Open MPI logo

Open MPI User's Mailing List Archives

  |   Home   |   Support   |   FAQ   |  

This web mail archive is frozen.

This page is part of a frozen web archive of this mailing list.

You can still navigate around this archive, but know that no new mails have been added to it since July of 2016.

Click here to be taken to the new web archives of this list; it includes all the mails that are in this frozen archive plus all new mails that have been sent to the list since it was migrated to the new archives.

Subject: Re: [OMPI users] 1.7 hangs with --preload-binary
From: Ralph Castain (rhc_at_[hidden])
Date: 2013-04-09 21:04:32


Unfortunately, that functionality fell into disrepair - it is scheduled to be repaired/replaced in v1.7.2.

On Apr 9, 2013, at 4:30 PM, Thomas Vacek <vacek008_at_[hidden]> wrote:

> I just built 1.7. If I distribute the binary, it works fine. If I use the --preload-binary option, it prints
>
> [warn] opal_libevent2019_event_base_loop: reentrant invocation. Only one event_base_loop can run on each event_base at once.
>
> many times per second.
>
> I have bit of an odd configuration: The admins in my organization are too busy to support us, so the installation is in a home folder rather than in the system folders. That also means no NFS, so it's kind of nice to have the binary distributed by the root node. I configured with --disable-dlopen, and use -x PATH -x LD_LIBRARY_PATH in the mpirun invocation. As I said, it works fine if the binary is pre-distributed, but the --preload-binary and ---preload-files options both cause this problem. This problem occurs with the head node and one slave---I haven't tried larger configurations.
>
> Thanks,
>
> Tom
> _______________________________________________
> users mailing list
> users_at_[hidden]
> http://www.open-mpi.org/mailman/listinfo.cgi/users