Re: [OMPI devel] UD BTL alltoall hangs
Thanks George. I figured out the problem (two of them actually) based on a pointer from Gleb (thanks Gleb). I have two types of send queues on the UD BTL -- one is per-module, and the other is per-endpoint. I had missed looking for stuck frags on the per-endpoint queues. So something is wrong with the per-endpoint queues and their interaction with the per-module queue. Disabling the per-endpoint queue makes the problem go away, and I'm not sure I liked having them in the first place. But this still left a similar problem at 2kb messages. I had static limits set for free list lengths based on the btl_ofud_sd_num MCA parameter. Switching the max to unlimited makes this problem go away too. Good enough to get some runs through for now :) Andrew George Bosilca wrote: Andrew, There is an option on the message queue stuff, that allow you to see all internal pending requests. On the current trunk, edit the file ompi/debuggers/ompi_dll.s at line 736 and set the p_info->show_internal_requests to 1. Now compile and install it, and then restart totalview. You should be able to get access to all pending requests, even those created by the collective modules. Moreover, the missing sends should be somewhere. If they are not in the BTL, and i they are not completed, then hopefully they are in the PML in the send_pending list. As the collective works on all other BTL I suppose the communication pattern is correct, so there is something happening with the requests when using the UD BTL. If the requests are not in the PML send_pending queue, the next thing you can do is to modify the receive handles in the OB1 PML, and print all incoming match header. You will have to somehow sort the output, but at least you can figure out, what is happening with the missing messages. george. On Sep 11, 2007, at 12:37 PM, Andrew Friedley wrote: First off, I've managed to reproduce this with nbcbench using only 16 procs (two per node), and setting btl_ofud_sd_num to 12 -- eases debugging with fewer procs to look at. ompi_coll_tuned_alltoall_intra_basic_linear is the alltoall routine that is being called. What I'm seeing from totalview is that some random number of procs (1-5 usually, varies from run to run) are sitting with a send and a recv outstanding to every other proc. The other procs however have moved on to the next collective. This is hard to see with the default nbcbench code since it calls only alltoall repeatedly -- adding a barrier after the MPI_Alltoall() call makes it easier to see, as the barrier has a different tag number and communication pattern. So what I see is a few procs stuck in alltoall, while the rest are waiting in the following barrier. I've also verified with totalview that there are no outstanding send wqe's at the UD BTL, and all procs are polling progress. The procs in the alltoall are polling in the opal_condition_wait() called from ompi_request_wait_all(). Not sure what to ask or where to look further other than, what should I look at to see what requests are outstanding in the PML? Andrew George Bosilca wrote: The first step will be to figure out which version of the alltoall you're using. I suppose you use the default parameters, and then the decision function in the tuned component say it is using the linear all to all. As the name state it, this means that every node will post one receive from any other node and then will start sending to every other node the respective fragment. This will lead to a lot of outstanding sends and receives. I doubt that the receive can cause a problem, so I expect the problem is coming from the send side. Do you have TotalView installed on your odin ? If yes there is a simple way to see how many sends are pending and where ... That might pinpoint [at least] the process where you should look to see what' wrong. george. On Aug 29, 2007, at 12:37 AM, Andrew Friedley wrote: I'm having a problem with the UD BTL and hoping someone might have some input to help solve it. What I'm seeing is hangs when running alltoall benchmarks with nbcbench or an LLNL program called mpiBench -- both hang exactly the same way. With the code on the trunk running nbcbench on IU's odin using 32 nodes and a command line like this: mpirun -np 128 -mca btl ofud,self ./nbcbench -t MPI_Alltoall -p 128-128 -s 1-262144 hangs consistently when testing 256-byte messages. There are two things I can do to make the hang go away until running at larger scale. First is to increase the 'btl_ofud_sd_num' MCA param from its default value of 128. This allows you to run with more procs/nodes before hitting the hang, but AFAICT doesn't fix the actual problem. What this parameter does is control the maximum number of outstanding send WQEs posted at the IB level -- when the limit is reached, frags are queued on an opal_list_t and later sent by progress as IB sends complete. The other way I've found is to play games with calling mca_btl_ud_component_p
Re: [OMPI devel] UD BTL alltoall hangs
