October 2014 - discuss - mailman3.cels.anl.gov

Re: [mpich-discuss] How to terminate MPI_Comm_accept
by Balaji, Pavan 08 Oct '14

08 Oct '14

Can you send us a simple program that reproduces the issue? — Pavan On Oct 8, 2014, at 2:54 PM, Roy, Hirak <Hirak_Roy(a)mentor.com> wrote: > Hi Pavan, > > I even tried duping the communicator (MPI_COMM_SELF à DUP1, DUP2) before I do any kind of MPI calls. > Still it does not work. > > Thanks, > Hirak > > > > You cannot do a Comm_dup when another thread is doing a collective call on the same communicator (e.g., Comm_accept). > > On Oct 8, 2014, at 2:36 PM, Roy, Hirak <Hirak_Roy at mentor.com> wrote: > > > Hi Pavan, > > > > I tried doing dup on both sides as well as on the client side only. > > Both cases, it did not work. > > > > Thanks, > > Hirak > > > > Here is the server-side > > > > MPI_Comm dupCom ; > > MPI_Comm_dup (MPI_COMM_SELF, &dupCom); > > do { > > MPI_Comm newComm ; > > MPI_Comm_accept (m_serverPort.c_str(), MPI_INFO_NULL, 0, dupCom, &newComm); > > int buf = 0 ; > > MPI_Status status ; > > Log ("Waiting on receive"); > > MPI_Recv(&buf, 1, MPI_INT, MPI_ANY_SOURCE, MPI_ANY_TAG, newComm, &status); > > Log ("Receive complete"); > > if (status.MPI_TAG == MPI_MSG_TAG_NEW_CONN) { > > m_clientComs[m_clientCount] = newComm ; > > m_clientCount++; > > } else if (status.MPI_TAG == MPI_MSG_TAG_SHUTDOWN) { > > Log ("Shutdown"); > > MPI_Comm_disconnect (&newComm); > > Log ("Disconnect"); > > break; > > } else { > > Log ("Unmatched Receive"); > > } > > } while(1) ; > > > > > > > > > > Are you doing MPI_Comm_dup while the other thread is doing MPI_Comm_accept? > > > > — Pavan > > > > > > From: Roy, Hirak > > Sent: Thursday, October 09, 2014 12:55 AM > > To: discuss at mpich.org > > Subject: Re: Re: Re: How to terminate MPI_Comm_accept > > > > Hi Pavan, > > > > Does not work. > > Crash in nemesis. > > Sock hangs in MPI_Recv/MPI_Send > > > > > > Thanks, > > Hirak > > > > MPI_Comm newComm ; > > MPI_Comm dupCom ; > > MPI_Comm_dup (MPI_COMM_SELF, &dupCom); > > MPI_Comm_connect (m_serverPort.c_str(), MPI_INFO_NULL, 0, dupCom, &newComm); > > Log ("Connect to Self"); > > int val = 0 ; > > MPI_Request req ; > > MPI_Send(&val, 1, MPI_INT, 0, MPI_MSG_TAG_SHUTDOWN, newComm); > > Log ("Successful"); > > MPI_Comm_disconnect(&newComm); > > > > > > > > > > On Oct 8, 2014, at 11:41 AM, Roy, Hirak <Hirak_Roy at mentor.com> wrote: > > > Is there any way to come out of the blocking accept? > > > > You can dup MPI_COMM_SELF and do the connect using that other communicator. > > > > — Pavan > > > > -- > > > > _________________________ > > _______________________________________________ > discuss mailing list discuss(a)mpich.org > To manage subscription options or unsubscribe: > https://lists.mpich.org/mailman/listinfo/discuss -- Pavan Balaji ✉️ http://www.mcs.anl.gov/~balaji _______________________________________________ discuss mailing list discuss(a)mpich.org To manage subscription options or unsubscribe: https://lists.mpich.org/mailman/listinfo/discuss

