[gt-user] Problems with mpich-g2 + globus + torque

[那日苏]

Hi, All, I have a cluster with 1 head node and 3 slave nodes, and their hostnames are: master:     m01.c01 slaves:     s01.c01     s02.c01     s03.c01 So I wanna build a small grid. I installed mpich-g2, globus, torque into my cluster, and the slaves share the mpich-g2 installation on the head node. I have pasted the "hello world" example with the mpich package, which proves my installation of globus is OK. Then I interfaced torque and globus and submitted the "hello world" job above with a RSL file like this: + ( &(resourceManagerContact="m01.c01/jobmanager-pbs")    (count=2)    (label="subjob 0")    (environment=(GLOBUS_DUROC_SUBJOB_INDEX 0)        (LD_LIBRARY_PATH /usr/local/gt-4.0.1/lib/))    (directory=/home/gt)    (executable=/home/gt/hello/hello) ) It worked very well and the output is: [[EMAIL PROTECTED] hello]$ globusrun -w -f hello.rsl hello, world hello, world Then I set the $mpirun in pbs.pm to $MPICH-G2_HOME/bin/mpirun, and submitted a mpich-g2 job: the classical "cpi" program with the mpich package. But it failed. This is the RSL file: + ( &(resourceManagerContact="m01.c01/jobmanager-pbs")    (count=4)    (jobtype=mpi)    (label="subjob 0")    (environment=(GLOBUS_DUROC_SUBJOB_INDEX 0)                 (LD_LIBRARY_PATH /usr/local/gt-4.0.1/lib/))    (directory="/home/gt/examples")    (executable="/home/gt/examples/cpi") ) The output is: [[EMAIL PROTECTED] examples]$ ./mpirun -globusrsl cpi.rsl     Submission of subjob (label = "subjob 0") failed because authentication with the remote server failed (error code 57)     Submission of subjob (label = "subjob 1") failed because the connection to the server failed (check host and port) (error code 62)     Submission of subjob (label = "subjob 2") failed because the connection to the server failed (check host and port) (error code 62)     Submission of subjob (label = "subjob 3") failed because the connection to the server failed (check host and port) (error code 62) So I googled it and someone said that I have to remove the line "(jobtype=mpi)" if I don't use Vender MPI. I did it and the errors were gone, but it seems like all the processes ran on the head nodes while none on the slaves： [gt@m01.c01 examples]$ ./mpirun -globusrsl cpi.rsl Process 3 on m01.c01 Process 2 on m01.c01 Process 1 on m01.c01 pi is approximately 3.1416009869231249, Error is 0.0000083333333318 wall clock time = 0.083140 Process 0 on m01.c01 Could anyone tell me what's wrong with it? Thanks in advance! Best Regards, Narisu, Beihang University, Beijing, China. Email:[EMAIL PROTECTED]