🚀 go-pugleaf

[Just kidding about the FAQ; yet again from COSMIC's official page.....]
[I do not know if international distribution of NQS is permitted.]

M88-10054 Sterling
NOS- NETWORK OUEUEING SYSTEM

B. KINGSBURY C-LANGUAGE Approximately 75,000 source
statements  9 Track 1600 BPI UNIX CPIO Format Magnetic
Tape
UNIX OPERATING SYSTEM
ARC-11750 Price: Program $6,000.00/Documentation $21.00

The Network Queueing System, NQS, provides batch and device
queueing facilities for various computers comprising a networked
UNIX environment. With the UNIX operating system as a common
interface, a user can invoke the NQS collection of user-space
programs to move freely around the different computer hardware
tied into a network. NQS provides facilities for remote queueing,
request routing, remote status, queue access controls, batch
request resource quota limits, and remote output return. NQS was
developed as part of an effort aimed at tying together diverse
UNIX based machines into NASA's Numerical Aerodynamic
Simulator Processing System Network. The NQS architecture was
written with the following design goals: 1) provide full support for
both batch and device requests, 2) support all of the resource
quotas enforceable by the underlying UNIX kernel implementation
that are relevant to any particular batch request and its
corresponding batch queue, 3) support remote queueing and routing
of batch and device requests throughout the NQS network, 4)
modularize request scheduling algorithms so that schedulers can
be easily modified at individual installations, S) support queue
access restrictions through user and group access lists for all
queues, 6) enable networked output return of both output and
error files to possibly remote machines, 7) allow mapping of
accounts across machine boundaries, 8) provide friendly
configuration modification mechanisms for each installation, 9)
support status operations across the network, without requiring a
user to log in on remote target machines, and 10) provide for file
staging, or copying of files for movement to the actual execution
rnachine. There are three types of queues supported by NQS.
Batch queues do not require a specific device, that is the batch
request can be routed around the network with no detrimental
effects. Device queues are for requests which require the direct
services of a specific device, such as a line printer. A pipe queue
exists to transport requests to other batch, device, or pipe queues
at possible remote machine destinations. The pipe queue
mechanism is responsible for routing and delivering requests to
other queues in spite of machine failures, lack of owner
authorization, insufficient space, disabled queues, and other
rejections. All NQS network conversations are performed using
the Berkeley socket mechanism as ported into the respective
vendor kernels. Computer networks which have successfully
implemented NOS include DEC VAX, Silicon Graphics IRIS, Amdahl
5840 mainframes and CRAY-2 machines. NQS is written in the C
language for interactive execution and has been implemented on
various computers operating under UNIX. NQS has a central
memory requirement of approximately 160K of 8 bit bytes on a
DEC VAX computer. The NQS daemon must reside on every
computer in the network. NQS requires the nmap utility, which is
supplied as part of the NQS package. This program was developed
in 1986.

In article <1991Jan4.181847.24382@nas.nasa.gov> eugene@wilbur.nas.nasa.gov (Eugene N. Miya) writes:
>[Just kidding about the FAQ; yet again from COSMIC's official page.....]
>[I do not know if international distribution of NQS is permitted.]
>
>M88-10054 Sterling
>NOS- NETWORK OUEUEING SYSTEM

[[sic]]

>
>B. KINGSBURY C-LANGUAGE Approximately 75,000 source
>statements  9 Track 1600 BPI UNIX CPIO Format Magnetic
>Tape
>UNIX OPERATING SYSTEM
>ARC-11750 Price: Program $6,000.00/Documentation $21.00
>
>The Network Queueing System, NQS, provides batch and device
>queueing facilities for various computers comprising a networked
>UNIX environment.
   .
   .
 [[etc.]]
   .


Thank you for your excellent write-up on NQS.  I have been using and teaching
NQS usage on Cray Research machines for over three years, yet I learned
a great deal about the background and organization of NQS that I had only
seen in fragments before.  However, the following is somewhat incomplete:

>Computer networks which have successfully
>implemented NOS include DEC VAX, Silicon Graphics IRIS, Amdahl
>5840 mainframes and CRAY-2 machines.

