Path: utzoo!mnetor!uunet!husc6!ncar!ames!mailrus!nrl-cmf!ukma!uflorida!beach.cis.ufl.edu!jdc
From: jdc@beach.cis.ufl.edu (Jeff Capehart)
Newsgroups: comp.os.vms
Subject: Re: Getting Information about Queues
Message-ID: <13556@uflorida.cis.ufl.EDU>
Date: 25 Mar 88 05:00:44 GMT
Sender: news@uflorida.cis.ufl.EDU
Reply-To: jdc@beach.cis.ufl.edu (Jeff Capehart)
Organization: UF CIS Department
Lines: 27
Keywords: record, lock, cluster, slow

I have a couple questions:   First, I am doing some shared file
access in Pascal using the put, get, find, and update procedures.
My problem is that if the record happens to be locked, I get a
huge system dump with an error that says....
error during GET operation, record currently locked by another user.
(filename, etc. follows, along with stack dump)
How can I both test and set the lock on the record so that I don't
run into this problem?  (Ah, the Philosopher's problem revisited!)
I am sure it is in a manual somewhere, but us peons don't have easy
access to those wonderful orange manuals.

Second question:   Why do VAXen all of a sudden get a lobotomy when
they are clustered?  This question may have been asked before, but
I forgot what the conclusion was.  We have a couple 780's that could
easily handle 80 users under vms 3.x and they slowed down a bit under
version 4.x but when they were clustered, they run incredibly slow
with only 20 users on.  A 750 will run faster than a clustered 780!!
Now, why is this so?  My only idea is that somehow the CI must take
over total control, and the processor dedicates its priorities to
making sure that the CI is happy.  Also VMS gets slower with each new
upgrade so we have to have a corresponding hardware upgrade to make
VMS at least resonably good with response time.


--
Jeff Capehart 		Internet: jdc@beach.cis.ufl.edu
University of Florida	UUCP:   ..!ihnp4!codas!uflorida!beach.cis.ufl.edu!jdc