[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Condor-users] AFS & Condor



On Fri, Feb 18, 2005 at 03:43:29PM -0600, Franco Milicchio wrote:
> Hi.
> 
> I'm planning to install condor on an AFS cell. What I'm planning to do, 
> is this:
> 
> - /afs/cell.name/proj/condor -> /afs/cell.name/proj/condor.@sys
> - install condor (linux 2.4) in /afs/cell.name/proj/condor.i386_linux24
> - install condor (linux 2.6) in /afs/cell.name/proj/condor.i386_linux26

Nice, but what's the point of having it kernel-dependent?
AFAIK, Condor is dependent on glibc versions (since standard universe
libs would replace it) and thus on Linux distros/versions

> - install condor (macosx) in /afs/cell.name/proj/condor.ppc_darwin_70
> - have all symlinks for bin, sbin & co on @sys
> 
> Every user using an afs account will have rights to read and write on
> /afs/cell.name/proj/condor-jobs, acl based on ip addresses will allow 
> writing on this volume.

YOu really want to have transient data on AFS?
My guts tell me that there'd be better ideas...

> Has anyone a configuration like this one?

Obviously there have been requests for AFS support, but if you read the
corresponding parts of section 3.3 in the manual (USE_AFS=...) you might
get the impression that it's no good for everything

> Is it necessary a condor user? I don't think so...

>From reading the docs, it's better to run the daemons *not* as root...

> I'm quite confused using AFS... but it's the smartest way of sharing 
> data across multiple machines...

AFS is powerful indeed, but
- standard universe doesn't rely on shared resources at all
- there's NFS

what's the point of using AFS for everything Condor? (It's probably OK
for the RELEASE_DIR, but everything else I would not put onto AFS. Try
to convert me.)

> No keyboard found. Press F1 to continue...
> (Almost every BIOS available in this world... even yours!)

Sure... had a good laugh the first time...

Cheers,
 Steffen

-- 
Steffen Grunewald * * * Merlin cluster admin (http://pandora.aei.mpg.de)
Albert-Einstein-Institut (MPI Gravitationsphysik, http://www.aei.mpg.de)
       Science Park Golm, Am Mühlenberg 1, 14476 Potsdam, Germany
e-mail: steffen.grunewald(*)aei.mpg.de * +49-331-567-{fon:7233,fax:7298}