Andrew, There is an option on the message queue stuff, that allow you to see all internal pending requests. On the current trunk, edit the file ompi/debuggers/ompi_dll.s at line 736 and set the p_info- >show_internal_requests to 1. Now compile and install it, and then restart totalview. You should be able to get access to all pending requests, even those created by the collective modules. Moreover, the missing sends should be somewhere. If they are not in the BTL, and i they are not completed, then hopefully they are in the PML in the send_pending list. As the collective works on all other BTL I suppose the communication pattern is correct, so there is something happening with the requests when using the UD BTL. If the requests are not in the PML send_pending queue, the next thing you can do is to modify the receive handles in the OB1 PML, and print all incoming match header. You will have to somehow sort the output, but at least you can figure out, what is happening with the missing messages. george. On Sep 11, 2007, at 12:37 PM, Andrew Friedley wrote: First off, I've managed to reproduce this with nbcbench using only 16 procs (two per node), and setting btl_ofud_sd_num to 12 -- eases debugging with fewer procs to look at. ompi_coll_tuned_alltoall_intra_basic_linear is the alltoall routine that is being called. What I'm seeing from totalview is that some random number of procs (1-5 usually, varies from run to run) are sitting with a send and a recv outstanding to every other proc. The other procs however have moved on to the next collective. This is hard to see with the default nbcbench code since it calls only alltoall repeatedly -- adding a barrier after the MPI_Alltoall() call makes it easier to see, as the barrier has a different tag number and communication pattern. So what I see is a few procs stuck in alltoall, while the rest are waiting in the following barrier. I've also verified with totalview that there are no outstanding send wqe's at the UD BTL, and all procs are polling progress. The procs in the alltoall are polling in the opal_condition_wait() called from ompi_request_wait_all(). Not sure what to ask or where to look further other than, what should I look at to see what requests are outstanding in the PML? Andrew George Bosilca wrote: The first step will be to figure out which version of the alltoall you're using. I suppose you use the default parameters, and then the decision function in the tuned component say it is using the linear all to all. As the name state it, this means that every node will post one receive from any other node and then will start sending to every other node the respective fragment. This will lead to a lot of outstanding sends and receives. I doubt that the receive can cause a problem, so I expect the problem is coming from the send side. Do you have TotalView installed on your odin ? If yes there is a simple way to see how many sends are pending and where ... That might pinpoint [at least] the process where you should look to see what' wrong. george. On Aug 29, 2007, at 12:37 AM, Andrew Friedley wrote: I'm having a problem with the UD BTL and hoping someone might have some input to help solve it. What I'm seeing is hangs when running alltoall benchmarks with nbcbench or an LLNL program called mpiBench -- both hang exactly the same way. With the code on the trunk running nbcbench on IU's odin using 32 nodes and a command line like this: mpirun -np 128 -mca btl ofud,self ./nbcbench -t MPI_Alltoall -p 128-128 -s 1-262144 hangs consistently when testing 256-byte messages. There are two things I can do to make the hang go away until running at larger scale. First is to increase the 'btl_ofud_sd_num' MCA param from its default value of 128. This allows you to run with more procs/nodes before hitting the hang, but AFAICT doesn't fix the actual problem. What this parameter does is control the maximum number of outstanding send WQEs posted at the IB level -- when the limit is reached, frags are queued on an opal_list_t and later sent by progress as IB sends complete. The other way I've found is to play games with calling mca_btl_ud_component_progress() in mca_btl_ud_endpoint_post_send (). In fact I replaced the CHECK_FRAG_QUEUES() macro used around btl_ofud_endpoint.c:77 with a version that loops on progress until a send WQE slot is available (as opposed to queueing). Same result -- I can run at larger scale, but still hit the hang eventually. It appears that when the job hangs, progress is being polled very quickly, and after spinning for a while there are no outstanding send WQEs or queued sends in the BTL. I'm not sure where further up things are spinning/blocking, as I can't produce the hang at less than 32 nodes / 128 procs and don't have a good way of debugging that (suggestions appreciated). Furthermore, both ob1 and dr PMLs result in the same behavior, except
Re: [OMPI devel] UD BTL alltoall hangs