1 0

Re: [mpich-discuss] How to terminate MPI_Comm_accept
by Roy, Hirak 08 Oct '14

08 Oct '14

Hi Pavan, I even tried duping the communicator (MPI_COMM_SELF --> DUP1, DUP2) before I do any kind of MPI calls. Still it does not work. Thanks, Hirak You cannot do a Comm_dup when another thread is doing a collective call on the same communicator (e.g., Comm_accept). On Oct 8, 2014, at 2:36 PM, Roy, Hirak <Hirak_Roy at mentor.com<https://lists.mpich.org/mailman/listinfo/discuss>> wrote: > Hi Pavan, > > I tried doing dup on both sides as well as on the client side only. > Both cases, it did not work. > > Thanks, > Hirak > > Here is the server-side > > MPI_Comm dupCom ; > MPI_Comm_dup (MPI_COMM_SELF, &dupCom); > do { > MPI_Comm newComm ; > MPI_Comm_accept (m_serverPort.c_str(), MPI_INFO_NULL, 0, dupCom, &newComm); > int buf = 0 ; > MPI_Status status ; > Log ("Waiting on receive"); > MPI_Recv(&buf, 1, MPI_INT, MPI_ANY_SOURCE, MPI_ANY_TAG, newComm, &status); > Log ("Receive complete"); > if (status.MPI_TAG == MPI_MSG_TAG_NEW_CONN) { > m_clientComs[m_clientCount] = newComm ; > m_clientCount++; > } else if (status.MPI_TAG == MPI_MSG_TAG_SHUTDOWN) { > Log ("Shutdown"); > MPI_Comm_disconnect (&newComm); > Log ("Disconnect"); > break; > } else { > Log ("Unmatched Receive"); > } > } while(1) ; > > > > > Are you doing MPI_Comm_dup while the other thread is doing MPI_Comm_accept? > > - Pavan > > > From: Roy, Hirak > Sent: Thursday, October 09, 2014 12:55 AM > To: discuss at mpich.org<https://lists.mpich.org/mailman/listinfo/discuss> > Subject: Re: Re: Re: How to terminate MPI_Comm_accept > > Hi Pavan, > > Does not work. > Crash in nemesis. > Sock hangs in MPI_Recv/MPI_Send > > > Thanks, > Hirak > > MPI_Comm newComm ; > MPI_Comm dupCom ; > MPI_Comm_dup (MPI_COMM_SELF, &dupCom); > MPI_Comm_connect (m_serverPort.c_str(), MPI_INFO_NULL, 0, dupCom, &newComm); > Log ("Connect to Self"); > int val = 0 ; > MPI_Request req ; > MPI_Send(&val, 1, MPI_INT, 0, MPI_MSG_TAG_SHUTDOWN, newComm); > Log ("Successful"); > MPI_Comm_disconnect(&newComm); > > > > > On Oct 8, 2014, at 11:41 AM, Roy, Hirak <Hirak_Roy at mentor.com> wrote: > > Is there any way to come out of the blocking accept? > > You can dup MPI_COMM_SELF and do the connect using that other communicator. > > - Pavan > > -- > > _________________________ _______________________________________________ discuss mailing list discuss(a)mpich.org To manage subscription options or unsubscribe: https://lists.mpich.org/mailman/listinfo/discuss

