Re: [Condor-users] SYSTEM_PERIODIC_HOLD

[Dan Bradley <dan@xxxxxxxxxxxx>]

Todd Tannenbaum wrote:
> Steven Timm wrote:
>   
> > system_periodic_hold is executed on the machine that reruns
> > the condor_schedd.  I think you have to restart the condor_schedd.
> >
> > Steve
> >     
>
> Restarting the schedd will certainly work, but a reconfig is all that is 
> required.  So you could reconfigure "on the fly" by making the change in 
> the config file and then doing
>    condor_reconfig <name of submit machine>
> from a machine with ADMINISTRATOR authorization according to the config 
> file (by default this means the central manager).
>   

That's true for idle jobs but not for running jobs.  (Perhaps this is an 
oversight.)

The workaround is to send SIGHUP to all running shadows in addition to 
running condor_reconfig.

--Dan