First off, I've managed to reproduce this with nbcbench using only 16 procs (two per node), and setting btl_ofud_sd_num to 12 -- eases debugging with fewer procs to look at. ompi_coll_tuned_alltoall_intra_basic_linear is the alltoall routine that is being called. What I'm seeing from totalview is that some random number of procs (1-5 usually, varies from run to run) are sitting with a send and a recv outstanding to every other proc. The other procs however have moved on to the next collective. This is hard to see with the default nbcbench code since it calls only alltoall repeatedly -- adding a barrier after the MPI_Alltoall() call makes it easier to see, as the barrier has a different tag number and communication pattern. So what I see is a few procs stuck in alltoall, while the rest are waiting in the following barrier. I've also verified with totalview that there are no outstanding send wqe's at the UD BTL, and all procs are polling progress. The procs in the alltoall are polling in the opal_condition_wait() called from ompi_request_wait_all(). Not sure what to ask or where to look further other than, what should I look at to see what requests are outstanding in the PML? Andrew George Bosilca wrote: The first step will be to figure out which version of the alltoall you're using. I suppose you use the default parameters, and then the decision function in the tuned component say it is using the linear all to all. As the name state it, this means that every node will post one receive from any other node and then will start sending to every other node the respective fragment. This will lead to a lot of outstanding sends and receives. I doubt that the receive can cause a problem, so I expect the problem is coming from the send side. Do you have TotalView installed on your odin ? If yes there is a simple way to see how many sends are pending and where ... That might pinpoint [at least] the process where you should look to see what' wrong. george. On Aug 29, 2007, at 12:37 AM, Andrew Friedley wrote: I'm having a problem with the UD BTL and hoping someone might have some input to help solve it. What I'm seeing is hangs when running alltoall benchmarks with nbcbench or an LLNL program called mpiBench -- both hang exactly the same way. With the code on the trunk running nbcbench on IU's odin using 32 nodes and a command line like this: mpirun -np 128 -mca btl ofud,self ./nbcbench -t MPI_Alltoall -p 128-128 -s 1-262144 hangs consistently when testing 256-byte messages. There are two things I can do to make the hang go away until running at larger scale. First is to increase the 'btl_ofud_sd_num' MCA param from its default value of 128. This allows you to run with more procs/nodes before hitting the hang, but AFAICT doesn't fix the actual problem. What this parameter does is control the maximum number of outstanding send WQEs posted at the IB level -- when the limit is reached, frags are queued on an opal_list_t and later sent by progress as IB sends complete. The other way I've found is to play games with calling mca_btl_ud_component_progress() in mca_btl_ud_endpoint_post_send (). In fact I replaced the CHECK_FRAG_QUEUES() macro used around btl_ofud_endpoint.c:77 with a version that loops on progress until a send WQE slot is available (as opposed to queueing). Same result -- I can run at larger scale, but still hit the hang eventually. It appears that when the job hangs, progress is being polled very quickly, and after spinning for a while there are no outstanding send WQEs or queued sends in the BTL. I'm not sure where further up things are spinning/blocking, as I can't produce the hang at less than 32 nodes / 128 procs and don't have a good way of debugging that (suggestions appreciated). Furthermore, both ob1 and dr PMLs result in the same behavior, except that DR eventually trips a watchdog timeout, fails the BTL, and terminates the job. Other collectives such as allreduce and allgather do not hang -- only alltoall. I can also reproduce the hang on LLNL's Atlas machine. Can anyone else reproduce this (Torsten might have to make a copy of nbcbench available)? Anyone have any ideas as to what's wrong? Andrew ___ devel mailing list de...@open-mpi.org http://www.open-mpi.org/mailman/listinfo.cgi/devel ___ devel mailing list de...@open-mpi.org http://www.open-mpi.org/mailman/listinfo.cgi/devel
Re: [OMPI devel] UD BTL alltoall hangs
George Bosilca wrote: Until then you should be using the latest command "tv8 mpirun -a -np 2 -bynode `pwd`/NPmpi". The `pwd` is really important for some reason, otherwise TotalView is unable to find the executable. The problem is that the name of the process will be "./NPmpi" and TotalView does not have access to the path where the executable was launched (at least that's the reason I think). Thanks George. That works except for one catch, when I'm asked on startup if I want to stop the parallel job (and hit yes), totalview waits forever trying to connect to a remote server. I see this on the xterm (shortened in a few places): Launching TotalView Debugger Servers with command: srun --jobid=0 -N1 -n1 -w`awk -F. 'BEGIN {ORS=","} {if (NR==1) ORS=""; print $1}' $PWD/TVT1Pa4Fjm` -l --input=none /usr/global/tools/totalview.8.1.0-1/linux-x86-64/bin/tvdsvr -callback_host atlas34 -callback_ports atlas31:16382 -set_pws 47319a24:4688a7a2 -verbosity info -working_directory $PWD/NetPIPE_3.6.2 srun: error: Invalid numeric value "0" for jobid. I got around this by hitting cancel in the 'waiting to connect' dialog, then setting my slurm jobid manually in file -> preferences -> bulk launch -> command instead of the %J filler, and restarting. Is there a better work around for this? Andrew