1 0

Re: [mpich-discuss] How to terminate MPI_Comm_accept
by Balaji, Pavan 08 Oct '14

08 Oct '14

You cannot do a Comm_dup when another thread is doing a collective call on the same communicator (e.g., Comm_accept). On Oct 8, 2014, at 2:36 PM, Roy, Hirak <Hirak_Roy(a)mentor.com> wrote: > Hi Pavan, > > I tried doing dup on both sides as well as on the client side only. > Both cases, it did not work. > > Thanks, > Hirak > > Here is the server-side > > MPI_Comm dupCom ; > MPI_Comm_dup (MPI_COMM_SELF, &dupCom); > do { > MPI_Comm newComm ; > MPI_Comm_accept (m_serverPort.c_str(), MPI_INFO_NULL, 0, dupCom, &newComm); > int buf = 0 ; > MPI_Status status ; > Log ("Waiting on receive"); > MPI_Recv(&buf, 1, MPI_INT, MPI_ANY_SOURCE, MPI_ANY_TAG, newComm, &status); > Log ("Receive complete"); > if (status.MPI_TAG == MPI_MSG_TAG_NEW_CONN) { > m_clientComs[m_clientCount] = newComm ; > m_clientCount++; > } else if (status.MPI_TAG == MPI_MSG_TAG_SHUTDOWN) { > Log ("Shutdown"); > MPI_Comm_disconnect (&newComm); > Log ("Disconnect"); > break; > } else { > Log ("Unmatched Receive"); > } > } while(1) ; > > > > > Are you doing MPI_Comm_dup while the other thread is doing MPI_Comm_accept? > > — Pavan > > > From: Roy, Hirak > Sent: Thursday, October 09, 2014 12:55 AM > To: discuss(a)mpich.org > Subject: Re: Re: Re: How to terminate MPI_Comm_accept > > Hi Pavan, > > Does not work. > Crash in nemesis. > Sock hangs in MPI_Recv/MPI_Send > > > Thanks, > Hirak > > MPI_Comm newComm ; > MPI_Comm dupCom ; > MPI_Comm_dup (MPI_COMM_SELF, &dupCom); > MPI_Comm_connect (m_serverPort.c_str(), MPI_INFO_NULL, 0, dupCom, &newComm); > Log ("Connect to Self"); > int val = 0 ; > MPI_Request req ; > MPI_Send(&val, 1, MPI_INT, 0, MPI_MSG_TAG_SHUTDOWN, newComm); > Log ("Successful"); > MPI_Comm_disconnect(&newComm); > > > > > On Oct 8, 2014, at 11:41 AM, Roy, Hirak <Hirak_Roy at mentor.com> wrote: > > Is there any way to come out of the blocking accept? > > You can dup MPI_COMM_SELF and do the connect using that other communicator. > > — Pavan > > -- > > _______________________________________________ > discuss mailing list discuss(a)mpich.org > To manage subscription options or unsubscribe: > https://lists.mpich.org/mailman/listinfo/discuss -- Pavan Balaji ✉️ http://www.mcs.anl.gov/~balaji _______________________________________________ discuss mailing list discuss(a)mpich.org To manage subscription options or unsubscribe: https://lists.mpich.org/mailman/listinfo/discuss

1 0

Re: [mpich-discuss] How to terminate MPI_Comm_accept
by Roy, Hirak 08 Oct '14

08 Oct '14

Hi Pavan, I tried doing dup on both sides as well as on the client side only. Both cases, it did not work. Thanks, Hirak Here is the server-side MPI_Comm dupCom ; MPI_Comm_dup (MPI_COMM_SELF, &dupCom); do { MPI_Comm newComm ; MPI_Comm_accept (m_serverPort.c_str(), MPI_INFO_NULL, 0, dupCom, &newComm); int buf = 0 ; MPI_Status status ; Log ("Waiting on receive"); MPI_Recv(&buf, 1, MPI_INT, MPI_ANY_SOURCE, MPI_ANY_TAG, newComm, &status); Log ("Receive complete"); if (status.MPI_TAG == MPI_MSG_TAG_NEW_CONN) { m_clientComs[m_clientCount] = newComm ; m_clientCount++; } else if (status.MPI_TAG == MPI_MSG_TAG_SHUTDOWN) { Log ("Shutdown"); MPI_Comm_disconnect (&newComm); Log ("Disconnect"); break; } else { Log ("Unmatched Receive"); } } while(1) ; Are you doing MPI_Comm_dup while the other thread is doing MPI_Comm_accept? - Pavan From: Roy, Hirak Sent: Thursday, October 09, 2014 12:55 AM To: discuss(a)mpich.org Subject: Re: Re: Re: How to terminate MPI_Comm_accept Hi Pavan, Does not work. Crash in nemesis. Sock hangs in MPI_Recv/MPI_Send Thanks, Hirak MPI_Comm newComm ; MPI_Comm dupCom ; MPI_Comm_dup (MPI_COMM_SELF, &dupCom); MPI_Comm_connect (m_serverPort.c_str(), MPI_INFO_NULL, 0, dupCom, &newComm); Log ("Connect to Self"); int val = 0 ; MPI_Request req ; MPI_Send(&val, 1, MPI_INT, 0, MPI_MSG_TAG_SHUTDOWN, newComm); Log ("Successful"); MPI_Comm_disconnect(&newComm); On Oct 8, 2014, at 11:41 AM, Roy, Hirak <Hirak_Roy at mentor.com<https://lists.mpich.org/mailman/listinfo/discuss>> wrote: > Is there any way to come out of the blocking accept? You can dup MPI_COMM_SELF and do the connect using that other communicator. - Pavan -- _______________________________________________ discuss mailing list discuss(a)mpich.org To manage subscription options or unsubscribe: https://lists.mpich.org/mailman/listinfo/discuss

