Hi folks Per the last telecon and some email threads, I have updated the errmgr framework to allow a greater range of response options and remove several loopback messages. See the commit message for more description of what was done:
https://svn.open-mpi.org/trac/ompi/changeset/23023 Please note that I had to touch all the PLM modules to change their function call to report a failed launch. I'm pretty sure I got those changes right, but I cannot verify it for all environments. So please give it a try when you can and let me know if you encounter a problem. I tested this change with both threaded and non-threaded OPAL configurations, and it worked just fine (including for comm_spawn). However, I would not claim that the testing was exhaustive, and it obviously only covered a couple of environments (ssh and slurm). So again, give it a try and let me know if you encounter a problem. Thanks Ralph