Re: [OMPI devel] UD BTL alltoall hangs
On Aug 29, 2007, at 7:05 PM, Andrew Friedley wrote: $ mpirun -debug -np 2 -bynode -debug-daemons ./NPmpi -- Internal error -- the orte_base_user_debugger MCA parameter was not able to be found. Please contact the Open RTE developers; this should not happen. -- Grepping for that param in ompi_info shows: MCA orte: parameter "orte_base_user_debugger" (current value: "totalview @mpirun@ -a @mpirun_args@ : ddt -n @np@ -start @executable@ @executable_argv@ @single_app@ : fxp @mpirun@ -a @mpirun_args@") This has been broken or a while. It's a long story to explain, but a fix is on the way. Until then you should be using the latest command "tv8 mpirun -a -np 2 -bynode `pwd`/NPmpi". The `pwd` is really important for some reason, otherwise TotalView is unable to find the executable. The problem is that the name of the process will be "./NPmpi" and TotalView does not have access to the path where the executable was launched (at least that's the reason I think). Once you do this, you should be good to go. george. What's going on? I also tried running totalview directly, using a line like this: totalview mpirun -a -np 2 -bynode -debug-daemons ./NPmpi Totalview comes up and seems to be running debugging the mpirun process, with only one thread. Doesn't seem to be aware that this is an MPI job with other MPI processes.. any ideas? Andrew George Bosilca wrote: The first step will be to figure out which version of the alltoall you're using. I suppose you use the default parameters, and then the decision function in the tuned component say it is using the linear all to all. As the name state it, this means that every node will post one receive from any other node and then will start sending to every other node the respective fragment. This will lead to a lot of outstanding sends and receives. I doubt that the receive can cause a problem, so I expect the problem is coming from the send side. Do you have TotalView installed on your odin ? If yes there is a simple way to see how many sends are pending and where ... That might pinpoint [at least] the process where you should look to see what' wrong. george. On Aug 29, 2007, at 12:37 AM, Andrew Friedley wrote: I'm having a problem with the UD BTL and hoping someone might have some input to help solve it. What I'm seeing is hangs when running alltoall benchmarks with nbcbench or an LLNL program called mpiBench -- both hang exactly the same way. With the code on the trunk running nbcbench on IU's odin using 32 nodes and a command line like this: mpirun -np 128 -mca btl ofud,self ./nbcbench -t MPI_Alltoall -p 128-128 -s 1-262144 hangs consistently when testing 256-byte messages. There are two things I can do to make the hang go away until running at larger scale. First is to increase the 'btl_ofud_sd_num' MCA param from its default value of 128. This allows you to run with more procs/nodes before hitting the hang, but AFAICT doesn't fix the actual problem. What this parameter does is control the maximum number of outstanding send WQEs posted at the IB level -- when the limit is reached, frags are queued on an opal_list_t and later sent by progress as IB sends complete. The other way I've found is to play games with calling mca_btl_ud_component_progress() in mca_btl_ud_endpoint_post_send (). In fact I replaced the CHECK_FRAG_QUEUES() macro used around btl_ofud_endpoint.c:77 with a version that loops on progress until a send WQE slot is available (as opposed to queueing). Same result -- I can run at larger scale, but still hit the hang eventually. It appears that when the job hangs, progress is being polled very quickly, and after spinning for a while there are no outstanding send WQEs or queued sends in the BTL. I'm not sure where further up things are spinning/blocking, as I can't produce the hang at less than 32 nodes / 128 procs and don't have a good way of debugging that (suggestions appreciated). Furthermore, both ob1 and dr PMLs result in the same behavior, except that DR eventually trips a watchdog timeout, fails the BTL, and terminates the job. Other collectives such as allreduce and allgather do not hang -- only alltoall. I can also reproduce the hang on LLNL's Atlas machine. Can anyone else reproduce this (Torsten might have to make a copy of nbcbench available)? Anyone have any ideas as to what's wrong? Andrew ___ devel mailing list de...@open-mpi.org http://www.open-mpi.org/mailman/listinfo.cgi/devel ___ devel mailing list de...@open-mpi.org http://www.open-mpi.org/mailman/listinfo.cgi/devel ___ devel mailing list de...@open-mpi.