1 0

Re: [mpich-discuss] How to terminate MPI_Comm_accept
by Balaji, Pavan 08 Oct '14

08 Oct '14

Are you doing MPI_Comm_dup while the other thread is doing MPI_Comm_accept? — Pavan On Oct 8, 2014, at 2:25 PM, Roy, Hirak <Hirak_Roy(a)mentor.com> wrote: > Hi Pavan, > > Does not work. > Crash in nemesis. > Sock hangs in MPI_Recv/MPI_Send > > > Thanks, > Hirak > > MPI_Comm newComm ; > MPI_Comm dupCom ; > MPI_Comm_dup (MPI_COMM_SELF, &dupCom); > MPI_Comm_connect (m_serverPort.c_str(), MPI_INFO_NULL, 0, dupCom, &newComm); > Log ("Connect to Self"); > int val = 0 ; > MPI_Request req ; > MPI_Send(&val, 1, MPI_INT, 0, MPI_MSG_TAG_SHUTDOWN, newComm); > Log ("Successful"); > MPI_Comm_disconnect(&newComm); > > > > > On Oct 8, 2014, at 11:41 AM, Roy, Hirak <Hirak_Roy at mentor.com> wrote: > > Is there any way to come out of the blocking accept? > > You can dup MPI_COMM_SELF and do the connect using that other communicator. > > — Pavan > > -- > > _______________________________________________ > discuss mailing list discuss(a)mpich.org > To manage subscription options or unsubscribe: > https://lists.mpich.org/mailman/listinfo/discuss -- Pavan Balaji ✉️ http://www.mcs.anl.gov/~balaji _______________________________________________ discuss mailing list discuss(a)mpich.org To manage subscription options or unsubscribe: https://lists.mpich.org/mailman/listinfo/discuss

1 0

Re: [mpich-discuss] How to terminate MPI_Comm_accept
by Roy, Hirak 08 Oct '14

08 Oct '14

Hi Pavan, Does not work. Crash in nemesis. Sock hangs in MPI_Recv/MPI_Send Thanks, Hirak MPI_Comm newComm ; MPI_Comm dupCom ; MPI_Comm_dup (MPI_COMM_SELF, &dupCom); MPI_Comm_connect (m_serverPort.c_str(), MPI_INFO_NULL, 0, dupCom, &newComm); Log ("Connect to Self"); int val = 0 ; MPI_Request req ; MPI_Send(&val, 1, MPI_INT, 0, MPI_MSG_TAG_SHUTDOWN, newComm); Log ("Successful"); MPI_Comm_disconnect(&newComm); On Oct 8, 2014, at 11:41 AM, Roy, Hirak <Hirak_Roy at mentor.com<https://lists.mpich.org/mailman/listinfo/discuss>> wrote: > Is there any way to come out of the blocking accept? You can dup MPI_COMM_SELF and do the connect using that other communicator. - Pavan -- _______________________________________________ discuss mailing list discuss(a)mpich.org To manage subscription options or unsubscribe: https://lists.mpich.org/mailman/listinfo/discuss

