From: X Y [mailto:[EMAIL PROTECTED]
Sent: Thu 16/02/2006 06:33
To: Bernard Li
Cc: [email protected]
Subject: RE: [Oscar-users] PVM jobs need to be forced with qrun to run !
Hi Bernard,
Now even mpi jobs just sit there, along with
pvm jobs. both have to be qrun
thru root.
Whats going on ? I mean can
torque/maui change behaviour on their on over
time.
My server is resonably
secure. I highly doubt any security breach or
something.
Btw what
could be quick short term solutions other than me sitting on the
terminal
qrun'ing users jobs. can qmgr thing be useful. can u suggest a
quick fix
(syntax etc..)
---
Regards
>From: "Bernard Li"
<[EMAIL PROTECTED]>
>To: "X Y" <[EMAIL PROTECTED]>,
<[email protected]>
>Subject: RE: [Oscar-users] PVM
jobs need to be forced with qrun to run !
>Date: Wed, 15 Feb 2006 22:10:18
-0800
>
>So does the job just sit there in the queue and do not
run? Do the logs
>(TORQUE, MAUI) say
anything?
>
>Cheers,
>
>Bernard
>
>________________________________
>
>From:
[EMAIL PROTECTED] on behalf of X Y
>Sent: Wed
15/02/2006 02:43
>To: [email protected]
>Subject:
[Oscar-users] PVM jobs need to be forced with qrun to run
!
>
>
>
>
> Hi,
> My
cluster
specs/config:
>
Oscar version: 4.1
>
OS : Redhat 9
(x86)
> with Default
Oscar installation
>
Compute Nodes: 32 nodes
>
> Im able to run my mpi jobs
fine. a soon as I qsub my mpi-jobs they get
>que-ed
> up
in the default que (workq) & run.
> but my pvm jobs wont
run. unless I su to root & manually (forcefully)
>qrun
>them. So
I
> doubt the problem is related to resources_default.nodes
being set as mpi
>ones are running fine.
> (btw its set
with the qmgr right?). the pvm pbsjobscript is attached
>below
>just
in case.
> Any suggestions/ideas are
welcome.
> Regards
> --
>
SD.
>
>
>
> pvmpbscript:
>
[EMAIL PROTECTED] server_priv]# cat /home/oscartst/pbs_script.pvm
>
************************************
>
#!/bin/sh
>
> ### Job name
> #PBS -N
pvmtest
>
> ### Output files
> #PBS -o
pvmtest.out
> #PBS -e pvmtest.err
>
>
### Queue name
> #PBS -q workq
>
> ###
Script Commands
> cd
$PBS_O_WORKDIR
>
> # generate pvm nodes
file
> echo "* ep=$PBS_O_WORKDIR wd=$PBS_O_WORKDIR" >
pvm_nodes
> cat $PBS_NODEFILE >>
pvm_nodes
>
> # start pvm daemon & wait for slave
daemons to start up
> pvmd pvm_nodes &
>
#sleep 10
>
> # run job
>
p=`pwd`
> cp master1.c slave1.c /tmp
> cd
/tmp
> gcc -I$PVM_ROOT/include master1.c
-L$PVM_ROOT/lib/$PVM_ARCH -lpvm3 -o
>
master1
> gcc -I$PVM_ROOT/include slave1.c
-L$PVM_ROOT/lib/$PVM_ARCH -lpvm3 -o
>slave1
> cp master1
slave1 $p
> cd $p
>
./master1
>
> # wait again to make sure everyone's
finished
> # then kill master pvm daemon
>
#sleep 5
> /usr/bin/killall -TERM
pvmd3
>
> # get rid of lock files & nodes
file
> uid=`id -u`
> tail +2 $PBS_NODEFILE
> pvm_nodes
> /bin/rm -f /tmp/pvm?.$uid
>
crm pvm_nodes:/tmp/pvmd.$uid > /dev/null
2>&1
> crm pvm_nodes:/tmp/pvml.$uid >
/dev/null 2>&1
> /bin/rm -f
pvm_nodes
> exit
>
*************************************
>
>_________________________________________________________________
>Express
yourself instantly with MSN Messenger! Download today - it's FREE!
>http://messenger.msn.click-url.com/go/onm00200471ave/direct/01/
>
>
>
>-------------------------------------------------------
>This
SF.net email is sponsored by: Splunk Inc. Do you grep through
log
>files
>for problems? Stop! Download the new AJAX
search engine that makes
>searching your log files as easy as surfing
the web. DOWNLOAD SPLUNK!
>http://sel.as-us.falkag.net/sel?cmd=lnk&kid=103432&bid=230486&dat=121642
>_______________________________________________
>Oscar-users
mailing list
>[email protected]
>https://lists.sourceforge.net/lists/listinfo/oscar-users
>
>
_________________________________________________________________
Don’t
just search. Find. Check out the new MSN Search!
http://search.msn.click-url.com/go/onm00200636ave/direct/01/
