Brady,
You might want to look at this thread on the Beowulf Mailling list (you
most
probably already have):
http://www.beowulf.org/pipermail/beowulf/2004-July/010311.html
I will have to set up a Gentoo cluster in the near future and would
definately like to keep all options opened as far as RSH/SSH connections are
concerned. So relaying your experience with RSH would definately be
appreciated.
Thanks,
Eric
Le 13 Novembre 2005 18:06, Brady Catherman a écrit :
> We have tons of bio apps that use stdout/stderr to send data. These
> programs can send gigs through SSH. We would prefer if it didn't have
> to get encrypted, passed over our secure network, and then be
> decrypted. It generates excess work for our Master node. Maintaining
> the keys also becomes an issue. We have 4 different platforms here.
> one little hickup with ssh means that everybody's accounts have to be
> fixed. If we use global rsh then we only have to fix it on the master/
> slave images. Plus is just makes PBS run smoother.
>
> Anyways, I think I got rlogin working. I can login password-less if I
> rlogin to localhost. Still can't do anything remotely though. Got
> BSU's cluster god to help me out a bit.. he's awesome with this
> stuff. When I get a final solution mind if I post to the list? The
> net seems to be full of people asking about this and nobody actually
> getting it to work.
>
>
> On Nov 13, 2005, at 2:59 PM, Brian D. Ropers-Huilman wrote:
>
> > -----BEGIN PGP SIGNED MESSAGE-----
> > Hash: RIPEMD160
> >
> > Might I ask what the intended use is?
> >
> > If you are pushing out commands, there are other tools for that.
> >
> > If you are deploying a job, we use SSH on our 512 node system just
> > fine.
> >
> > Brady Catherman said the following on 2005-11-13 16:46:
> >> ssh -vvv doesn't tell me where the failure happens. For now I have
> >> given up on ssh and am just working on RSH. Seems that nobody
> >> knows how
> >> to set RSH up on clusters. Problem is that 500+ ssh sessions can
> >> eat p
> >> CPU time on the master in a hurry. Problem is that RSH has no
> >> debugging
> >> information at all.. Closest I have gotten is getting xinetd to
> >> warn on
> >> startup errors. Whats a guy got to do to get information from his
> >> daemons? =)
> >>
> >> (BTW, Why does everybody shrug off rsh? For clusters its perfectly
> >> acceptable and if the cluster is large enough then its nearly
> >> required.)
> >>
> >>
> >> On Nov 13, 2005, at 2:30 PM, Donnie Berkholz wrote:
> >>
> >> Brady Catherman wrote:
> >> | And whats up with the SSH/RSH developers not giving some form of
> >> | debugging method? I mean.. The only thing you get for logging is
> >> login
> >> | notices. I can't find any way to get useful debugging out of RSH.
> >>
> >> Does `ssh -vvv` not give any useful debugging info?
> >>
> >> | Whats most frustrating is that this is effortless on Fedora. Very
> >> | little file modification is needed.
> >>
> >> It might be interesting to look at any patches they apply that
> >> they've
> >> failed to send upstream.
> >>
> >> Thanks,
> >> Donnie
> > - --
> > [email protected] mailing list
> >>>
> >
> > - --
> > Brian D. Ropers-Huilman, Director .:. High Performance
> > Computing
> > Networking, Infrastructure, & Research / Information Technology
> > Support
> > Frey Computing Services, Rm. 323
> > [EMAIL PROTECTED]
> > Louisiana State University +1 225.578.3272
> > (V)
> > Baton Rouge, LA 70803-1900 USA +1 225.578.5362
> > (F)
> > http://www.lsu.edu/ http://
> > www.cct.lsu.edu/
> > -----BEGIN PGP SIGNATURE-----
> > Version: GnuPG v1.4.1 (GNU/Linux)
> > Comment: Using GnuPG with Thunderbird - http://enigmail.mozdev.org
> >
> > iD8DBQFDd8U5wRr6eFHB5lgRA1ghAJ0YiZTkLNsEzZUooPoLG2y1CSN4ZQCgy5OX
> > uVKS8igs67EPaJRDS+XTG4g=
> > =GY1S
> > -----END PGP SIGNATURE-----
> > --
> > [email protected] mailing list
> >
>
--
Eric Thibodeau
Neural Bucket Solutions Inc.
T. (514) 736-1436
C. (514) 710-0517
--
[email protected] mailing list
