LAM/MPI General User's Mailing List Archives

From: Alexander L. Belikoff (ABEL_at_[hidden])
Date: 2006-10-11 11:40:22

• Next message: Josh Lehan: "LAM: Patch for lamhalt lamd segfault if tkill not found"
• Previous message: Jeff Squyres: "Re: LAM: Error message from LAM-7.1.2"
• In reply to: Jeff Squyres: "Re: LAM: Error message from LAM-7.1.2"
• Next in thread: Jeff Squyres: "Re: LAM: Error message from LAM-7.1.2"
• Reply: Jeff Squyres: "Re: LAM: Error message from LAM-7.1.2"

Jeff Squyres wrote:
>
> These error messages mean that processes 2-7 tried to do a receive from
> someone who they later found out were dead, so they aborted.
>
What would be a "standard" (that is, a portable) way for one of the peer
processes to get notified about such a death? For example, if one of
processes dies, I'd like the process of rank 0 to know it in order to
change the strategy.

Cheers,
-- Sasha

• Next message: Josh Lehan: "LAM: Patch for lamhalt lamd segfault if tkill not found"
• Previous message: Jeff Squyres: "Re: LAM: Error message from LAM-7.1.2"
• In reply to: Jeff Squyres: "Re: LAM: Error message from LAM-7.1.2"
• Next in thread: Jeff Squyres: "Re: LAM: Error message from LAM-7.1.2"
• Reply: Jeff Squyres: "Re: LAM: Error message from LAM-7.1.2"