Hi,
Each of our partner institutes maintains its own computer pool and there are several different queuing systems in use. Our users are not familiar with the other queuing systems and vice versa. So we've agreed to use Condor as our common interface for job submission. Condor should be able to communicate with the various "underlaying" clusters in order to prevent our users from having to memorize several different batch file formats, etc. Of course, our users should still be able to use their familiar local queuing system directly. The software which is used in our local queuing system usually generates large amounts of data. Our local users can access their output files though our local network. But how would I handle output file transfer to non-local users? Should I set up home directories for them? Or should I periodically sync the output files (e.g. over night)? I'm afraid that allowing our software to write directly to the local directories would generate too much traffic. Lukas From: HTCondor-users [htcondor-users-bounces@xxxxxxxxxxx] on behalf of Miha Ahronovitz [myinnervoice@xxxxxxxxx]
Sent: 09 January 2014 17:27 To: HTCondor-Users Mail List Subject: Re: [HTCondor-users] Condor/SGE cluster Lukas, what is reason you want to use HTCondor as "overlay cluster" to the SGE cluster? Why not SGE as well?
Miha
|