Mike,<br> The hang seems to be occuring when the MPI library is trying to create the 2-level communicator, during the init phase. Can you try running the test with <span class="subsectionToc"><a href="http://mvapich.cse.ohio-state.edu/support/user_guide_mvapich2-1.4rc1.html#x1-16000011.74" id="QQ2-1-160">MV2_USE_SHMEM_COLL</a></span>=0. This will ensure that a flat communicator is used for the subsequent MPI calls. This might help us isolate the problem. <br>
<br>Thanks,<br>Krishna <br> <br><br><div class="gmail_quote">On Tue, Jul 14, 2009 at 5:04 PM, Mike Heinz <span dir="ltr"><<a href="mailto:michael.heinz@qlogic.com">michael.heinz@qlogic.com</a>></span> wrote:<br>
<blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">
<div link="blue" vlink="purple" lang="EN-US">
<div>
<p>We’re having a very odd problem with our fabric, where,
out of the entire cluster, machine “A” can’t run mvapich2
programs with machine “B”, and machine “C” can’t
run programs with machine “D” – even though “A”
can run with “D” and “B” can run with “C” –
and the rest of the fabric works fine.</p>
<p> </p>
<p style="text-indent: -0.25in;"><span>1)<span style="font-family: "Times New Roman"; font-style: normal; font-variant: normal; font-weight: normal; font-size: 7pt; line-height: normal; font-size-adjust: none; font-stretch: normal;">
</span></span>There are no IB errors anywhere on the fabric that I
can find, and the machines in question all work correctly with mvapich1 and
low-level IB tests.</p>
<p style="text-indent: -0.25in;"><span>2)<span style="font-family: "Times New Roman"; font-style: normal; font-variant: normal; font-weight: normal; font-size: 7pt; line-height: normal; font-size-adjust: none; font-stretch: normal;">
</span></span>The problem occurs whether using mpd or rsh.</p>
<p style="text-indent: -0.25in;"><span>3)<span style="font-family: "Times New Roman"; font-style: normal; font-variant: normal; font-weight: normal; font-size: 7pt; line-height: normal; font-size-adjust: none; font-stretch: normal;">
</span></span>If I attach to the running processes, both machines
appear to be waiting for a read operation to complete. (See below)</p>
<p> </p>
<p>Can anyone make a suggestion on how to debug this? </p>
<p> </p>
<p>Stack trace for node 0:</p>
<p> </p>
<p>#0 0x000000361160abb5 in pthread_spin_lock () from
/lib64/libpthread.so.0</p>
<p>#1 0x00002aaaab08fb6c in mthca_poll_cq
(ibcq=0x2060980, ne=1,</p>
<p> wc=0x7fff9d835900) at src/cq.c:468</p>
<p>#2 0x00002aaaaab5d8d8 in MPIDI_CH3I_MRAILI_Cq_poll (</p>
<p> vbuf_handle=0x7fff9d8359d8, vc_req=0x0,
receiving=0, is_blocking=1)</p>
<p> at /usr/include/infiniband/verbs.h:934</p>
<p>#3 0x00002aaaaab177fa in MPIDI_CH3I_read_progress
(vc_pptr=0x7fff9d8359e0,</p>
<p> v_ptr=0x7fff9d8359d8, is_blocking=1) at
ch3_read_progress.c:143</p>
<p>#4 0x00002aaaaab17464 in MPIDI_CH3I_Progress
(is_blocking=1,</p>
<p> state=<value optimized out>) at
ch3_progress.c:202</p>
<p>#5 0x00002aaaaab5bc4e in MPIC_Wait
(request_ptr=0x2aaaaae19800)</p>
<p> at helper_fns.c:269</p>
<p>#6 0x00002aaaaab5c043 in MPIC_Sendrecv (sendbuf=0x217fc50,
sendcount=2,</p>
<p> sendtype=1275069445, dest=1, sendtag=7,
recvbuf=0x217fc58, recvcount=2,</p>
<p> recvtype=1275069445, source=1, recvtag=7,
comm=1140850688,</p>
<p> status=0x7fff9d835b60) at
helper_fns.c:125</p>
<p>#7 0x00002aaaaaafe387 in MPIR_Allgather
(sendbuf=<value optimized out>,</p>
<p> sendcount=<value optimized out>,
sendtype=1275069445, recvbuf=0x217fc50,</p>
