Re: [4] Error: Unable to open a GM port !

On Sun, 2003-11-09 at 10:54, Camilo Ruiz Méndez wrote:
> hello;
>   i am having problems with the mpich.
> i was debugging a problem and then [4] Error: Unable to open a GM port !
> 
> i am running /gm-1.6.3_Linux
> 
> The output of my script is  the following, basically i run
> mpirun.ch_gm -v -np $PROCESOS hello.mpichgm
> 
> -----
>   ./scriptHello
> Program binary is: /home/camilo/calculo/mpi3d/hello.mpichgm
> Machines file is /opt/mpich-1.2.4..8a/share/machines.ch_gm.LINUX
> Shared memory for intra-nodes coms is enabled.
> GM receive mode used: polling.
> 6 processes will be spawned:
>          Process 0 (/home/camilo/calculo/mpi3d/hello.mpichgm ) on 
> neperiano.usal.es
>          Process 1 (/home/camilo/calculo/mpi3d/hello.mpichgm ) on 
> neperiano.usal.es
>          Process 2 (/home/camilo/calculo/mpi3d/hello.mpichgm ) on 
> neperiano.usal.es
>          Process 3 (/home/camilo/calculo/mpi3d/hello.mpichgm ) on 
> neperiano.usal.es
>          Process 4 (/home/camilo/calculo/mpi3d/hello.mpichgm ) on 
> cat02.usal.es
>          Process 5 (/home/camilo/calculo/mpi3d/hello.mpichgm ) on 
> cat02.usal.es
> Open a socket on neperiano.usal.es...
> Got a first socket opened on port 8000.
> Got a second socket opened on port 8001.
> Shared memory file: /tmp/gmpi_shmem-470713.tmp
> 
> rsh neperiano.usal.es cd /home/camilo/calculo/mpi3d ; env  
> GMPI_MASTER=neperiano.usal.es GMPI_PORT1=8000 GMPI_PORT2=8001 
> GMPI_SHMEM=1 GMPI_MAGIC=470713 GMPI_ID=0 GMPI_NP=6 GMPI_BOARD=-1 
> /home/camilo/calculo/mpi3d/hello.mpichgm
> rsh -n neperiano.usal.es cd /home/camilo/calculo/mpi3d ; env  
> GMPI_MASTER=neperiano.usal.es GMPI_PORT1=8000 GMPI_PORT2=8001 
> GMPI_SHMEM=1 GMPI_MAGIC=470713 GMPI_ID=1 GMPI_NP=6 GMPI_BOARD=-1 
> /home/camilo/calculo/mpi3d/hello.mpichgm
> rsh -n neperiano.usal.es cd /home/camilo/calculo/mpi3d ; env  
> GMPI_MASTER=neperiano.usal.es GMPI_PORT1=8000 GMPI_PORT2=8001 
> GMPI_SHMEM=1 GMPI_MAGIC=470713 GMPI_ID=2 GMPI_NP=6 GMPI_BOARD=-1 
> /home/camilo/calculo/mpi3d/hello.mpichgm
> rsh -n neperiano.usal.es cd /home/camilo/calculo/mpi3d ; env  
> GMPI_MASTER=neperiano.usal.es GMPI_PORT1=8000 GMPI_PORT2=8001 
> GMPI_SHMEM=1 GMPI_MAGIC=470713 GMPI_ID=3 GMPI_NP=6 GMPI_BOARD=-1 
> /home/camilo/calculo/mpi3d/hello.mpichgm
> rsh -n cat02.usal.es cd /home/camilo/calculo/mpi3d ; env  
> GMPI_MASTER=neperiano.usal.es GMPI_PORT1=8000 GMPI_PORT2=8001 
> GMPI_SHMEM=1 GMPI_MAGIC=470713 GMPI_ID=4 GMPI_NP=6 GMPI_BOARD=-1 
> /home/camilo/calculo/mpi3d/hello.mpichgm
> rsh -n cat02.usal.es cd /home/camilo/calculo/mpi3d ; env  
> GMPI_MASTER=neperiano.usal.es GMPI_PORT1=8000 GMPI_PORT2=8001 
> GMPI_SHMEM=1 GMPI_MAGIC=470713 GMPI_ID=5 GMPI_NP=6 GMPI_BOARD=-1 
> /home/camilo/calculo/mpi3d/hello.mpichgm
> MPI Id 3 is using GM port 2, board 0, GM_id 1.
> MPI Id 2 is using GM port 4, board 0, GM_id 1.
> MPI Id 1 is using GM port 5, board 0, GM_id 1.
> MPI Id 0 is using GM port 6, board 0, GM_id 1.
> MPI Id 5 is using GM port 7, board 0, GM_id 14.
> [4] Error: Unable to open a GM port !
> Received SIGINT. Cleaning up...
> 
> If i run the info board i get the following.
> 
>   /opt/gm-1.6.3_Linux/binary/bin/gm_board_info
> GM build ID is "1.6.3_Linux 
> [email protected]:/opt/gm-1.6.3_Linux Fri Oct 18 12:50:14 CEST 
> 2002."
> 
> Board number 0:
>    lanai_clockval    = 0x082082a0
>    lanai_cpu_version = 0x0900 (LANai9.0)
>    lanai_board_id    = 00:60:dd:7e:f6:e3
>    lanai_sram_size   = 0x00200000 (2048K bytes)
>    max_lanai_speed   = 134 MHz
>    product_code      = 111
>    serial_number     = 122102
>      (should be labeled: "M3F-PCI64B-2-122102")
> LANai time is 0x5881b0cbc0 ticks, or about 2832 minutes since reset.
> This is node 1 (neperiano.usal.es)  node_type=0
> Board has room for 8 ports,  3000 nodes/routes,  16384 cache entries
>            Port token cnt: send=29, recv=248
> Port: Status  PID
>     0:   BUSY 28964  (this process [gm_board_info])
>     1:   BUSY  1030
>     3:   BUSY    -1

Go look on node neperiano.usal.es and see what process
is running as pid 1030.  You probably have a runaway
process started from a different job.  The next job
doesn't have enough ports available to start as
you have configured your machinefile.

Craig