This strikes me as very reasonable. That is, the PML should be fixed,
but to keep the issue from being a 1.3.2 blocker we should bump the
mpool_sm_min_size default back up again so that 1.3.2 is no worse than
I put back SVN r20944 with this change. osu_bw now runs (for me).
I filed CMR 1870 to add this change to the 1.3.2 branch. I guess I
need a code review. Could someone review the code for r20944 and
annotate the CMR? It's a one-line/several-character change that bumps
the min default size from 0 to 64M.
At this point, I assume 1857 is no longer a blocker, but in the long
term the PML should be fixed.
Lenny Verkhovsky wrote:
Changing default value is an easy fix. This fix will not add new
possible bugs/dead locks/pathes where noone has gone before on the PML
This fix can be added to Open MPI 1.3 that currently is blocked
due to OSU failure.
PML fix can be done later (IMHO)
On Sat, Apr 4, 2009 at 1:46 AM, Eugene Loh <Eugene.Loh@sun.com>
next on this ticket? It's supposed to be a blocker. Again, the issue
is that osu_bw deluges a receiver with rendezvous messages, but the
receiver does not have enough eager frags to acknowledge them all. We
see this now that the sizing of the mmap file has changed and there's
less headroom to grow the free lists. Possible fixes are:
A) Just make the mmap file default size larger (though less overkill
than we used to have).
B) Fix the PML code that is supposed to deal with cases like this. (At
least I think the PML has code that's intended for this purpose.)
Eugene Loh wrote:
osu_bw, process 0 pumps lots of Isend's to process 1, and process 1 in
turn sets up lots of matching Irecvs. Many messages are in flight.
The question is what happens when resources are exhausted and OMPI
cannot handle so much in-flight traffic. Let's specifically consider
the case of long, rendezvous messages. There are at least two
1) When the sender no longer has any fragments (nor can grow its free
list any more), it queues a send up with add_request_to_send_pending()
and somehow life is good. The PML seems to handle this case
2) When the receiver -- specifically
mca_pml_ob1_recv_request_ack_send_btl() -- no longer has any fragments
to send ACKs back to confirm readiness for rendezvous, the
resource-exhaustion signal travels up the call stack to
mca_pml_ob1_recv_request_ack_send(), who does a
MCA_PML_OB1_ADD_ACK_TO_PENDING(). In short, the PML adds the ACK to
pckt_pending. Somehow, this code path doesn't work.
The reason we see the problem now is that I added "autosizing" of the
shared-memory area. We used to mmap *WAY* too much shared-memory for
small-np jobs. (Yes, that's a subjective statement.) Meanwhile, at
large-np, we didn't mmap enough and jobs wouldn't start. (Objective
statement there.) So, I added heuristics to size the shared area
"appropriately". The heuristics basically targetted the needs of
MPI_Init(). If you want fragment free lists to grow on demand after
MPI_Init(), you now basically have to bump mpool_sm_min_size up
I'd like feedback on a fix. Here are two options:
A) Someone (could be I) increases the default resources. E.g., we
could start with a larger eager free list. Or, I could change those
"heuristics" to allow some amount of headroom for free lists to grow on
demand. Either way, I'd appreciate feedback on how big to set these
B) Someone (not I, since I don't know how) fixes the ob1 PML to handle
scenario 2 above correctly.