-
Notifications
You must be signed in to change notification settings - Fork 900
fix grpcomm errors bug 1215 #1254
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Conversation
@annu13 I'm puzzled - I see the new code, but I don't see where the old code was removed. We were obviously setting all these infrastructure up prior to this commit, so I assume the old code is being removed, yes? |
java: try do dlopen libmpi with the full path Since OS X 10.11 (aka El Capitan) DYLD_LIBRARY_PATH is no more propagated to children, so try to dlopen libmpi with the full path using the directory of libmpi_java Fixes open-mpi#1220 Thanks Alexander Daryin for reporting this Update symbol-hiding script btl/sm: rename file after file descriptor has been closed. Thanks George for spotting this.
oops! that was a merge mistake, thanks a bunch for quickly pointing this. In fact I didn't add any code, I just moved the existing job setup section before comm setup. |
@annu13 it seems that you haven't rebased your OMPI branch. And tried to resolve something by hands. This doesn't seems right. Your PR must have only your changes while this one: |
@rhc54 you made a PR from your master branch, and that will get you in trouble ... The way we work is by making a branch first, and the make a pr with that branch. As pointed by @artpol84 , there should be only one commit so you need to rebase, |
I think you meant @annu13 , not me. |
This reverts commit 1a8c351.
closing this as it contains incorrect commits. |
@ggouaillardet I got into this mess in multiple ways and now I am unable
I am trying to remove your changes from annu13/ompi@1a8c351, with no luck. I hope it works now. I am sorry hope if this has caused any inconvenience Thanks, |
…count-ireduce mpi/c: Add each check for count==0 in nonblocking reduce interface
This commit is a potential fix for grpcomm issue:
#1215
The race condition identified can be avoided by setting up the job structure before starting the communications in ess_base_std_orted.c