BTL checkpoint friendly

[ompiteam]

So far only self and tcp seems to be check-friendly. SM has a small issue, but that might be removed with a little work. All others BTL has to be investigated. A testing file will be added shortly once I figure out how to do it ...

[ompiteam]

Imported from trac issue 19. Created by bosilca on 2006-05-19T14:05:54, last modified: 2008-10-23T20:59:56

• bosilca attached ft-test.c on 2006-05-19 14:17:48

[ompiteam]

Trac comment by jjhursey on 2007-03-13 10:52:54:

It seems that the attached program tests to see if Open MPI can successfully fork, and have the child shut down the OMPI layer. So for this bug is it sufficient to show that a BTL can operate correctly under fork conditions to be ''checkpoint friendly''? I guess I'm just searching for what you mean by a ''checkpoint friendly BTL'' since it can mean loads of different things.

[ompiteam]

Trac comment by jsquyres on 2008-10-23 11:24:19:

Josh has done a bunch of work to make various BTL's checkpointable. Is this ticket necessary? It's not really clear what the exact intent of this ticket is.

[ompiteam]

Trac comment by bosilca on 2008-10-23 20:59:56:

This program try to clean the OMPI internal state before taking a checkpoint. If a BTL pass this test, then it can be safely used with BLCR to do a uncoordinated checkpoint. I'm in favor of keeping this open ...