[mpich-discuss] MPICH2 1.0.6 'hangs'

[chong tan]

HW            :  INTEL 4XQuad,  128G memory box.
processes    :  6
Threaded    : no
process size : 7G each

when it happened :  first MPI_Send()/MPI_Recv() call, 5 minutes into the run

description :

   between the master and each slaves the max message sizes are around 240K bytes.  However, the
first MPI_Send()/MPI_Recv() only has all slaves send 12 bytes to the master, and receiving 12 bytes
from master.  By design, when this exchange is completed, we are good to go.  

However, on this particular test,  we got stuck at MPI_Send()/MPI_Recv() call for more than 70 minutes
for reason I don't understand.  

has anyone run into situation like this ?  I ran into one with version 1., but this is the first time I run
into this problem with 1.0.6

any suggestion will be helpful.

thanks
tan


__________________________________________________
Do You Yahoo!?
Tired of spam?  Yahoo! Mail has the best spam protection around 
http://mail.yahoo.com 
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mcs.anl.gov/pipermail/mpich-discuss/attachments/20090819/79ceb278/attachment.htm>