topic James, in Intel® MPI Library

problem with mpi_comm_accept

Dave_K_ — Mon, 07 Jan 2013 19:37:55 GMT

I am trying to set up a server client pair that establishes a connection (after being launched independently) using mpi_comm_accept and connect. I successfully have the server wait for a connection request using MPI_Comm_accept. The client successfully connects to using MPI_Comm_connect. Both the server and the client return without any error but the a negative handle is returned in 'newcomm' to both the server and the client. I launch both using mpiexec and have mpd running.

I cannot figure out what is wrong and it is probably something simple. Any ideas?

SERVER:

Integer*4 ierr,rt_gang_comm

…

call MPI_OPEN_PORT(MPI_INFO_NULL,portA,ierr)

write(contape,*) 'Gang Master port = ' // trim(portA)

… server writes the name to a file

call MPI_COMM_ACCEPT(portA,MPI_INFO_NULL,0,MPI_COMM_SELF,rt_gang_comm, ierr)

if (ierr) then

stop ‘serever problem’

else

write(contape,*) 'MPI_COMM_WORLD = ',MPI_COMM_WORLD,' rt_gang_com = ',rt_gang_comm

endif

call MPI_Bcast(i_rec,n_i_s,MPI_INTEGER4,id_mpi_rank,rt_gang_comm, ierr)

if (ierr)then

callMPI_ERROR_STRING(ierr,errs,err_len,ierr1)

stop

endif

SERVER OUTPUT:

Gang Master port = tag#0$rdma_port0#5124$rdma_host0#2:0:0:192:168:11:220:0:0:0:

0:0:0:0:0$

MPI_COMM_WORLD = 1140850688 rt_gang_com = -2080374784

CLIENT:

Integer*4 ierr,intercom

… client reads port name from the server written file

write(contape,*)'Opening port', trim(portA)

call MPI_COMM_CONNECT(portA, MPI_INFO_NULL, 0, MPI_COMM_WORLD, intercomm, ierr)

if (ierr) then

stop ‘client problem’

else

print*,"MPI_COMM_WORLD=",MPI_COMM_WORLD,"intercomm=", intercom

endif

CLIENT OUTOPUT:

Opening port tag#0$rdma_port0#5124$rdma_host0#2:0:0:192:168:11:220:0:0:0:0:0:0:0:0$

MPI_COMM_WORLD= 1140850688 intercomm= -2080374783

A subsequent broadcasr hangs. They are trying to communicate over a communicator that differs by one.

I cannot find anything around that can seem to help so I would appreciate even ideas to try.

Thanks,

Dave

A few particulars:

Dave_K_ — Mon, 07 Jan 2013 19:57:14 GMT

A few particulars: ifort version 12.1.0 CentOS release 5.7 impi/4.0.3/lib64

If you should happen to use

TimP — Mon, 07 Jan 2013 22:57:59 GMT

If you should happen to use an architecture option -xHost on AVX hardware, or one including AVX, I don't know whether it will work, since CentOS 5.7 doesn't support AVX.

Thanks for you input Tim. I

Dave_K_ — Tue, 08 Jan 2013 13:44:48 GMT

Thanks for you input Tim. I was compiling with -xSSE4.2 because it helped with optimization (at some point). I removed it and recompiled with defaults for both the client and server and still end up with a negative pointer.

Are there any other ideas out

Dave_K_ — Tue, 08 Jan 2013 16:54:47 GMT

Are there any other ideas out there? I am so stumped.

Hi Dave,

James_T_Intel — Tue, 08 Jan 2013 17:04:38 GMT

Hi Dave, I've tested the basic functionality of your program, and it appears to work even with a negative value for the communicator. That is simply a handle to the communicator, and as long as the two have the same communicator, the connection should work. From the server side: [plain] $ mpirun -n 1 ./server MPI_COMM_WORLD = 1140850688 newcomm = -2080374784 value = 25 $ cat portname.txt tag#0$rdma_port0#25033$rdma_host0#2:0:0:36:101:26:30:0:0:0:0:0:0:0:0$ [/plain] And the client side: [plain] $ mpirun -n 1 ./client MPI_COMM_WORLD = 1140850688 newcomm = -2080374784 value = 25 [/plain] If you're still having problems, please run with I_MPI_DEBUG=5 and attach the output. Sincerely, James Tullos Technical Consulting Engineer Intel® Cluster Tools

James,

Dave_K_ — Tue, 08 Jan 2013 20:15:05 GMT

James, When I run it the negative communicators differ by one as reflected in the output. Both the client and server have valid communicators but the do not match so the each hange on a broadcast. Dave

James,

Dave_K_ — Tue, 08 Jan 2013 20:38:20 GMT

James, My example was snipped out so the output does not match excatly but it is attached.

James,

Dave_K_ — Tue, 08 Jan 2013 21:07:03 GMT

James, My example was snipped out so the output does not match excatly but it is attached.

James

Dave_K_ — Tue, 08 Jan 2013 21:20:43 GMT

James attached is debug=10 Dave

James

Dave_K_ — Tue, 08 Jan 2013 21:20:46 GMT

James attached is debug=10 Dave

Don't think my last

Dave_K_ — Tue, 08 Jan 2013 21:22:34 GMT

Don't think my last attachment made it

Hi Dave,

James_T_Intel — Wed, 09 Jan 2013 16:49:44 GMT

Hi Dave, I don't see anything immediately wrong in your output (other than the mismatched communicator handle). Can you try using the current version of the Intel® MPI Library (Version 4.1)? Can you send me a reproducer code to test here? Sincerely, James Tullos Technical Consulting Engineer Intel® Cluster Tools

James,

Dave_K_ — Thu, 10 Jan 2013 15:41:02 GMT

James, I sent the perfect test case to you but not have heard back? Dave

Hi Dave,

James_T_Intel — Thu, 10 Jan 2013 19:40:26 GMT

Hi Dave, I have not received the test case. Did you send it via private message? How large is the file? Sincerely, James Tullos Technical Consulting Engineer Intel® Cluster Tools