Hi,
On Thu, 5 Apr 2007, Stern, Julie wrote:
> My parallel installation barfed. Error messages below. I'm on a linux cluster.
> Can anyone tell me what went wrong and what do I do with a broken pipe (call a plumber? )?
:)
> bgc 301% make test.parallel > paralleltest2.log
> [0] MPI Abort by user Aborting program !
> [0] Aborting program!
> make[1]: [test.sander.BASIC] Error 1 (ignored)
> [0] MPI Abort by user Aborting program !
> [0] Aborting program!
> /opt/mpich-1.2.6-icc/bin/mpirun: line 1: 21764 Broken pipe /home/jstern/amber9_parallel/exe/sander.MPI "-O" "-i" "gbin" "-c" "eq1.x" "-o" "mdout.column_fft" -p4pg /home/jstern/amber9_parallel/test/4096wat/PI21642 -p4wd /home/jstern/amber9_parallel/test/4096wat
> make: *** [test.sander.BASIC.MPI] Error 1
>
>
> In the log file:
> cd bintraj; ./Run.bintraj
> sander and ptraj: test sander netCDF output and ptraj netCDF input
> p0_21633: p4_error: : 1
> p2_2873: p4_error: net_recv read: probable EOF on socket: 1
> ./Run.bintraj: Program error
> make[1]: Leaving directory `/home/jstern/amber9_parallel/test'
> export TESTsander=/home/jstern/amber9_parallel/exe/sander.MPI; cd 4096wat; ./Run.column_fft
> p0_21764: p4_error: : 1
> Broken pipe
> ./Run.column_fft: Program error
> Broken pipe
The error messages are generic ones from mpich indicating a failure
in communication. Have you eliminated the usual suspects:
can you rsh or ssh without a request for a password between the nodes ?
did you run the mpich test programs ?
what happens when the number of processors is 1 ?
have you successfully installed and tested serial sander ?
Scott
-----------------------------------------------------------------------
The AMBER Mail Reflector
To post, send mail to amber.scripps.edu
To unsubscribe, send "unsubscribe amber" to majordomo.scripps.edu
Received on Sun Apr 08 2007 - 06:07:31 PDT