<p> recvcount=2, recvtype=1275069445,
comm_ptr=0x2aaaaae1c1e0)</p>
<p> at allgather.c:192</p>
<p>#8 0x00002aaaaaafeff9 in PMPI_Allgather
(sendbuf=0xffffffffffffffff,</p>
<p> sendcount=2, sendtype=1275069445,
recvbuf=0x217fc50, recvcount=2,</p>
<p> recvtype=1275069445, comm=1140850688) at
allgather.c:866</p>
<p>---Type <return> to continue, or q <return> to
quit---</p>
<p>#9 0x00002aaaaab3b00b in PMPI_Comm_split
(comm=1140850688, color=0, key=0,</p>
<p> newcomm=0x2aaaaae1c2f4) at
comm_split.c:196</p>
<p>#10 0x00002aaaaab3cd84 in create_2level_comm
(comm=1140850688, size=2,</p>
<p> my_rank=<value optimized out>) at
create_2level_comm.c:142</p>
<p>#11 0x00002aaaaab6877d in PMPI_Init (argc=0x7fff9d835e7c,
argv=0x7fff9d835e70)</p>
<p> at init.c:146</p>
<p>#12 0x0000000000400b2f in main (argc=3, argv=0x7fff9d835fb8)
at bw.c:27</p>
<p> </p>
<p>Stack trace for node 1:</p>
<p> </p>
<p>#0 0x00002ac3cbdac2d2 in MPIDI_CH3I_read_progress
(vc_pptr=0x7fffdee81020,</p>
<p> v_ptr=0x7fffdee81018, is_blocking=1) at
ch3_read_progress.c:143</p>
<p>#1 0x00002ac3cbdabf44 in MPIDI_CH3I_Progress
(is_blocking=1,</p>
<p> state=<value optimized out>) at
ch3_progress.c:202</p>
<p>#2 0x00002ac3cbdf060e in MPIC_Wait
(request_ptr=0x2ac3cbfae2a0)</p>
<p> at helper_fns.c:269</p>
<p>#3 0x00002ac3cbdf0a03 in MPIC_Sendrecv
(sendbuf=0xf79028, sendcount=2,</p>
<p> sendtype=1275069445, dest=0, sendtag=7,
recvbuf=0xf79020, recvcount=4,</p>
<p> recvtype=1275069445, source=0, recvtag=7,
comm=1140850688,</p>
<p> status=0x7fffdee811a0) at
helper_fns.c:125</p>
<p>#4 0x00002ac3cbd92ddb in MPIR_Allgather
(sendbuf=<value optimized out>,</p>
<p> sendcount=<value optimized out>,
sendtype=1275069445, recvbuf=0xf79020,</p>
<p> recvcount=2, recvtype=1275069445,
comm_ptr=0x2ac3cbfb0c80)</p>
<p> at allgather.c:192</p>
<p>#5 0x00002ac3cbd93a45 in PMPI_Allgather
(sendbuf=0xffffffffffffffff,</p>
<p> sendcount=2, sendtype=1275069445,
recvbuf=0xf79020, recvcount=2,</p>
<p> recvtype=1275069445, comm=1140850688) at
allgather.c:866</p>
<p>#6 0x00002ac3cbdcf91b in PMPI_Comm_split
(comm=1140850688, color=1, key=0,</p>
<p> newcomm=0x2ac3cbfb0d94) at
comm_split.c:196</p>
<p>#7 0x00002ac3cbdd18f4 in create_2level_comm
(comm=1140850688, size=2,</p>
<p> my_rank=<value optimized out>) at
create_2level_comm.c:142</p>
<p>#8 0x00002ac3cbdfd0a5 in PMPI_Init
(argc=0x7fffdee814bc, argv=0x7fffdee814b0)</p>
<p> at init.c:146</p>
<p>---Type <return> to continue, or q <return> to
quit---</p>
<p>#9 0x0000000000400bcf in main (argc=3, argv=0x7fffdee815f8)
at bw.c:27</p>
<p><span style="font-size: 10pt;">--</span><span style="font-size: 12pt; font-family: "Times New Roman","serif";"></span></p>
<p><span style="font-size: 10pt;">Michael
Heinz</span><span style="font-size: 12pt; font-family: "Times New Roman","serif";"></span></p>
<p><span style="font-size: 10pt;">Principal
Engineer, Qlogic Corporation</span><span style="font-size: 12pt; font-family: "Times New Roman","serif";"></span></p>
<p><span style="font-size: 10pt;">King
of Prussia, Pennsylvania</span></p>
</div>
</div>
<br>_______________________________________________<br>
mvapich-discuss mailing list<br>
<a href="mailto:mvapich-discuss@cse.ohio-state.edu">mvapich-discuss@cse.ohio-state.edu</a><br>
<a href="http://mail.cse.ohio-state.edu/mailman/listinfo/mvapich-discuss" target="_blank">http://mail.cse.ohio-state.edu/mailman/listinfo/mvapich-discuss</a><br>
<br></blockquote></div><br><br clear="all"><br>-- <br>In the middle of difficulty, lies opportunity<br>