Mailing List Archives
Authenticated access
|
|
|
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [Condor-users] Cluster design questions...
- Date: Mon, 6 Mar 2006 10:54:15 -0500
- From: "Jonathan D. Proulx" <jon@xxxxxxxxxxxxx>
- Subject: Re: [Condor-users] Cluster design questions...
On Mon, Feb 13, 2006 at 02:38:21PM -0600, Zachary Miller wrote:
:On Mon, Feb 13, 2006 at 03:17:46PM -0500, Jonathan D. Proulx wrote:
:>
:> Is it possible for a machines to have multiple filesystem domains?
:
:technically, no. but there's a quasi-hacky workaround for most situations.
:
:the job's filesystem_domain doesn't actually have to match the machine's. in
:reality, it must just be a substring of the machine's filesystem_domain.
:
:for example:
: job filesystem_domain is "edu".
: machine filesystem_domain is "wisc.edu"
:
:that machine can run that job.
Thanks for the info, since you're relying on this at UWisc can I
assume it's likely to stay that way for the foreseeable future?
Let me see if I understand (since I'm about to actually try and put
this all together).
I have a collection of machines called borg<nn>. They have AFS a
standard NFS automounting and a shared GFS storage system.
I define these as FS_DOMAIN=borg.nfs.csail.mit.edu. I expect jobs
started here will stay on "borg" systems unless the FS_DOMAIN is
specified mor egenericly for the job. I also expects jobs for the
FS_DOMAINs "csail.mit.edu" and "nfs.csail.mit.edu" will be matched to
these resources and that "borg.csail.mit.edu" would not match.
-Jon