1 0

Re: [mpich-discuss] How to terminate MPI_Comm_accept
by Balaji, Pavan 08 Oct '14

08 Oct '14

On Oct 8, 2014, at 11:41 AM, Roy, Hirak <Hirak_Roy(a)mentor.com> wrote: > Is there any way to come out of the blocking accept? You can dup MPI_COMM_SELF and do the connect using that other communicator. — Pavan -- Pavan Balaji ✉️ http://www.mcs.anl.gov/~balaji _______________________________________________ discuss mailing list discuss(a)mpich.org To manage subscription options or unsubscribe: https://lists.mpich.org/mailman/listinfo/discuss

1 0

Re: [mpich-discuss] How to terminate MPI_Comm_accept
by Roy, Hirak 08 Oct '14

08 Oct '14

Thanks Huiwei for trying this out. Looks like this is not allowed. Look at Pavan's reply. -Hirak Hi Hirak, I can reproduce your error with the attached program using one process: mpicc -g -o mpi_comm_accept mpi_comm_accept.c -pthread mpiexec -n 1 ./mpi_comm_accept I found MPI_Send and MPI_Recv were not using the same communicator to communicate, that's why MPI_Recv will never receive the message. So either the communicator creation was wrong or the application was wrong. The standard said MPI_Comm_accept and MPI_Comm_connect are used for "establishing contact between two groups of processes that do not share an existing communicator". But in this case, thread 1 and 2 do share an existing communicator and want to create a new communicator based on the common communicator. I don't know if it is allowed. If it is allowed, then MPI_Comm_accept and MPI_Comm_connect should be fixed to support multiple thread case; if it is not allowed, we may need to change the application to use another way to terminate MPI_Comm_accept. Thanks, - Huiwei On Oct 8, 2014, at 12:14 AM, Roy, Hirak <Hirak_Roy at mentor.com<https://lists.mpich.org/mailman/listinfo/discuss>> wrote: > Hi Pavan, > > Here is my code for thread2 : > > do { > MPI_Comm newComm ; > MPI_Comm_accept (m_serverPort.c_str(), MPI_INFO_NULL, 0, MPI_COMM_SELF, &newComm); > Log ("Accepted a connection"); > int buf = 0 ; > MPI_Status status ; > MPI_Recv(&buf, 1, MPI_INT, MPI_ANY_SOURCE, MPI_ANY_TAG, newComm, &status); > > if (status.MPI_TAG == MPI_MSG_TAG_NEW_CONN) { > m_clientComs[m_clientCount] = newComm ; > m_clientCount++; > } else if (status.MPI_TAG == MPI_MSG_TAG_SHUTDOWN) { > Log ("Shutdown"); > //MPI_Comm_disconnect (&newComm); > Log ("Disconnect"); > break; > } else { > Log ("Unmatched Receive"); > } > } while(1) ; > > > Here is my code for thread1 to terminate thread2 : > > MPI_Comm newComm ; > MPI_Comm_connect (m_serverPort.c_str(), MPI_INFO_NULL, 0, MPI_COMM_SELF, &newComm); > Log ("Connect to Self"); > int val = 0 ; > MPI_Request req ; > MPI_Send(&val, 1, MPI_INT, 0, MPI_MSG_TAG_SHUTDOWN, newComm); > Log ("Successful"); > //MPI_Status stat ; > //MPI_Wait(&req, &stat); > Log ("Complete"); > > //MPI_Comm_disconnect(&newComm); > > > > > The MPI_Send/Recv waits. > I am using sock channel. > For nemesis, I get the following crash : > Assertion failed in file ./src/mpid/ch3/channels/nemesis/include/mpid_nem_inline.h at line 58: vc_ch->is_local > internal ABORT - process 0 > > I tried non-blocking send and receive followed by wait. However, that also does not solve the problem. > > Thanks, > Hirak > > > > ----- > > Hirak, > > Your approach should work fine. I'm not sure what issue you are facing. I assume thread 1 is doing this: > > while (1) { > MPI_Comm_accept(..); > MPI_Recv(.., tag, ..); > if (tag == REGULAR_CONNECTION) > continue; > else if (tag == TERMINATION) { > MPI_Send(..); > break; > } > } > > In this case, all clients do an MPI_Comm_connect and then send a message with tag = REGULAR_CONNECTION. When thread 2 is done with its work, it'll do an MPI_Comm_connect and then send a message with tag = TERMINATION, wait for a response from thread 1, and call finalize. > > - Pavan > _______________________________________________ discuss mailing list discuss(a)mpich.org To manage subscription options or unsubscribe: https://lists.mpich.org/mailman/listinfo/discuss

