Print

Print


On ה', 2005-11-17 at 15:58 +0000, Stephen Childs wrote:
> We have a couple of sites where we want to be able to submit Grid jobs 
> into an existing Condor pool. I have been trawling the usual sources of 
> gold such as the LCG-ROLLOUT archives and the GOC wiki to try and find out 
> what is needed to set this up and so far have found 
> http://goc.grid.sinica.edu.tw/gocwiki/Condor_setup which briefly describes 
> the process of setting up a Condor-only site. However at the sites in 
> question we have existing PBS configurations which we want to retain, so 
> we want to run both in parallel. It looks like under LCFG at least only 
> one type of jobmanager is supported at a time (i.e. LSF or Condor or PBS). 
> Is this still the case with Yaim and/or Quattor? Has anyone tried this 
> kind of thing and lived to tell the tale?

Hello, Stephen!

I've been trying to set this thing up for a long time, still
unsuccessfully, yet cannot stop. As far as I can see the reasons for our
failures were so far 99% of firewall blocks issues: our uni's firewall
policy blocks outgoing traffic.  We're still "fighting" the firewall
guys. 

To the condor point: I assume you have (plan to have) a 2.6 site.

do refer to
http://www.cs.technion.ac.il/Labs/DSL/projects/EGEE/files/lcg_condor.pdf

I created 'good' files, and then overwrote them with a script.

Please see if you see any inconsistencies in the document - it is the
first draft. Most of the valuable data in it is the knowledge &
expertise of BUDAPEST Hungarian site manageer, Mr. Hernath Szabolcs.
I will somehow transfer this information into LCG web space as soon as
my site is up and running ( i.e. the filrewall issues are closed. )

We can with no problem submit into a condor pool with LCG2 middleware
and credentials, but unfortunately, if the jobs were submitted
externally via RB, our system could not notify the submitter about
anything back since the outbound connectivity to the RB on e.g. port
7772 was blocked by the university firewall policy, because this is a
common IRC port. 


Regards,
Max.

> 
> 
> Stephen
-- 
Maxim Kovgan
Distributed Systems and Data Mining Laboratory
Computer Science, Technion, ISRAEL
phone: +972-48293864