RE: gmID conflicts on static mapping

Leonardo Alves de Paula e Silva <[email protected]> Mon, 14 Apr 2003 12:30:53 -0300 (BRT)
Newsgroups gmane.network.myrinet.general
Message-ID <[email protected]>
Thanks for your help, Ruth.

No. I've read README-linux and I did exactly what it is written 
there since the first time. 

But ok, assuming that I've run mapper on two machines by mistake, 
removing static.map, static,routes and static.host from that 
mistaken node, re-running mapper just on one node and copying the 
right files then re-running file_mapper, the problem will 
disappear?     

Thanks.

-- 
Leonardo Alves de Paula e Silva                [email protected]
Msc. Student                       Institute of Informatics/UFRGS
GPPD                                 http://www-gppd.inf.ufrgs.br

Address:
Av. Bento Gonçalves, 9500 Bloco IV    PO Box 15064      91501-970
Phone: +55 51 3316-6168 (Lab. 207)          Fax: +55 51 3316-7308 
Porto Alegre - RS                                          Brazil

On 10/04/03  22:28, Ruth Sivilotti wrote:

> Leonardo,
> 
> The ID conflict usually arises when there
> are conflicting mapper *.hosts files on two
> machines in the cluster.
> Have you run the active mapper on more than
> one machine?
> 
> See:
> 
>  http://www.myri.com/cgi-bin/fom?file=131
> 
> And if that doesn't help, let us know!
> 
> --Ruth
> 
> > -----Original Message-----
> > From: [email protected] [mailto:[email protected]] On 
> > Behalf Of Leonardo Alves de Paula e Silva
> > Sent: Thursday, April 10, 2003 2:50 PM
> > To: [email protected]
> > Subject: [Myrinet] gmID conflicts on static mapping
> > 
> > 
> > Hi,
> > 
> > I'm using static mapping to route the messges on our Myrinet and I 
> > had a weird problem with file_mapper.
> > 
> > I have 16 nodes connected ad 15 have done the mapping without 
> > problems. I have syncronized all nodes using rsynch, so the 
> > system installation is the same in all nodes.
> > 
> > On this problematic one, the command 
> > 
> > l10:/usr/local/gm-1.6.3# ./file_mapper ../etc/gm/file.args. 
> > 
> > never ends and when I press ^C it outputs segmentation fault.
> > 
> > I have decided to look the routing table unsing gm_boar_info 
> > output in all nodes. 
> >  
> > On l10, the problematic, its routing table don't have a line to 
> > the mapper, l1. Instead, the gmID 1 is attributed to l10 mac 
> > address and it shows a message which I could not understand:
> > 
> > Route table for this node follows:
> > The mapper 48-bit ID was: 00:60:dd:7f:26:ff
> > gmID MAC Address                                 gmName Route
> > ---- ----------------- -------------------------------- 
> > ---------------------
> >    1 00:60:dd:7f:26:ff *** node_id->unique_id->node_id is 
> > 0x5. ***                       
> > localhost 80 (this node) (mapper)
> > 
> > On the other nodes, l1 node have gmID of 1 and it is showed on the 
> > first line. That is the output:  
> > 
> > Route table for this node follows:
> > The mapper 48-bit ID was: 00:60:dd:7f:27:16
> > gmID MAC Address                                 gmName Route
> > ---- ----------------- -------------------------------- 
> > ---------------------
> >    1 00:60:dd:7f:27:16                        localhost 80 (this 
> > node) (mapper) 
> >  
> > I have tried to map again and I have this output of 
> > 
> > l1:/usr/local/gm-1.6.3/sbin# ./mapper ../etc/gm/static.args 
> > 18:43:54 starting to map. My clock says Thu Apr 10 18:43:54 2003
> > 18:43:55 id conflict gmId 1, address 0060dd7f26ff
> > 18:43:55 this id has been seen before on host localhost
> > 18:43:56 map time was 2869039 lanai ticks, which is 1434 
> > millisecs.
> > 18:43:56 found 16 hosts and 6 switches (22 nodes)
> > 18:43:56 localhost has an id conflict
> > 18:43:56 configuring
> > 18:43:56 cleared conflict on  localhost (address 0060dd7f26ff)
> > 18:43:56 route computation took 0 milliseconds
> > 18:43:56 configuration took 20 milliseconds
> > 18:43:56 map-once or configure-once enabled. Done mapping.
> > 
> > How can I avoid this gmId conflict?
> > 
> > -- 
> > Leonardo Alves de Paula e Silva                [email protected]
> > Msc. Student                       Institute of Informatics/UFRGS
> > GPPD                                 http://www-gppd.inf.ufrgs.br
> > 
> > Address:
> > Av. Bento Gonçalves, 9500 Bloco IV    PO Box 15064      91501-970
> > Phone: +55 51 3316-6168 (Lab. 207)          Fax: +55 51 3316-7308 
> > Porto Alegre - RS                                          Brazil
> > 
> > 
> > _______________________________________________
> > Myrinet mailing list
> > [email protected]
> > http://email.osc.edu/mailman/listinfo/myrinet
> > 
> > 
> 
>