On Fri, 20 Nov 2009, gonzoprosperity wrote:

> My next thought was to specify a reserve timeout
> 
>   job = queue.reserve(20) # seconds
> 
> So that at least every 20 seconds there will be a loop iteration any
> SIGTERM sent in the interim can be caught and handled.

This is the method I've adopted for our Python-based workers, I've basically
got a SIGINT/SIGTERM handler that basically sets "cont = False" and within the
next TIMEOUT period the worker will gracefully die.


Cheers,
-R. Tyler Ballance
--------------------------------------
 GitHub: http://github.com/rtyler
Twitter: http://twitter.com/agentdero
   Blog: http://unethicalblogger.com

Attachment: pgpPK5myI9woH.pgp
Description: PGP signature

Reply via email to