NQS can run on any CRI machine running the UNICOS(tm), CRI's UNIX(tm)-based
OS.  This includes (potentially) all of the CRAY X-MP and CRAY Y-MP series,
and many of the remaining CRAY-1 computers, as well as the CRAY-2.  Of
course many Cray-1's and X-MP's in the field (and a handful of Y-MP's) do
not run UNICOS, and hence cannot run NQS.  There are also a handful of
Cray sites that run UNICOS but choose not to run NQS.

CRI also supports a subset of NQS called RQS, whose purpose is to pass
batch jobs from a smaller UNIX box to a CRI machine.  For example, I have
used RQS to submit from a Sun-3/60 to a CRAY.  I am not sure what other
UNIX-based systems RQS will support.


---------------------------------------------------------------------------
Roger Glover         ! Disclaimer:  "Nope.  No way.  Impossible.  I couldn't
Software Instructor  !  have said that."
Cray Research, Inc.  ! "Like neutrinos through the salt mine, so are the
		     !  NANO-SECONDS OF OUR CLOCK PULSE"

In article <012352.8054@timbuk.cray.com> glover@sequoia.cray.com (Roger Glover) writes:
>
>
>In article <1991Jan4.181847.24382@nas.nasa.gov> eugene@wilbur.nas.nasa.gov (Eugene N. Miya) writes:
>>
>>M88-10054 Sterling
>>NOS- NETWORK OUEUEING SYSTEM
>

  While on the subject of NQS - can somebody out there help me?

     I have two problems so far

	a) WHen I start a substantially large process, NQS kills the job
	   saying "not enough memory". Obviously the process is hitting
	   some memory limits imposed by NQS. I want to circumvent it.
	   (Yes, I know about the options one can specify to alter
	   limits)
        b) I am unable to queue a job from one machine onto another.
	   i.e. remote submission does not seem to work.

   THanks

   renu raman
--
--------------------------------
   Renukanthan Raman				ARPA:ram@sun.com
   M/S 18-412, 2500 Garcia Avenue,              TEL :415-336-1813
   Sun Microsystems, Mt. View,  CA 94043

In article <012352.8054@timbuk.cray.com> glover@sequoia.cray.com (Roger Glover) writes:
>CRI also supports a subset of NQS called RQS, whose purpose is to pass
>batch jobs from a smaller UNIX box to a CRI machine.  For example, I have
>used RQS to submit from a Sun-3/60 to a CRAY.  I am not sure what other
>UNIX-based systems RQS will support.

What I would like to be able to do is the following:

Say I have an account on a remote supercomputer which runs NQS.  I
want to be able to submit a job on my local workstation (also running
NQS or similar) and have it transferred to the remote machine to run
under my account there and when its done have the log file transferred
back here It would also be nice to be able to 'qstat' 'remote's
queues.

The problem is that 'remote' has thousands of users scattered across
the country, and the management of the machine isn't about to
configure every user's local system into the NQS setup, quite
understandably, I think.

Is there/will there be a mechanism by which this would be possible?
As far as I can tell, even CRI's NQS requires that every machine it is
willing to speak to be defined (via qmapmgr(1M) ).  How about a more
"casual" way of submitting jobs?  The verification method for remote
jobs is just using .rhosts, so why require that all network hosts be
listed?  Why not accept my connection as long as I can identify myself
sufficiently using the same method?

And to bring the the message I referenced, how does something like RQS
fit into this?  I've never heard of it & the Unicos 5.1 sites I have
access to don't have a man page or anything (there is an NQS man
page).

Thanks for info and ideas.
-- 
David Bernholdt			bernhold@qtp.ufl.edu
Quantum Theory Project		bernhold@ufpine.bitnet
University of Florida
Gainesville, FL  32611		904/392 6365

In article <012352.8054@timbuk.cray.com> glover@sequoia.cray.com
(Roger Glover) writes:
>Thank you for your excellent write-up on NQS.

Actually, it is not my write up.  That is the 'approved' write-up from
the COSMIC catalog  (NQS and CLIPS are the most asked for pieces
of software from the net).  Inevitability, someone sends numerous
requests for info on it.  And in the days of the FSF and the LPF,
the costs are regarded by some as "ridiculous."

"All I every got was 'Not my bailwick....'"  -- C. Stoll

--e. nobuo miya, NASA Ames Research Center, eugene@orville.nas.nasa.gov
  {uunet,mailrus,other gateways}!ames!eugene