Re: [OMPI devel] UD BTL alltoall hangs
Thanks for the suggestion; though that appears to hang with no output whatsoever. Andrew Aurelien Bouteiller wrote: You should try mpirun -np 2 -bynode totalview ./NPmpi Aurelien Le 29 août 07 à 13:05, Andrew Friedley a écrit : OK, I've never used totalview before. So doing some FAQ reading I got an xterm on an Atlas node (odin doesn't have totalview AFAIK). Trying a simple netpipe run just to get familiar with things results in this: $ mpirun -debug -np 2 -bynode -debug-daemons ./NPmpi -- Internal error -- the orte_base_user_debugger MCA parameter was not able to be found. Please contact the Open RTE developers; this should not happen. -- Grepping for that param in ompi_info shows: MCA orte: parameter "orte_base_user_debugger" (current value: "totalview @mpirun@ -a @mpirun_args@ : ddt -n @np@ -start @executable@ @executable_argv@ @single_app@ : fxp @mpirun@ -a @mpirun_args@") What's going on? I also tried running totalview directly, using a line like this: totalview mpirun -a -np 2 -bynode -debug-daemons ./NPmpi Totalview comes up and seems to be running debugging the mpirun process, with only one thread. Doesn't seem to be aware that this is an MPI job with other MPI processes.. any ideas? Andrew George Bosilca wrote: The first step will be to figure out which version of the alltoall you're using. I suppose you use the default parameters, and then the decision function in the tuned component say it is using the linear all to all. As the name state it, this means that every node will post one receive from any other node and then will start sending to every other node the respective fragment. This will lead to a lot of outstanding sends and receives. I doubt that the receive can cause a problem, so I expect the problem is coming from the send side. Do you have TotalView installed on your odin ? If yes there is a simple way to see how many sends are pending and where ... That might pinpoint [at least] the process where you should look to see what' wrong. george. On Aug 29, 2007, at 12:37 AM, Andrew Friedley wrote: I'm having a problem with the UD BTL and hoping someone might have some input to help solve it. What I'm seeing is hangs when running alltoall benchmarks with nbcbench or an LLNL program called mpiBench -- both hang exactly the same way. With the code on the trunk running nbcbench on IU's odin using 32 nodes and a command line like this: mpirun -np 128 -mca btl ofud,self ./nbcbench -t MPI_Alltoall -p 128-128 -s 1-262144 hangs consistently when testing 256-byte messages. There are two things I can do to make the hang go away until running at larger scale. First is to increase the 'btl_ofud_sd_num' MCA param from its default value of 128. This allows you to run with more procs/nodes before hitting the hang, but AFAICT doesn't fix the actual problem. What this parameter does is control the maximum number of outstanding send WQEs posted at the IB level -- when the limit is reached, frags are queued on an opal_list_t and later sent by progress as IB sends complete. The other way I've found is to play games with calling mca_btl_ud_component_progress() in mca_btl_ud_endpoint_post_send (). In fact I replaced the CHECK_FRAG_QUEUES() macro used around btl_ofud_endpoint.c:77 with a version that loops on progress until a send WQE slot is available (as opposed to queueing). Same result -- I can run at larger scale, but still hit the hang eventually. It appears that when the job hangs, progress is being polled very quickly, and after spinning for a while there are no outstanding send WQEs or queued sends in the BTL. I'm not sure where further up things are spinning/blocking, as I can't produce the hang at less than 32 nodes / 128 procs and don't have a good way of debugging that (suggestions appreciated). Furthermore, both ob1 and dr PMLs result in the same behavior, except that DR eventually trips a watchdog timeout, fails the BTL, and terminates the job. Other collectives such as allreduce and allgather do not hang -- only alltoall. I can also reproduce the hang on LLNL's Atlas machine. Can anyone else reproduce this (Torsten might have to make a copy of nbcbench available)? Anyone have any ideas as to what's wrong? Andrew ___ devel mailing list de...@open-mpi.org http://www.open-mpi.org/mailman/listinfo.cgi/devel ___ devel mailing list de...@open-mpi.org http://www.open-mpi.org/mailman/listinfo.cgi/devel ___ devel mailing list de...@open-mpi.org http://www.open-mpi.org/mailman/listinfo.cgi/devel ___ devel mailing list de...@open-mpi.
