Bruno, No, I was looking for the earlier part of the log where mpiexec launches proxies. You may cut out the dump of environment variables if those are sensitive. -- Hui ________________________________ From: [email protected] <[email protected]> Sent: Thursday, November 17, 2022 8:35 AM To: Zhou, Hui <[email protected]>; [email protected] <[email protected]> Subject: Re: [mpich-discuss] wrong number of processes on host Is this log OK for you ? [pgid: 0] got PMI command: cmd=barrier_in [pgid: 0] got PMI command: cmd=put sharedFilename[0]=/dev/shm/mpich_shar_tmpAOPuur [pgid: 0] got PMI command: cmd=barrier_in [pgid: 0] got PMI command: cmd=put sharedFilename[4]=/dev/shm/mpich_shar_tmp76zj4w [pgid: 0] got PMI command: cmd=barrier_in PMI response to fd 13 pid 4: cmd=keyval_cache sharedFilename[2]=/dev/shm/mpich_shar_tmppFa2Wf sharedFilename[0]=/dev/shm/mpich_shar_tmpAOPuur sharedFilename[4]=/dev/shm/mpich_shar_tmp76zj4w PMI response to fd 12 pid 4: cmd=keyval_cache sharedFilename[2]=/dev/shm/mpich_shar_tmppFa2Wf sharedFilename[0]=/dev/shm/mpich_shar_tmpAOPuur sharedFilename[4]=/dev/shm/mpich_shar_tmp76zj4w PMI response to fd 14 pid 4: cmd=keyval_cache sharedFilename[2]=/dev/shm/mpich_shar_tmppFa2Wf sharedFilename[0]=/dev/shm/mpich_shar_tmpAOPuur sharedFilename[4]=/dev/shm/mpich_shar_tmp76zj4w PMI response to fd 13 pid 4: cmd=barrier_out PMI response to fd 12 pid 4: cmd=barrier_out PMI response to fd 14 pid 4: cmd=barrier_out [pgid: 0] got PMI command: cmd=put P2-businesscard=description#l-neobi-1$port#39551$ifname#10.193.21.24$ P3-businesscard=description#l-neobi-1$port#49083$ifname#10.193.21.24$ [pgid: 0] got PMI command: cmd=barrier_in [pgid: 0] got PMI command: cmd=put P0-businesscard=description#l-neobi-4$port#57647$ifname#10.193.21.65$ P1-businesscard=description#l-neobi-4$port#46287$ifname#10.193.21.65$ [pgid: 0] got PMI command: cmd=barrier_in [pgid: 0] got PMI command: cmd=put P4-businesscard=description#l-neobi-3$port#48597$ifname#10.193.21.26$ P5-businesscard=description#l-neobi-3$port#47573$ifname#10.193.21.26$ [pgid: 0] got PMI command: cmd=barrier_in PMI response to fd 13 pid 5: cmd=keyval_cache P2-businesscard=description#l-neobi-1$port#39551$ifname#10.193.21.24$ P3-businesscard=description#l-neobi-1$port#49083$ifname#10.193.21.24$ P0-businesscard=description#l-neobi-4$port#57647$ifname#10.193.21.65$ P1-businesscard=description#l-neobi-4$port#46287$ifname#10.193.21.65$ P4-businesscard=description#l-neobi-3$port#48597$ifname#10.193.21.26$ P5-businesscard=description#l-neobi-3$port#47573$ifname#10.193.21.26$ PMI response to fd 12 pid 5: cmd=keyval_cache P2-businesscard=description#l-neobi-1$port#39551$ifname#10.193.21.24$ P3-businesscard=description#l-neobi-1$port#49083$ifname#10.193.21.24$ P0-businesscard=description#l-neobi-4$port#57647$ifname#10.193.21.65$ P1-businesscard=description#l-neobi-4$port#46287$ifname#10.193.21.65$ P4-businesscard=description#l-neobi-3$port#48597$ifname#10.193.21.26$ P5-businesscard=description#l-neobi-3$port#47573$ifname#10.193.21.26$ PMI response to fd 14 pid 5: cmd=keyval_cache P2-businesscard=description#l-neobi-1$port#39551$ifname#10.193.21.24$ P3-businesscard=description#l-neobi-1$port#49083$ifname#10.193.21.24$ P0-businesscard=description#l-neobi-4$port#57647$ifname#10.193.21.65$ P1-businesscard=description#l-neobi-4$port#46287$ifname#10.193.21.65$ P4-businesscard=description#l-neobi-3$port#48597$ifname#10.193.21.26$ P5-businesscard=description#l-neobi-3$port#47573$ifname#10.193.21.26$ PMI response to fd 13 pid 5: cmd=barrier_out PMI response to fd 12 pid 5: cmd=barrier_out PMI response to fd 14 pid 5: cmd=barrier_out On 17/11/2022 15:30, Zhou, Hui wrote: That is strange. Try pass -v option (i.e. mpiexec -v ...) to obtain a console log when that happens. -- Hui ________________________________ From: bruno via discuss <[email protected]><mailto:[email protected]> Sent: Thursday, November 17, 2022 8:11 AM To: [email protected]<mailto:[email protected]> <[email protected]><mailto:[email protected]> Cc: [email protected]<mailto:[email protected]> <[email protected]><mailto:[email protected]> Subject: [mpich-discuss] wrong number of processes on host Hi, I am using mpich on a hadoop cluster with yarn. It is not a smooth integration but it is working. Flowing an old post, I am using the manual launcher and the option -disable-hostname-propagation (https://lists.mpich.org/mailman/htdig/devel/2016-July/000717.html) The command line to launch by binary is : mpiexec -launcher manual -disable-hostname-propagation -n 6 -f ./hosts /path/to/my/bin And the file hosts contains host1:2 host2:2 host3:2 I expect to have 2 processes on each host but most of the time it launches 3 processes on host1, 1 on host2 and 2 on host3 Any ideas? Bruno _________________________________________________________________________________________________________________________ Ce message et ses pieces jointes peuvent contenir des informations confidentielles ou privilegiees et ne doivent donc pas etre diffuses, exploites ou copies sans autorisation. Si vous avez recu ce message par erreur, veuillez le signaler a l'expediteur et le detruire ainsi que les pieces jointes. Les messages electroniques etant susceptibles d'alteration, Orange decline toute responsabilite si ce message a ete altere, deforme ou falsifie. Merci. This message and its attachments may contain confidential or privileged information that may be protected by law; they should not be distributed, used or copied without authorisation. If you have received this email in error, please notify the sender and delete this message and its attachments. As emails may be altered, Orange is not liable for messages that have been modified, changed or falsified. Thank you. _______________________________________________ discuss mailing list [email protected]<mailto:[email protected]> To manage subscription options or unsubscribe: https://lists.mpich.org/mailman/listinfo/discuss _________________________________________________________________________________________________________________________ Ce message et ses pieces jointes peuvent contenir des informations confidentielles ou privilegiees et ne doivent donc pas etre diffuses, exploites ou copies sans autorisation. Si vous avez recu ce message par erreur, veuillez le signaler a l'expediteur et le detruire ainsi que les pieces jointes. Les messages electroniques etant susceptibles d'alteration, Orange decline toute responsabilite si ce message a ete altere, deforme ou falsifie. Merci. This message and its attachments may contain confidential or privileged information that may be protected by law; they should not be distributed, used or copied without authorisation. If you have received this email in error, please notify the sender and delete this message and its attachments. As emails may be altered, Orange is not liable for messages that have been modified, changed or falsified. Thank you.