Request: Batch system info
Tony Hackenberg
vvawh at convx1.lerc.nasa.gov
Wed Aug 29 14:23:51 AEST 1990
In article <996 at lee.SEAS.UCLA.EDU> scw at seashell.seas.ucla.edu (Stephen C Woods/) writes:
>This doesn't seem to have gotten out when I posted it, sorry if it's a dup.
>
> We are looking for a Batch system to be run on AIX/370 on a 3090. Rather
>than reinvent the wheel (for the umpteenth time) we thought that it would be
>a good idea to poll the net. Please reply directly to me (scw at seas.ucla.edu)
>and in a couple of weeks I'll summarize to the net.
>
> Our needs are as follows (given source we can add any of these):
>
> 1 Graded priorities, jobs of a lesser priority will not be released when
> jobs of a higher priority are running. (exception, see (3) below).
>
> 2 Resource limits (al la BSD limits and or SYSV ulimit) with the extensions
> that processes that have exceeded some Hard limit (at say 110% of
> allocated) will be killed nicely (SIGTERM) then (at say 120%) terminated
> with extreme prejudice (SIGKILL).
>
> 3 Timed release queues, jobs that will be run at specific times (late
> at night, and on weekends), these jobs will be suspended when the time
> that they are supposed to be running in expires.
>system). We are interested in pointers to something that my boss heard
>about called NQS, but please let us know about anything else.
>
CRAY Research, Inc., Sterling Software, Inc., and NASA/Ames Research Center
jointly worked on a product called NQS (Network Queueing System). It is a
batch subsystem for UNICOS (CRAY's operating system for its supercomputers).
Other vendors have picked up NQS and offered it as part of their product
(CONVEX).
A public domain version (or a govt copyrighted version that costs
nothing but needs some work) may be available from NASA/Ames.
Email me at vvawh at convx1.lerc.nasa.gov for contacts.
More information about the Comp.unix.questions
mailing list