Batch systems for DS5500 / Ultrix -- summary of responses.

[Paul Leyland]

About a month ago, I posted an article asking for information about
batch queueing systems to run under Ultrix on a DEC 5500.  The following
people responded:

	Keith Refson <keith at earth.ox.ac.uk>
	"Richard A. Lethin" <lethin at ai.mit.edu>
	"Takeuchi T. OS2 U1" <takeuchi at osku01.enet.dec.com>
	gwyn at smoke.brl.mil (Doug Gwyn)
	jiml at edu.wisc.slh (James E. Leinweber)

Thanks guys!

I explained that I already knew about the system-supplied at(1)
(almost unbelievably primitive), SUPERNAT, a commercial (read: REAL
MONEY) system from SO Systems and Condor (a distributed batch system,
best used to soak up idle cycles).  The latter is available from
cs.wisc.edu and has been described previously.

Three other systems have been described to me -- NQS, MDQS and batchd.

Here's what was said about NQS:  (from keith at earth.ox.ac.uk)

There is, of course NQS, NASA's distributed batch system upon which
> the convex batch is based.  It, or enhancements of it appear to be on
> the way to becoming a de-facto standard for unix systems.  It is used
> on crays and vaxes and we use it on our Stardent systems.  It does
> lack some functionality -- the queues are first-in first-out and there
> is no means I have ever found of re-ordering a queue.  However there
> is a commercial concern who are marketing and supporting a
> (presumably enhanced) version.

lethin at ai.mit.edu also mentioned NQS.

As for MDQS, two people (takeuchi at osku01.enet.dec.com and
gwyn at smoke.brl.mil) replied:

> Try picking up arch/mdqs.tar.Z by anonymous FTP from FTP.BRL.MIL.

and:

> There is MDQS(Multi Device Queuing System) on gatekeeper.dec.com
> probably
>  
> 	gatekeeper.dec.com:pub/net/mdqs.tar.Z
>  
> This is free and source include I think.


Finally, jiml at edu.wisc.slh, told me about batchd.  He gave the
following brief description and also included the manual page, which I
won't quote here because of its length.

> We run a heavily modified version of a batch job queuing system
> obtained off usenet from U. of Waterloo in 1985.  It supports different
> queues with different policies, controllable by time of day and day of
> week, with uncaught output mailed back to the submitting user.  It was
> developed under 4.2 BSD on a vax, and we are running it under Ultrix
> 4.1 on a vaxstation, but we haven't tried compiling it on Ultrix or a
> DECstation yet.  I expect the porting effort would be minimal, and the
> price is right - free.  Let me know if you would like more
> information.  I'm appending a copy of the man page we wrote for the
> daemon.

Once again, thanks to all who replied.  I'm tending towards NQS,
partly because of local usage in Oxford, but I will ftp over the
others for deeper investigation.

Paul Leyland.