firemodels / fds

Fire Dynamics Simulator
https://pages.nist.gov/fds-smv/
Other
670 stars 626 forks source link

Mpiexec noerror but error #6937

Closed Behnam-Nouri closed 6 years ago

Behnam-Nouri commented 6 years ago

Hi,

the last time i used FDS and MPI was 2016 and there it worked. I updated to 6.7 and i get some problems. I removed all previous MPI and FDS installation. However i get this.

C:\Program Files\firemodels\FDS6\bin>mpiexec -n 2 fds [unset]: Error reading initack on 444 Error on readline:: No error [unset]: write_line error; fd=444 buf=:cmd=init pmi_version=1 pmi_subversion=1 : system msg for write_line failure : No error [unset]: Unable to write to PMI_fd [unset]: write_line error; fd=444 buf=:cmd=barrier_in : system msg for write_line failure : No error [unset]: write_line error; fd=444 buf=:cmd=get_ranks2hosts : system msg for write_line failure : No error [unset]: expecting cmd="put_ranks2hosts", got cmd="" Fatal error in PMPI_Init_thread: Other MPI error, error stack: MPIR_Init_thread(805): fail failed MPID_Init(1743)......: channel initialization failed MPID_Init(2144)......: PMI_Init returned -1 [unset]: write_line error; fd=444 buf=:cmd=abort exitcode=68204815 : system msg for write_line failure : No error [unset]: Error reading initack on 468 Error on readline:: No error [unset]: write_line error; fd=468 buf=:cmd=init pmi_version=1 pmi_subversion=1 : system msg for write_line failure : No error [unset]: Unable to write to PMI_fd [unset]: write_line error; fd=468 buf=:cmd=barrier_in : system msg for write_line failure : No error [unset]: write_line error; fd=468 buf=:cmd=get_ranks2hosts : system msg for write_line failure : No error [unset]: expecting cmd="put_ranks2hosts", got cmd="" Fatal error in PMPI_Init_thread: Other MPI error, error stack: MPIR_Init_thread(805): fail failed MPID_Init(1743)......: channel initialization failed MPID_Init(2144)......: PMI_Init returned -1 [unset]: write_line error; fd=468 buf=:cmd=abort exitcode=68204815 : system msg for write_line failure : No error>

Behnam-Nouri commented 6 years ago

It doesnt work with any of my working minimal examples.. I have a windows 8 machine with Xeon CPU.

C:\Program Files\firemodels\FDS6\bin>mpiexec -info HYDRA build details: Version: 2018 Update 2 Release Date: 20180125 Process Manager: pmi Bootstrap servers available: service ssh fork Resource management kernels available: user slurm pbs hpc Checkpointing libraries available: blcr Demux engines available: select

C:\Program Files\firemodels\FDS6\bin>mpiexec -validate SUCCESS

C:\Program Files\firemodels\FDS6\bin>

mcgratta commented 6 years ago

Try adding -localonly to the mpiexec command line.

Behnam-Nouri commented 6 years ago

C:\Program Files\firemodels\FDS6\bin>mpiexec -localonly -n 2 test_mpi [unset]: Error reading initack on 468 Error on readline:: No error [unset]: write_line error; fd=468 buf=:cmd=init pmi_version=1 pmi_subversion=1 :+1: and so on like the other error.

sbenkorichi commented 6 years ago

It's been some time now since I've submitted a ticket to Intel regarding this issue, but I haven't heard of them up to date. You can follow it here

mcgratta commented 6 years ago

The only thing I can suggest is to uninstall FDS, check for and kill any hydra services running in the task manager under Services, and then reinstall and try again. We have been having similar problems here as well.

Behnam-Nouri commented 6 years ago

Hi, after deleting all mpich entries in regedit. And killing the hydra service and deleting and reinstalling the whole package it works now. Seems to be an error with an old MPICH installation. Maybe this should be included in the next version installer. Check old MPICH installations.

mcgratta commented 6 years ago

Yes, it seems to be difficult to untangle these various MPI communicators.