1 0

Re: [mpich-discuss] How to terminate MPI_Comm_accept
by Roy, Hirak 08 Oct '14

08 Oct '14

Pavan, Is there any way to come out of the blocking accept? Thanks, Hirak Hmm. This will not work. You are simultaneously doing two collectives on the same communicator (MPI_COMM_SELF) from two different threads: MPI_Comm_accept and MPI_Comm_connect. This is not allowed by the MPI standard. - Pavan P.S.: Sorry, I should have caught that yesterday. It's amazing what a good night's sleep can do! On Oct 8, 2014, at 12:14 AM, Roy, Hirak <Hirak_Roy at mentor.com<https://lists.mpich.org/mailman/listinfo/discuss>> wrote: > Hi Pavan, > > Here is my code for thread2 : > > do { > MPI_Comm newComm ; > MPI_Comm_accept (m_serverPort.c_str(), MPI_INFO_NULL, 0, MPI_COMM_SELF, &newComm); > Log ("Accepted a connection"); > int buf = 0 ; > MPI_Status status ; > MPI_Recv(&buf, 1, MPI_INT, MPI_ANY_SOURCE, MPI_ANY_TAG, newComm, &status); > > if (status.MPI_TAG == MPI_MSG_TAG_NEW_CONN) { > m_clientComs[m_clientCount] = newComm ; > m_clientCount++; > } else if (status.MPI_TAG == MPI_MSG_TAG_SHUTDOWN) { > Log ("Shutdown"); > //MPI_Comm_disconnect (&newComm); > Log ("Disconnect"); > break; > } else { > Log ("Unmatched Receive"); > } > } while(1) ; > > > Here is my code for thread1 to terminate thread2 : > > MPI_Comm newComm ; > MPI_Comm_connect (m_serverPort.c_str(), MPI_INFO_NULL, 0, MPI_COMM_SELF, &newComm); > Log ("Connect to Self"); > int val = 0 ; > MPI_Request req ; > MPI_Send(&val, 1, MPI_INT, 0, MPI_MSG_TAG_SHUTDOWN, newComm); > Log ("Successful"); > //MPI_Status stat ; > //MPI_Wait(&req, &stat); > Log ("Complete"); > > //MPI_Comm_disconnect(&newComm); > > > > > The MPI_Send/Recv waits. > I am using sock channel. > For nemesis, I get the following crash : > Assertion failed in file ./src/mpid/ch3/channels/nemesis/include/mpid_nem_inline.h at line 58: vc_ch->is_local > internal ABORT - process 0 > > I tried non-blocking send and receive followed by wait. However, that also does not solve the problem. > > Thanks, > Hirak > > > > ----- > > Hirak, > > Your approach should work fine. I'm not sure what issue you are facing. I assume thread 1 is doing this: > > while (1) { > MPI_Comm_accept(..); > MPI_Recv(.., tag, ..); > if (tag == REGULAR_CONNECTION) > continue; > else if (tag == TERMINATION) { > MPI_Send(..); > break; > } > } > > In this case, all clients do an MPI_Comm_connect and then send a message with tag = REGULAR_CONNECTION. When thread 2 is done with its work, it'll do an MPI_Comm_connect and then send a message with tag = TERMINATION, wait for a response from thread 1, and call finalize. > > - Pavan > _______________________________________________ discuss mailing list discuss(a)mpich.org To manage subscription options or unsubscribe: https://lists.mpich.org/mailman/listinfo/discuss

