Re: [Condor-users] Running Jobs on central-manager

[Alain Roy <roy@xxxxxxxxxxx>]

On Feb 11, 2009, at 7:35 AM, Samir Cury wrote:
> According to what i read in some presentations, and the most  
> relevant, this topic on the list "What makes a node execute node or  
> not a execute	node", i just have to put on the DAEMON_LIST the  
> STARTD, and restart condor_master, so if the central-manager (the  
> machine itself) is correctly pointed on condor_config, it will  
> communicate with the collector, and become an execute node.
>
> So, i did it, but i couldn't see the central manager on  
> condor_status list, am i missing something?

Before I answer your question...

In general, you have to be really careful about running jobs on your  
central manager. A lot of people set up Condor with host permissions,  
which means that to have administrator privileges you only need to be  
on a certain computer. (There are stronger authentication methods  
available, but people often use host permissions anyway.) A common  
configuration is to say "If you are on the central manager, you have  
administrator privileges for the whole pool.) If you do this, jobs  
that run on the central manager have elevated privileges with respect  
to Condor.

If you're running Condor just to try it out, or if you only have  
trusted users, maybe you don't care. But I think it's pointing out so  
that people don't make a mistake that can affect their security.

On to your question...

1) Look in the MasterLog on the central manager: did the master  
restart? Did it start the condor_startd?

2) Do you see a condor_startd process?

3) Do you see any errors in the StartLog (the log file for the  
condor_startd)?

4) Do you see any errors in the CollectorLog? This is where the  
condor_startd sends information to let the pool know that it exists.

-alain