Re: [OMPI devel] UD BTL alltoall hangs
You should try mpirun -np 2 -bynode totalview ./NPmpi Aurelien Le 29 août 07 à 13:05, Andrew Friedley a écrit : OK, I've never used totalview before. So doing some FAQ reading I got an xterm on an Atlas node (odin doesn't have totalview AFAIK). Trying a simple netpipe run just to get familiar with things results in this: $ mpirun -debug -np 2 -bynode -debug-daemons ./NPmpi -- Internal error -- the orte_base_user_debugger MCA parameter was not able to be found. Please contact the Open RTE developers; this should not happen. -- Grepping for that param in ompi_info shows: MCA orte: parameter "orte_base_user_debugger" (current value: "totalview @mpirun@ -a @mpirun_args@ : ddt -n @np@ -start @executable@ @executable_argv@ @single_app@ : fxp @mpirun@ -a @mpirun_args@") What's going on? I also tried running totalview directly, using a line like this: totalview mpirun -a -np 2 -bynode -debug-daemons ./NPmpi Totalview comes up and seems to be running debugging the mpirun process, with only one thread. Doesn't seem to be aware that this is an MPI job with other MPI processes.. any ideas? Andrew George Bosilca wrote: The first step will be to figure out which version of the alltoall you're using. I suppose you use the default parameters, and then the decision function in the tuned component say it is using the linear all to all. As the name state it, this means that every node will post one receive from any other node and then will start sending to every other node the respective fragment. This will lead to a lot of outstanding sends and receives. I doubt that the receive can cause a problem, so I expect the problem is coming from the send side. Do you have TotalView installed on your odin ? If yes there is a simple way to see how many sends are pending and where ... That might pinpoint [at least] the process where you should look to see what' wrong. george. On Aug 29, 2007, at 12:37 AM, Andrew Friedley wrote: I'm having a problem with the UD BTL and hoping someone might have some input to help solve it. What I'm seeing is hangs when running alltoall benchmarks with nbcbench or an LLNL program called mpiBench -- both hang exactly the same way. With the code on the trunk running nbcbench on IU's odin using 32 nodes and a command line like this: mpirun -np 128 -mca btl ofud,self ./nbcbench -t MPI_Alltoall -p 128-128 -s 1-262144 hangs consistently when testing 256-byte messages. There are two things I can do to make the hang go away until running at larger scale. First is to increase the 'btl_ofud_sd_num' MCA param from its default value of 128. This allows you to run with more procs/nodes before hitting the hang, but AFAICT doesn't fix the actual problem. What this parameter does is control the maximum number of outstanding send WQEs posted at the IB level -- when the limit is reached, frags are queued on an opal_list_t and later sent by progress as IB sends complete. The other way I've found is to play games with calling mca_btl_ud_component_progress() in mca_btl_ud_endpoint_post_send (). In fact I replaced the CHECK_FRAG_QUEUES() macro used around btl_ofud_endpoint.c:77 with a version that loops on progress until a send WQE slot is available (as opposed to queueing). Same result -- I can run at larger scale, but still hit the hang eventually. It appears that when the job hangs, progress is being polled very quickly, and after spinning for a while there are no outstanding send WQEs or queued sends in the BTL. I'm not sure where further up things are spinning/blocking, as I can't produce the hang at less than 32 nodes / 128 procs and don't have a good way of debugging that (suggestions appreciated). Furthermore, both ob1 and dr PMLs result in the same behavior, except that DR eventually trips a watchdog timeout, fails the BTL, and terminates the job. Other collectives such as allreduce and allgather do not hang -- only alltoall. I can also reproduce the hang on LLNL's Atlas machine. Can anyone else reproduce this (Torsten might have to make a copy of nbcbench available)? Anyone have any ideas as to what's wrong? Andrew ___ devel mailing list de...@open-mpi.org http://www.open-mpi.org/mailman/listinfo.cgi/devel ___ devel mailing list de...@open-mpi.org http://www.open-mpi.org/mailman/listinfo.cgi/devel ___ devel mailing list de...@open-mpi.org http://www.open-mpi.org/mailman/listinfo.cgi/devel
Re: [OMPI devel] UD BTL alltoall hangs
