RE: Error - Unable to open a GM Port.

Make sure all your cards have link.  You'll get that if you use the newer MPICH (1.2.5..8 or later) and one of the nodes in the job doesn't have Myrinet connectivity.

   Chris


> -----Original Message-----
> From: Sid Carter [mailto:[email protected]] 
> Sent: Monday, September 15, 2003 10:07 AM
> To: Craig Tierney
> Cc: [email protected]
> Subject: Re: [Myrinet] Error - Unable to open a GM Port.
> 
> 
> On Mon, 2003-09-15 at 19:30, Craig Tierney wrote:
> > > glibc-2.2.4-32.3
> > > gm-1.6.4-1
> > > mpif90 for 1.2.5 (release) of : 2003/01/13 16:21:53
> > > 
> > > "Error: Unable to open a GM port !"
> > Run gm_board_info and see if you have other processes
> > still running on the node.  There is a finite number
> > of available ports.  There might be some runaway processes 
> using them.
> 
> Craig,
> 
> We have already done that and we do have free ports 
> available. Enclosing more info, FYI.
> 
> ==============================
> [gopal@hydra3 jason]$ more qgrid2.o822 /* this is the output of the 
> application I ran */
> Generating public/private rsa1 key pair.
> Your identification has been saved in /tmp/keys.vUfXwg/tmpid. 
> Your public key has been saved in /tmp/keys.vUfXwg/tmpid.pub. 
> The key fingerprint is: 
> e4:f6:05:d4:43:8a:64:23:5e:9d:84:ba:7f:fa:a9:67 
> gopal@compute-0-8 spawn /usr/bin/ssh-add 
> /tmp/keys.vUfXwg/tmpid Enter passphrase for 
> gopal@compute-0-8: Identity added: /tmp/keys.vUfXwg/tmpid 
> (gopal@compute-0-8)
> 
> [2] Error: Unable to open a GM port !
> [0] Error: Unable to open a GM port !
> [6] Error: Unable to open a GM port !
> [3] Error: Unable to open a GM port !
> [5] Error: Unable to open a GM port !
> [1] Error: Unable to open a GM port !
> [7] Error: Unable to open a GM port !
> [4] Error: Unable to open a GM port !
> [gopal@hydra3 jason]$ ssh compute-0-8
> Last login: Thu Sep 11 10:58:40 2003 from hydra3
> Rocks Compute Node
> Rocks 2.3.2 (Annapurna)
> Kickstarted on Thu Jul 31 13:15:23 SGT 2003
> Profile built 13:07:01 31 Jul 2003
> [gopal@compute-0-8 gopal]$ /usr/sbin/gm_board_info
> GM build ID is "1.6.4_Linux_and_AIX 
> @compute-0-8:/usr/src/redhat/BUILD/gm-1.6.4_Linux_and_AIX Thu Jul 31 
> 13:19:40 SGT 2003."
> 
> Board number 0:
>    lanai_clockval    = 0x082082a0
>    lanai_cpu_version = 0x0900 (LANai9.0)
>    lanai_board_id    = 00:60:dd:7e:f6:fc
>    lanai_sram_size   = 0x00200000 (2048K bytes)
>    max_lanai_speed   = 134 MHz
>    product_code      = 111
>    serial_number     = 122077
>      (should be labeled: "M3F-PCI64B-2-122077")
> LANai time is 0x696f222413a ticks, or about 53982 minutes 
> since reset. This is node 15 (compute-0-8)  node_type=0 Board 
> has room for 8 ports,  3000 nodes/routes,  16384 cache entries
>            Port token cnt: send=61, recv=248
> Port: Status  PID
>     0:   BUSY  8664  (this process [gm_board_info])    /* 
> port is being 
> used only by gm_board_info */
> Route table for this node follows:
> The mapper 48-bit ID was: 00:60:dd:7e:f7:08
> gmID MAC Address                                 gmName Route
> ---- ----------------- -------------------------------- 
> ---------------------
>    11 00:60:dd:7e:f7:02                      compute-0-3 b6
>    12 00:60:dd:7e:f7:06                      compute-0-6 b9
>    14 00:60:dd:7e:f7:0e                      compute-0-7 bd
>    15 00:60:dd:7e:f6:fc                      compute-0-8 80 
> (this node)
>    17 00:60:dd:7e:f7:01                     compute-0-10 be
>    19 00:60:dd:7e:f6:fe                     compute-0-13 bb
>    20 00:60:dd:7e:f7:08                     compute-0-14 bc (mapper)
>    21 00:60:dd:7e:f9:84                           hydra3 b3
>    99 00:60:dd:7e:f9:6b                      compute-0-2 b5
>   102 00:60:dd:7e:f6:f9                      compute-0-1 b4
>   103 00:60:dd:7e:f7:10                      compute-0-5 b8
>   105 00:60:dd:7e:f8:36                      compute-0-4 b7
>   106 00:60:dd:7e:f8:4e                      compute-0-9 bf
>   107 00:60:dd:7e:f9:7c                     compute-0-11 ba
>   108 00:60:dd:7e:f9:83                     compute-0-12 b2
> ===============================
> 
> Any help regarding the same would be appreciated.
> 
> Thanks
> Regards
>            Sid
> _______________________________________________
> Myrinet mailing list
> [email protected]
> http://email.osc.edu/mailman/listinfo/myrinet
>