LAM MPI hangs the hamachi driver

Keith Underwood keithu@parl.clemson.edu
Mon Feb 28 19:27:51 2000


Well, I've been working on the driver a bit.  We have a few G-NIC II cards
and saw some interesting behavior.  I'm not really ready to release what
I've done to the world, but I'll send you a separate email with an updated
driver as an attachment.  If anyone else needs it right now, just ask.
I'll be finishing up in the next week or so, but I've just started trying
some other "interesting" experiments with the driver.

					Keith

On Mon, 28 Feb 2000, Anthony Caola wrote:

> 
> Hello all -
> 
> I have a parallel MPI code that appears to be hanging the hamachi driver and 
> G-NIC II card.  The symptoms right now are like this:
> 
> 1 - When the code becomes communication intensive, one of the nodes will drop 
> off the network.  ifdown and ifup'ing eth0 sometimes clears the trouble.
> 2 - If I do anything to slow my code - compile at the -g level or turn on 
> debugging in the hamachi driver (options debug=6 hamachi) - everything 
> finishes successfully, but slower.
> 
> I'm going to start tearing apart this problem and try to get to the bottom of 
> this, but was hoping someone might have seen some kind of 'overflow' type of 
> problem in the past.  Maybe all I need to do is increase one of the tunable 
> parameters. . .
> 
> Our setup is as follows:
> 
> 16 dual processor pentium xeon III's (dell 610 precision workstations) with 
> G-NIC II's running Linux 2.2.12.  We use a PowerRail 2200 switch for the 
> interconnect.
> 
> Thanks!
> 
> Anthony
> 
> 
> 
> Anthony Caola                Massachusetts Institute of Technology
> Phone:  (617) 253-6547       Department of Chemical Engineering
> Fax:    (617) 258-8224       25 Ames St., Building 66-250
> Email:  caola@mit.edu        Cambridge, MA  02139
> 
> 
>  | To unsubscribe, send mail to Majordomo@cesdis.gsfc.nasa.gov, and within the
>  |  body of the mail, include only the text:
>  |   unsubscribe this-list-name youraddress@wherever.org
>  | You will be unsubscribed as speedily as possible.
> 
> 

 | To unsubscribe, send mail to Majordomo@cesdis.gsfc.nasa.gov, and within the
 |  body of the mail, include only the text:
 |   unsubscribe this-list-name youraddress@wherever.org
 | You will be unsubscribed as speedily as possible.