OK, I've never used totalview before. So doing some FAQ reading I got an xterm on an Atlas node (odin doesn't have totalview AFAIK). Trying a simple netpipe run just to get familiar with things results in this: $ mpirun -debug -np 2 -bynode -debug-daemons ./NPmpi -- Internal error -- the orte_base_user_debugger MCA parameter was not able to be found. Please contact the Open RTE developers; this should not happen. -- Grepping for that param in ompi_info shows: MCA orte: parameter "orte_base_user_debugger" (current value: "totalview @mpirun@ -a @mpirun_args@ : ddt -n @np@ -start @executable@ @executable_argv@ @single_app@ : fxp @mpirun@ -a @mpirun_args@") What's going on? I also tried running totalview directly, using a line like this: totalview mpirun -a -np 2 -bynode -debug-daemons ./NPmpi Totalview comes up and seems to be running debugging the mpirun process, with only one thread. Doesn't seem to be aware that this is an MPI job with other MPI processes.. any ideas? Andrew George Bosilca wrote: The first step will be to figure out which version of the alltoall you're using. I suppose you use the default parameters, and then the decision function in the tuned component say it is using the linear all to all. As the name state it, this means that every node will post one receive from any other node and then will start sending to every other node the respective fragment. This will lead to a lot of outstanding sends and receives. I doubt that the receive can cause a problem, so I expect the problem is coming from the send side. Do you have TotalView installed on your odin ? If yes there is a simple way to see how many sends are pending and where ... That might pinpoint [at least] the process where you should look to see what' wrong. george. On Aug 29, 2007, at 12:37 AM, Andrew Friedley wrote: I'm having a problem with the UD BTL and hoping someone might have some input to help solve it. What I'm seeing is hangs when running alltoall benchmarks with nbcbench or an LLNL program called mpiBench -- both hang exactly the same way. With the code on the trunk running nbcbench on IU's odin using 32 nodes and a command line like this: mpirun -np 128 -mca btl ofud,self ./nbcbench -t MPI_Alltoall -p 128-128 -s 1-262144 hangs consistently when testing 256-byte messages. There are two things I can do to make the hang go away until running at larger scale. First is to increase the 'btl_ofud_sd_num' MCA param from its default value of 128. This allows you to run with more procs/nodes before hitting the hang, but AFAICT doesn't fix the actual problem. What this parameter does is control the maximum number of outstanding send WQEs posted at the IB level -- when the limit is reached, frags are queued on an opal_list_t and later sent by progress as IB sends complete. The other way I've found is to play games with calling mca_btl_ud_component_progress() in mca_btl_ud_endpoint_post_send (). In fact I replaced the CHECK_FRAG_QUEUES() macro used around btl_ofud_endpoint.c:77 with a version that loops on progress until a send WQE slot is available (as opposed to queueing). Same result -- I can run at larger scale, but still hit the hang eventually. It appears that when the job hangs, progress is being polled very quickly, and after spinning for a while there are no outstanding send WQEs or queued sends in the BTL. I'm not sure where further up things are spinning/blocking, as I can't produce the hang at less than 32 nodes / 128 procs and don't have a good way of debugging that (suggestions appreciated). Furthermore, both ob1 and dr PMLs result in the same behavior, except that DR eventually trips a watchdog timeout, fails the BTL, and terminates the job. Other collectives such as allreduce and allgather do not hang -- only alltoall. I can also reproduce the hang on LLNL's Atlas machine. Can anyone else reproduce this (Torsten might have to make a copy of nbcbench available)? Anyone have any ideas as to what's wrong? Andrew ___ devel mailing list de...@open-mpi.org http://www.open-mpi.org/mailman/listinfo.cgi/devel ___ devel mailing list de...@open-mpi.org http://www.open-mpi.org/mailman/listinfo.cgi/devel
Re: [OMPI devel] UD BTL alltoall hangs
