[GE users] MPICH2 and SGE 6.2u2 tight integration fails

gustgr rondina at gmail.com
Tue Nov 17 15:25:02 GMT 2009


> Have you checked things are working outside of SGE. i.e. Doing mpdboot 
> and vertifying the mpd daemons are running and using mpdrun to run an 
> example application?

Hello Mark,

I just tested the hello world outside of SGE and it seems to be working:

grondina at master:~/mpi-test/mpich2> cat mpd.hosts 
node20.cluster
node38.cluster
node50.cluster
node39.cluster
grondina at master:~/mpi-test/mpich2> mpdboot -n 5 -f mpd.hosts 
grondina at master:~/mpi-test/mpich2> mpicc -o mpihello mpihello.c
grondina at master:~/mpi-test/mpich2> mpiexec -n 5 ./mpihello
Hello World from Node 0.
Hello World from Node 2.
Hello World from Node 3.
Hello World from Node 1.
Hello World from Node 4.
^Cgrondina at master:~/mpi-test/mpich2> mpdringtest 100
time for 100 loops = 0.0452499389648 seconds
grondina at master:~/mpi-test/mpich2> mpdtrace
master
node39
node38
node50
node20
grondina at master:~/mpi-test/mpich2> mpdtrace -l
master_42411 (192.168.0.254)
node39_34842 (192.168.0.39)
node38_50437 (192.168.0.38)
node50_55329 (192.168.0.50)
node20_50406 (192.168.0.20)
grondina at master:~/mpi-test/mpich2> mpdallexit
grondina at master:~/mpi-test/mpich2>

I was checking startmpich2.sh and the following line is the one responsible for starting the mpd daemon on the nodes:

$SGE_ROOT/mpich2_mpd/bin/$ARC/start_mpich2 -n $host $MPICH2_ROOT/bin/mpd

I did some echo's and all the variables are valid and defined. The locations (both for start_mpich2 and the MPICH2_ROOT directory) are visible from the nodes (through NFS).


Gustavo

------------------------------------------------------
http://gridengine.sunsource.net/ds/viewMessage.do?dsForumId=38&dsMessageId=227464

To unsubscribe from this discussion, e-mail: [users-unsubscribe at gridengine.sunsource.net].



More information about the gridengine-users mailing list