[comp.unix.admin] NQS vs MDQS

bernhold@red8 (David E. Bernholdt) (06/18/91)

In article <1991Jun16.174703.18419@ms.uky.edu> sean@ms.uky.edu (Sean Casey) writes:
>I wonder how it compares with MDQS?

My personal opinions...

We used MDQS here for some time before we got NQS.  My perception as
maintainer and user of it on our system is that it was primarily
designed to support printer queuing on a very inhomogeneous network,
with a fair variety of different output devices.  The batch support is
very basic and seems to have been almost an afterthought.

As provided by BRL, the batch server has no concept of resources
limits and does not deal very well with cases where the user wants to
abort a running job -- we generally found that you had to signal the
process to kill it rather than using the appropriate MDQS commands to
manipulate the queues.

Network clients are provided, but as we never used them, I can't
remark on them.

On the other hand, the documentation is very good and the program
seems well structured.  Anyone with the inclination and time should be
able to turn MDQS into something functionally equavalent to the modern
NQS as used by Cray, etc. Since the MDQS documentation lists an email
address for questions to the BRL support staff (without promising any
real support, of course), I asked if they'd be interested in having an
improved batch server if someone ever got around to working on it.
They basically said no, they didn't need such a time, and wouldn't
support it for their users.

We're now running NQS as available from COSMIC.  It seems much more
appropriate to being a batch queueing system.  Since I'm not the one
who brought it up, I can't comment on the code or documentation.  I do
know that it took a little porting (MDQS went up almost immediately).
It would be much nicer to have all the capabilities that Cray and
others have added, but even the COSMIC version offers resource limits,
etc. to be configured for whatever machines support them.
-- 
David Bernholdt			bernhold@qtp.ufl.edu
Quantum Theory Project		bernhold@ufpine.bitnet
University of Florida
Gainesville, FL  32611		904/392 6365