[R-sig-hpc] snow, socket cluster: problem with path to rscript

[Dirk Eddelbuettel]

On 17 April 2009 at 12:53, Steve Weston wrote:
| On Fri, Apr 17, 2009 at 11:28 AM, Dirk Eddelbuettel <edd at debian.org> wrote:
| 
| > For what it is worth, I cannot do that either on Ubuntu at work, yet snow
| > works just fine:
| >
| > edd at l1:~$ telnet l2 10187
| > Trying xxx.xx.50.99...
| > telnet: Unable to connect to remote host: Connection refused
| > edd at l1:~$ telnet l1 10187
| > Trying xxx.xx.50.97...
| > telnet: Unable to connect to remote host: Connection refused
| 
| Perhaps I'm confused, but it looks like your telnets are going the
| wrong way.  The telnets are supposed to be acting like the slave
| processes connecting back to the master.  So they would be
| running on l1 and l2, but connecting back to the master, where
| "socketConnection(port=10871, server=TRUE)" is executing.

Correct, Luke pointed that out is his post that arrived at about the same
time.  When there is server to connect to, it should be little wonder that
the telnet process fails....   Sloppy thinking on my part.

| At this point, I would be running R under gdb, and maybe doing
| some packet sniffing.  I'm not sure what else to suggest.

Yes, Matthieu should probably try some 'manual' R socket connection example
to prove that these work in his setup before trying to employ them with snow.

Dirk

-- 
Three out of two people have difficulties with fractions.