[Swift-devel] Re: InternalHostname in sites file

[Michael Wilde]

Thanks for clarifying. Jon and/or David, can you address this with a cookbook entry on Coasters that heads towards a users guide section?

We should tell users what they can run on their cluster (eg ping or telnet-style connect tests) to validate the setting of internalHostName.

- Mike



----- Original Message -----
> Right. If the head node has multiple network interfaces, only one of
> which is visible from the worker nodes.
> 
> The choice of which interface is the one that worker nodes can connect
> to is a matter of the particular cluster. It's not particularly easy
> to
> have an automated mechanism that figures it out. We tried some scheme
> to
> pass all the interface addresses to the worker and let it try to
> connect
> to all of them in order, but that didn't work very well. Of course,
> there might be a scheme that works, but I didn't want to spend too
> much
> time on that.
> 
> So that's why it's needed. To clarify to the workers which exact
> interface on the head node they are to try to connect to.
> 
> Mihael
> 
> On Thu, 2011-06-02 at 14:10 -0500, Jonathan S Monette wrote:
> > Mihael,
> >    I believe we have talked about this before but why is it
> >    necessary
> > for an InternalHostname to be specified for PADS? I know that the
> > address that coasters connects to is wrong but I do not remember why
> > that was. Could you give an explanation on why internalHostname
> > needs
> > to be set?
> >
> > --
> >
> >
> > Any intelligent fool can make things bigger and more complex... It
> > takes a touch of genius - and a lot of courage to move in the
> > opposite
> > direction.
> >
> > - Albert Einstein
> >
> >
> >
> 
> 
> _______________________________________________
> Swift-devel mailing list
> Swift-devel at ci.uchicago.edu
> http://mail.ci.uchicago.edu/mailman/listinfo/swift-devel

-- 
Michael Wilde
Computation Institute, University of Chicago
Mathematics and Computer Science Division
Argonne National Laboratory