Re: [gentoo-cluster] Gentoo clustering "?? la" www.clustermatic.org (PXE booted nodes)

[Brady Catherman]

Ouch.. that would totally kill us here.

We have four types of programs:
CPU hogs that just eat up cpu time but don't do much with file access  
or communications.
NFS hogs, that load huge files off NFS shares and then process them  
(We are talking 4TB here)
Communication hogs that are chatty as all get out but don't do much  
with files (GA and such)
And scripts that run tons of local commands on data.


We had to go with a generic approach. Plus our NFS daemons already  
get hung pretty easily so  we try and reduce the NFS load as much as  
we can.



On Nov 21, 2005, at 10:38 AM, Donnie Berkholz wrote:

> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
>
> Brady Catherman wrote:
> | My experience with NFS mounted roots is that they can bombard your
> | network with packets. A simple little script can manage to generate
> | enough traffic to actually slow down other services just by hitting
> | tons of services. Plus if you have spoolers and such you end up
> | generating a ton of traffic or memory. The advantage to a local
> install
> | is that you can cut down on the network traffic drastically.
> Granted,
> | if your applications are all embarrassingly parallel and  don't
> do a ton
> | of disk IO then NFS root works great.. Many of the  applications
> we use
> | here would utterly destroy the network if run  from a NFS mounted
> root..
> | The advantage of rebuilding is consistency  without the
> disadvantage of
> | NFS roots.
>
> That's why you should have a separate 100Mb network for NFS traffic
> and
> other crap like that, then have a 1Gb network for your actual
> calculations, MPI traffic, etc.
> -----BEGIN PGP SIGNATURE-----
> Version: GnuPG v1.4.2 (GNU/Linux)
>
> iD8DBQFDghRCXVaO67S1rtsRAlpkAJ4vs6ROMe1TF4IcwNoyANTMtDjmfQCgwcGI
> xzSx4qUzq2cLje32i0DD0hw=
> =Ps1N
> -----END PGP SIGNATURE-----
> --
> gentoo-cluster@g.o mailing list
>

-- 
gentoo-cluster@g.o mailing list