Fwd: [ofa-general] Re: [OMPI users] openMPI over uDAPL doesn't work

[Jeff Squyres]

Re-forwarding to OMPI list; because of the OMPI list anti-spam  
checks, Arlin's post didn't make it through to our list when he  
originally posted.



Begin forwarded message:

> From: Arlin Davis <ardavis at ichips.intel.com>
> Date: May 8, 2007 3:09:02 PM EDT
> To: Jeff Squyres <jsquyres at cisco.com>
> Cc: Open MPI Users <users at open-mpi.org>, OpenFabrics General  
> <general at lists.openfabrics.org>
> Subject: Re: [ofa-general] Re: [OMPI users] openMPI over uDAPL  
> doesn't work
>
> Jeff Squyres wrote:
>
>> I'm forwarding this to the OpenFabrics general list -- as it just   
>> came up the other day, we know that Open MPI's UDAPL support works  
>> on  Solaris, but we have done little/no testing of it on OFED (I   
>> personally know almost nothing about UDPAL).
>>
>> Can the UDAPL OFED wizards shed any light on the error messages  
>> that  are listed below?  In particular, these seem to be worrysome:
>>
>>>  setup_listener Permission denied
>>
>>  setup_listener Address already in use
>
> These failures are from rdma_cm_bind indicating the port is already  
> bound to this IA address. How are you creating the service point?
> dat_psp_create or dat_psp_create_any? If it is psp_create_any then  
> you will see some failures until it  gets to a free port. That is  
> normal. Just make sure your create call returns DAT_SUCCESS.
>
>>>  create_qp Address already in use
>>
> This is a real problem with the bind, port is already in use. Not  
> sure why this would fail since the current version of OFED uDAPL  
> uses a wildcard port when binding and uses the address from the  
> open;  I remember an issue a while back with rdma_cm and wildcard  
> ports. What version of OFED are you using?
>
> -arlin


-- 
Jeff Squyres
Cisco Systems