Skip to content

Conversation

@hpcraink
Copy link
Member

@hpcraink hpcraink commented Apr 7, 2016

Per the discussion in the pull request in pmix/master#71 on Mac OSX, the temporary directory can get too long for the socket's sun_path (which is on LINUX 108 chars, on Mac OSX 103 chars), making Open MPI fail with cryptic: "bind() failed"...

Ralph suggested moving the pull request to OMPI.

So, now fail gracefully and have OMPI nicely complain to the user:

USER@SUPER-MacBook-Pro:~/C/MPI_TESTS> ./mpi_wtick
[SUPER-MacBook-Pro.local:28645] [[46320,0],0] bind() failed on error Address already in use (48)
--------------------------------------------------------------------------
The call to pmix_init_server() failed. This may be due to your
system's restriction for Unix's socket's path-length.

orte_proc_session_dir: /var/folders/30/c1t3_6m546n62qrckq5fs_wc0000gp/T/openmpi-sessions-502@USER-SUPER-MacBook-Pro_0/46320/0/0

Please try to set TMPDIR to something short (like /tmp) or change
Your computer's name (see uname -n).

hpcraink added 2 commits April 7, 2016 22:12
   openpmix/openpmix#71

Have OMPI's current version of pmix120 nicely fail in case of
too long sun_path (longer than 108 or in case of OSX 103 chars).
And have OMPI return proper error messages with hints how to
amend.
orte/orted/pmix/pmix_server.c reference orterun.
@rhc54 rhc54 merged commit 5b8a40a into open-mpi:master Apr 8, 2016
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants