|
|
Choosing A Webhost: |
Re: Small problems using Torque + MAUI (start delay, error message, redirec: msg#00190clustering.torque.user
I've added the line 'set server scheduling = true" and the delay has desappeared! Thank you! I have a small cluster, so reducing the poll interval should be not a problem, but it is unnecessary, now. What about the message "'unknown': I need something more specific." when exiting from an interactive session? As an example, this is what happens: -------------------------------- [francesco@epsilon ~]$ qsub -I qsub: waiting for job 1783.epsilon to start qsub: job 1783.epsilon ready [francesco@node6 ~]$ top 'unknown': I need something more specific. [francesco@node6 ~]$ exit logout 'unknown': I need something more specific. qsub: job 1783.epsilon completed [francesco@epsilon ~]$ -------------------------------- I guess it's something related to the terminal, but how can I resove the problem? Francesco Dave Jackson ha scritto: >Francesco, > > Moab/Maui use both polling and event driven interfaces to manage job >scheduling. First off, if your cluster is smaller than 200 nodes, you >should have no problem reducing your poll interval to 10 seconds or >lower. > > Secondly, Moab/Maui can load info using TORQUE's event interface. In >qmgr, make certain that the line 'set server scheduling' is set to true. >This enables TORQUE to send events to the scheduler. When the scheduler >detects this info, it immediately reloads workload info and attempts to >schedule. > > Please let us know if this addresses your issues. > >Dave > >On Mon, 2006-01-16 at 15:51 +0100, Francesco Del Citto wrote: > > >>Dear Torque users, >>I'm happly using Torque (2.0.0p0) and MAUI (3.2.6p13) on a linux cluster >>(Fedora Core 3 - kernel 2.6.11), but I have some question: >> >>1) >>A problem I've encountered is a uncomfortable delay from when a job is >>submited to whe it starts. >>The delay varies more or less between 1 to 30 seconds, and I guess it's >>related to the line >>RMPOLLINTERVAL 00:00:30 >>in maui.cfg, but it was the suggested value for maui, so I haven't >>changed it. >>How can I reduce this delay? >> >>2) >>Another problem I get is a message when exiting from an interactive job: >><< >>[francesco@node3 ~]$ exit >>logout >>'unknown': I need something more specific. >> >> >>What does it mean? >> >>3) >>The last problem I've encountered is while redirecting the standard >>output of a program to a file. >>A command line like >>"mpiexec program_to_execute > filelog.out 2>&1" >>doesn't always works. With some executable it redirects all the output >>to the file named filelog.out, while with other ones it redirects only >>the output of mpiexec, if it has any. >>This happens both in interactive mode and in batch mode. >>If it could be useful, I get the problem with a self made application >>written in Fortran95 and compiled with the Intel Fortran Compiler 9. >>Any suggestions? >> >>Thank you very much for your patience! >>Francesco >> >>_______________________________________________ >>torqueusers mailing list >>torqueusers@xxxxxxxxxxxxxxxx >>http://www.supercluster.org/mailman/listinfo/torqueusers >> >> > > > >
|
|
| <Prev in Thread] | Current Thread | [Next in Thread> |
|---|---|---|
| Previous by Date: | RE: Compiling xpbsmon, Bisbal, Prentice |
|---|---|
| Next by Date: | RE: Compiling xpbsmon, Diego Vadell |
| Previous by Thread: | Re: Small problems using Torque + MAUI (start delay, error message, redirecting standard output), Dave Jackson |
| Next by Thread: | Re: Small problems using Torque + MAUI (start delay, error message, redirecting standard output), Garrick Staples |
| Indexes: | [Date] [Thread] [Top] [All Lists] |
Free MagazinesCisco NewsReceive a free quarterly e-newsletter with exclusive articles on how Cisco IT uses its own products and solutions to enable the business. subscribe Systems Management News, the newspaper for IT systems administration and data center managers! Each issue of Systems Management News is chock-full of news and analysis to help you understand what's happening in your field. subscribe The Enterprise Newsweekly eWeek is the essential technology information source for builders of e-business. subscribe Oracle Magazine Oracle Magazine contains technology strategy articles, sample code, tips, Oracle and partner news, how to articles for developers and DBAs, and more. Oracle (NASDAQ: ORCL) is the world's largest enterprise software company. subscribe Total Telecom Total Telecom is "The Economist of the communications industry". subscribe |