Open MPI logo

Open MPI User's Mailing List Archives

  |   Home   |   Support   |   FAQ   |   all Open MPI User's mailing list

Subject: Re: [OMPI users] Problem with running MPI programs on machines with multiple interfaces
From: Avinash Malik (amalik_at_[hidden])
Date: 2011-01-24 11:27:30


Hello,
        
        Please don't worry about this for now, the problem stems from
        iptable rules. But, I still think putting usb0 into the reject
        list should disable the ip-address associated with it.

Regards,
        
>>>>> "Avinash" == Avinash Malik <amalik_at_[hidden]> writes:

    Avinash> Hello,
        
    Avinash> I have two mahcines each having 3 live interfaces:
    Avinash> lo, eth0 (interanet) and usb0 (internet). eth0 cannot
    Avinash> access usb0 on the other machine (and vice-veras). Now,
    Avinash> when I try to run the MPI program with these two hosts I
    Avinash> cannot get any output, even --mca btl_base_verbose 30 does
    Avinash> not give any output. If I set hostfile to have only
    Avinash> localhost, then everything runs fine.

    Avinash> I tried out the same code and hostfile with two
    Avinash> other machines with two interfaces: lo and eth1, which can
    Avinash> access each other. The program runs fine on these machines.

    Avinash> Next, I setup btl_tcp_if_exclude to lo,usb0 (on the
    Avinash> first arch) and also the ip-address/mask, but this does not
    Avinash> work either. When I run the program on one machine and do
    Avinash> "ps aux | grep mpi" on the other I can see --hnp-uri being
    Avinash> set to the usb0's ip-address, which it should not, because
    Avinash> I have set usb0 to be exluded in the btl_tcp_if_exclude
    Avinash> list. So, what exactly am I doing wrong here?

    Avinash> I read the otimization FAQ and saw how openmpi
    Avinash> builds the bipartite graphs for connection. But, as I said
    Avinash> before, eth0 cannot access usb0's ip and vice-versa, how
    Avinash> can I get rid of the usb0 ip-address showing up in
    Avinash> --hnp-uri, because this is the only difference between the
    Avinash> working and the non-working archs.

    Avinash> Regards, -- Avinash Malik
    Avinash> _______________________________________________ users
    Avinash> mailing list users_at_[hidden]
    Avinash> http://www.open-mpi.org/mailman/listinfo.cgi/users

-- 
Avinash Malik