The first step will be to figure out which version of the alltoall you're using. I suppose you use the default parameters, and then the decision function in the tuned component say it is using the linear all to all. As the name state it, this means that every node will post one receive from any other node and then will start sending to every other node the respective fragment. This will lead to a lot of outstanding sends and receives. I doubt that the receive can cause a problem, so I expect the problem is coming from the send side. Do you have TotalView installed on your odin ? If yes there is a simple way to see how many sends are pending and where ... That might pinpoint [at least] the process where you should look to see what' wrong. george. On Aug 29, 2007, at 12:37 AM, Andrew Friedley wrote: I'm having a problem with the UD BTL and hoping someone might have some input to help solve it. What I'm seeing is hangs when running alltoall benchmarks with nbcbench or an LLNL program called mpiBench -- both hang exactly the same way. With the code on the trunk running nbcbench on IU's odin using 32 nodes and a command line like this: mpirun -np 128 -mca btl ofud,self ./nbcbench -t MPI_Alltoall -p 128-128 -s 1-262144 hangs consistently when testing 256-byte messages. There are two things I can do to make the hang go away until running at larger scale. First is to increase the 'btl_ofud_sd_num' MCA param from its default value of 128. This allows you to run with more procs/nodes before hitting the hang, but AFAICT doesn't fix the actual problem. What this parameter does is control the maximum number of outstanding send WQEs posted at the IB level -- when the limit is reached, frags are queued on an opal_list_t and later sent by progress as IB sends complete. The other way I've found is to play games with calling mca_btl_ud_component_progress() in mca_btl_ud_endpoint_post_send (). In fact I replaced the CHECK_FRAG_QUEUES() macro used around btl_ofud_endpoint.c:77 with a version that loops on progress until a send WQE slot is available (as opposed to queueing). Same result -- I can run at larger scale, but still hit the hang eventually. It appears that when the job hangs, progress is being polled very quickly, and after spinning for a while there are no outstanding send WQEs or queued sends in the BTL. I'm not sure where further up things are spinning/blocking, as I can't produce the hang at less than 32 nodes / 128 procs and don't have a good way of debugging that (suggestions appreciated). Furthermore, both ob1 and dr PMLs result in the same behavior, except that DR eventually trips a watchdog timeout, fails the BTL, and terminates the job. Other collectives such as allreduce and allgather do not hang -- only alltoall. I can also reproduce the hang on LLNL's Atlas machine. Can anyone else reproduce this (Torsten might have to make a copy of nbcbench available)? Anyone have any ideas as to what's wrong? Andrew ___ devel mailing list de...@open-mpi.org http://www.open-mpi.org/mailman/listinfo.cgi/devel
[OMPI devel] UD BTL alltoall hangs
I'm having a problem with the UD BTL and hoping someone might have some input to help solve it. What I'm seeing is hangs when running alltoall benchmarks with nbcbench or an LLNL program called mpiBench -- both hang exactly the same way. With the code on the trunk running nbcbench on IU's odin using 32 nodes and a command line like this: mpirun -np 128 -mca btl ofud,self ./nbcbench -t MPI_Alltoall -p 128-128 -s 1-262144 hangs consistently when testing 256-byte messages. There are two things I can do to make the hang go away until running at larger scale. First is to increase the 'btl_ofud_sd_num' MCA param from its default value of 128. This allows you to run with more procs/nodes before hitting the hang, but AFAICT doesn't fix the actual problem. What this parameter does is control the maximum number of outstanding send WQEs posted at the IB level -- when the limit is reached, frags are queued on an opal_list_t and later sent by progress as IB sends complete. The other way I've found is to play games with calling mca_btl_ud_component_progress() in mca_btl_ud_endpoint_post_send(). In fact I replaced the CHECK_FRAG_QUEUES() macro used around btl_ofud_endpoint.c:77 with a version that loops on progress until a send WQE slot is available (as opposed to queueing). Same result -- I can run at larger scale, but still hit the hang eventually. It appears that when the job hangs, progress is being polled very quickly, and after spinning for a while there are no outstanding send WQEs or queued sends in the BTL. I'm not sure where further up things are spinning/blocking, as I can't produce the hang at less than 32 nodes / 128 procs and don't have a good way of debugging that (suggestions appreciated). Furthermore, both ob1 and dr PMLs result in the same behavior, except that DR eventually trips a watchdog timeout, fails the BTL, and terminates the job. Other collectives such as allreduce and allgather do not hang -- only alltoall. I can also reproduce the hang on LLNL's Atlas machine. Can anyone else reproduce this (Torsten might have to make a copy of nbcbench available)? Anyone have any ideas as to what's wrong? Andrew