Original qestion:
I'm looking for a free product to manage queue batch on systems like
NQS does on big systems.
Thanks to"
Bob Shaw bshaw@bobasun.spdc.ti.com
Andrew Moffat amof@SubaruSparcDev.subaru1.com
Gyorgy Simon simon@tigem.it
Mike Salehi mrs@cadem.mc.xerox.com
Mattias Zhabinskiy mattias@txc.com
David Fetrow fetrow@biostat.washington.edu
Robert Lopez Robert.Lopez@abq.sc.philips.com
Robin Landis robin.landis@imail.exim.gov
Response:
There is in fact GNQS from the GNU group. Below the global description
"Generic NQS is a network queuing system for spreading batch jobs across a
network of machines. It is designed to be simple to install on a
heterogeneous network of machines, and has optimizations for running on
the high end, symmetric multiprocessing servers that are currently on the
market. It is available for many more UNIX variants than any other
comparable product, and inter-operates with other NQS systems, including
Cray's NQE."
The Generic NQS web site is: http://www.shef.ac.uk/~nqs/
Other products is DQS (Distributed Queueing System ) at :
http://www.scri.fsu.edu/~pasko/dqs.html
"The Distributed Queueing System is designed as a management tool to aid in
computational resource distribution across a network. DQS provides architecture
transparency for both users and administration across a heterogeneous
environment, allowing for seamless interaction for multiple architectures.
Highly mutable custom site configurationsare possible under DQS.
This abilty to customize DQS leads to effective resource distribution and
increased network throughput."
Similar product is Condor a package which lets you stop a program, save its
state to disk, run it from where it left off on the same (or different) machine.
It includes load balancing. But you need to recompile your program with the
condor library.
" The goal of the Condor project is to develop, implement, deploy, and evaluate
mechanisms and policies that support High Throughput Computing (HTC) on large
collections of distributively owned computing resources. Guided by both the
technological and sociological challenges of such a computing environment, the
Condor Team has been building software tools that enable scientists and
engineers to increase their computing throughput."
Adress: http://www.cs.wisc.edu/condor
Bob Shaw points to LoadBalancer from unison software. (http://www.unison.com ).
He says that it's *very* easy to install/configure and use and takes very
little admin time.
Mattias Zhabinskiy give me three good adresses:
a) For Queueing and Scheduling Page
http://hepwww.ph.man.ac.uk/~mcnab/QueShed
This is a global homepage entry where the most importants products
on this topic are related
b) Clustering Computing Review:
http://www.npac.syr.edu/techreports/hypertext/sccs-748/cluster-review.html#RTFToC123
c) DQS (see above)
Andrew Moffat speak about the BATCH scheduler class from Sun at
http://opcom.sun.ca/special_info.html (the opcom page).
The Opcom Software Specials (http://opcom.sun.ca/specials.html) are
special products on differents subjects.
Here is the description of the BaTch and Fixed Scheduling Classes (BT-FX)
"The BaTch and FiXed scheduling classes are loadable scheduling classes for
Solaris 2.x. The BaTch Class (BT) allows you to make optimal use of unused
cycles or idle time, without impacting normal timesharing activities.
This allows the scheduling of computing intensive applications to be run
in the "true" background, while retaining normal performance characteristics
on foreground activities. The FiXed Class (FX) ensures that applications get
a higher percentage of CPU time than normal timesharing jobs. It has the added
advantage of giving some of the processing time benefits of a real-time
scheduling class, without the inherent dangers and problems of this type of
class (like preempting normal Kernel activities). BT and FX may be used
alone or together."
Robin Landis talk me about an article by Hal Stern titled:"It Slices, It Dices:"(Learn How to Carve Your Own Time Slices or Dice Up the SUNOS 5.x Dispatch
Table into Schedule Classes. Unfortunately he has lost the adress of the
internet link. He tried to send me a copy by regular mail but yet I havn't
received it.
Many thanks to all people.
+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
+ Rene OCCELLI +
+ I.U.S.T.I. C.N.R.S. U.M.R. 6595 +
+ Technopole de Chateau Gombert +
+ 5 Rue Enrico FERMI +
+ 13453 MARSEILLE Cedex 13 France +
+ Tel: (33)04 91 10 69 37 04 91 10 69 38 +
+ Fax: (33)04 91 10 69 69 +
+ Email: rene@iusti.univ-mrs.fr +
+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++