[torqueusers] PVM, mpiexec, tm, and pbsdsh

Michael Gutteridge mgutteri at fhcrc.org
Tue Dec 6 16:35:00 MST 2005


I built pbsdsh with your patch.  As you said, it works, though I wasn't
able to make it work as a drop-in to PVM_RSH.  I could be using it
incorrectly.

I would set "PVM_RSH" to <path_to_pbsdsh>/pbsdsh, then start pvm.  When
I tried to add a host, I got complaints about "<hostname> not found".
It looks like pbsdsh requires "-h" to be run on a single host.  PVM
seems to issues a "system" call with the arguments being:

[0] pbsdsh
[1] <hostname>
[2] pvmd -s -d 0x0 -n<hostname> 1 8c6b34c0:8019 4080 2 8c6b34ba:0000

So I really munged the pbsdsh source to take the hostname as the first
thing on the command line.  Basically after argument processing, it just
sets targethost to "*(argv+optind)".  I called this new mess "pbsrsh",
anyway.

And that seems to work somewhat:

> momnode:(~) >pbsrsh sisternode pwd
> /home/me
> 

However, pvm's addhost still doesn't work.  The problem appears to be
that pvm sends the command line as a single argument instead of an
array.  This is probably my fault, 'cause I suck at C. I think what I
have to do is re-parameterize the contents of argv[] before passing to
tm_spawn.

Is it worth munging up pbsdsh to be a syntactically-equivalent cousin to
rsh, as in "pbsdsh <hostname> [arguments] <program>"?

Anyway, thanks for the patches.  I'm getting a better Idea of how TM
works through all this.  Writing a pvm hoster might not be too bad....

Michael


On Tue, 2005-12-06 at 02:06 -0800, Garrick Staples wrote:
> On Mon, Dec 05, 2005 at 11:56:43PM -0800, Garrick Staples alleged:
> > On Mon, Dec 05, 2005 at 11:05:34PM -0800, mgutteri at fhcrc.org alleged:
> > > 
> > > I was reviewing the pvmd3 manpage and saw this:
> > > 
> > >   The following options are used by the master pvmd when starting slaves 
> > >   and are only of interest to someone writing a hoster.  Don?t just go 
> > >   using them, now.
> > >        -s     Start pvmd in slave mode.  Hostfile cannot be used, five   
> > >               additional parameters  must  be  supplied:  master pvmd 
> > >               index, master IP, master MTU, slave pvmd index, and slave IP.
> > > 
> > > This suggests to me that perhaps one could use pbsdsh to start the pvmd slaves
> > > this way.  I am uncertain about how to get the parameters that would be
> > > provided to the slave processses, though I'm going to look into this "hoster"
> > > business though...
> > > 
> > > However, Garrick may well beat me to it if he can clone rsh functionality into
> > > pbsdsh 8-)..
> > 
> > Check this out... it needs some polishing, but it works.
> 
> Here we go.  This is working really well.  If you don't need stdio,
> there's almost an rsh replacement.
> 
> _______________________________________________
> torqueusers mailing list
> torqueusers at supercluster.org
> http://www.supercluster.org/mailman/listinfo/torqueusers
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: This is a digitally signed message part
Url : http://www.supercluster.org/pipermail/torqueusers/attachments/20051206/b38377c2/attachment.bin


More information about the torqueusers mailing list