1 0

Re: [mpich-discuss] How to terminate MPI_Comm_accept
by Balaji, Pavan 08 Oct '14

08 Oct '14

Hmm. This will not work. You are simultaneously doing two collectives on the same communicator (MPI_COMM_SELF) from two different threads: MPI_Comm_accept and MPI_Comm_connect. This is not allowed by the MPI standard. — Pavan P.S.: Sorry, I should have caught that yesterday. It’s amazing what a good night’s sleep can do! On Oct 8, 2014, at 12:14 AM, Roy, Hirak <Hirak_Roy(a)mentor.com> wrote: > Hi Pavan, > > Here is my code for thread2 : > > do { > MPI_Comm newComm ; > MPI_Comm_accept (m_serverPort.c_str(), MPI_INFO_NULL, 0, MPI_COMM_SELF, &newComm); > Log ("Accepted a connection"); > int buf = 0 ; > MPI_Status status ; > MPI_Recv(&buf, 1, MPI_INT, MPI_ANY_SOURCE, MPI_ANY_TAG, newComm, &status); > > if (status.MPI_TAG == MPI_MSG_TAG_NEW_CONN) { > m_clientComs[m_clientCount] = newComm ; > m_clientCount++; > } else if (status.MPI_TAG == MPI_MSG_TAG_SHUTDOWN) { > Log ("Shutdown"); > //MPI_Comm_disconnect (&newComm); > Log ("Disconnect"); > break; > } else { > Log ("Unmatched Receive"); > } > } while(1) ; > > > Here is my code for thread1 to terminate thread2 : > > MPI_Comm newComm ; > MPI_Comm_connect (m_serverPort.c_str(), MPI_INFO_NULL, 0, MPI_COMM_SELF, &newComm); > Log ("Connect to Self"); > int val = 0 ; > MPI_Request req ; > MPI_Send(&val, 1, MPI_INT, 0, MPI_MSG_TAG_SHUTDOWN, newComm); > Log ("Successful"); > //MPI_Status stat ; > //MPI_Wait(&req, &stat); > Log ("Complete"); > > //MPI_Comm_disconnect(&newComm); > > > > > The MPI_Send/Recv waits. > I am using sock channel. > For nemesis, I get the following crash : > Assertion failed in file ./src/mpid/ch3/channels/nemesis/include/mpid_nem_inline.h at line 58: vc_ch->is_local > internal ABORT - process 0 > > I tried non-blocking send and receive followed by wait. However, that also does not solve the problem. > > Thanks, > Hirak > > > > ----- > > Hirak, > > Your approach should work fine. I’m not sure what issue you are facing. I assume thread 1 is doing this: > > while (1) { > MPI_Comm_accept(..); > MPI_Recv(.., tag, ..); > if (tag == REGULAR_CONNECTION) > continue; > else if (tag == TERMINATION) { > MPI_Send(..); > break; > } > } > > In this case, all clients do an MPI_Comm_connect and then send a message with tag = REGULAR_CONNECTION. When thread 2 is done with its work, it’ll do an MPI_Comm_connect and then send a message with tag = TERMINATION, wait for a response from thread 1, and call finalize. > > — Pavan > > _______________________________________________ > discuss mailing list discuss(a)mpich.org > To manage subscription options or unsubscribe: > https://lists.mpich.org/mailman/listinfo/discuss -- Pavan Balaji ✉️ http://www.mcs.anl.gov/~balaji _______________________________________________ discuss mailing list discuss(a)mpich.org To manage subscription options or unsubscribe: https://lists.mpich.org/mailman/listinfo/discuss

1 0