAX.SRI.COM
X-Gateway-Source-Info: USENET

In article <18497735@MVB.SAIC.COM>, EVERHART@arisia.gce.com writes:
> Heard at DECUS (conversation) of situations where cluster hangs had
> been cured by moving the axp vms version to the 1.5-1H1 subrelease.
> Sounds like a bug...

	Yes, there is such a bug.  The symptoms on my cluster were two nodes
would lose connection to each other, resulting in loss of quorum.  The cluster
would partition and never recover on its own (this was a VAX V6.0 / AXP V1.5
cluster).  It seemed to happen during times of very high cluster traffic (like
a cross-cluster BACKUP with ANU NEWS doing things, etc...).

	AXP V1.5-1H1 was installed and has been running wonderfully ever since.

	I have not actually seen it occur on pure AXP clusters, though...

- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - 
David L. Cathey		                |INET: davidc@montagar.com
Montagar Software Concepts              |UUCP: ...!montagar!davidc
P. O. Box 260772, Plano TX  75026-0772  |Fone: (214)-618-2117