Ticket #912 (new bug)

Opened 3 weeks ago

Last modified 3 weeks ago

fortran test MPI_Errhandler_fatal fails with SMPD on the nightlies

Reported by: jayesh Owned by: jayesh
Priority: minor Milestone: mpich2-1.3
Component: mpich2 Keywords:
Cc:

Description

fortran/env_manager/functional/MPI_Errhandler_fatal

Make step:
make[1]: Entering directory `/sandbox/mpich2/nightly/output/comb15.ubuntu-hardy-amd64/test_1/IntelMPITEST/Test/fortran/env_manager/functional/MPI_Errhandler_fatal'
/sandbox/mpich2/nightly/output/comb15.ubuntu-hardy-amd64/test_1/mpich2-installed/bin/mpif77 -I/sandbox/mpich2/nightly/output/comb15.ubuntu-hardy-amd64/test_1/IntelMPITEST/include -I/sandbox/mpich2/nightly/output/comb15.ubuntu-hardy-amd64/test_1/IntelMPITEST/include -I/sandbox/mpich2/nightly/output/comb15.ubuntu-hardy-amd64/test_1/IntelMPITEST/Test/fortran/env_manager/functional/MPI_Errhandler_fatal   -fno-globals -Wno-globals -o node /sandbox/mpich2/nightly/output/comb15.ubuntu-hardy-amd64/test_1/IntelMPITEST/Test/fortran/env_manager/functional/MPI_Errhandler_fatal/node.F  /sandbox/mpich2/nightly/output/comb15.ubuntu-hardy-amd64/test_1/IntelMPITEST/lib/libmpitestf_mpich2.a -lm
make[1]: Leaving directory `/sandbox/mpich2/nightly/output/comb15.ubuntu-hardy-amd64/test_1/IntelMPITEST/Test/fortran/env_manager/functional/MPI_Errhandler_fatal'
Run step:
Fatal error in MPI_Send: Invalid rank, error stack:
MPI_Send(174): MPI_Send(buf=0x7fffc2e7a554, count=1, MPI_INTEGER, dest=2, tag=0, comm=0x84000000) failed
MPI_Send(99).: Invalid rank has value 2 but must be nonnegative and less than 2

job aborted:
rank: node: exit code[: error message]
0: bb33: -2
1: bb33: 1: process 1 exited without calling finalize

Attachments

Change History

Changed 3 weeks ago by jayesh

  • priority changed from major to minor
  • milestone changed from mpich2-1.2.1 to mpich2-1.3

The test passes when run locally on crush (unable to reproduce locally on crush). This looks to me like one of those output buffering problems that we have seen with SMPD before. The test program prints a (atleast) "PASSED" string in case of success and performs a barrier before doing an invalid MPI_Send() to abort the program. This output in stdout of the process (although the output is flushed on an abort in the client side it looks like the SMPD manager does not process this output) is not forwarded correctly to the stdout for the job. The mtest script looks for "PASSED" string to determine whether the test passed. Since we are too close to the next release (and since it appears to be a minor bug) I am pushing this ticket to the next release.

-Jayesh

WARNING! You need to either login using OpenID here or enter your email address here before you can create or edit tickets. Otherwise the ticket will get treated as spam. More information on creating tickets can be found here.

Don't forget to add your email address to the cc list to make sure that you get updated of the ticket status.

Add/Change #912 (fortran test MPI_Errhandler_fatal fails with SMPD on the nightlies)

Author



Change Properties
Action
as new
 
Note: See TracTickets for help on using tickets.