Open Stack

On 1/27/15 9:13 AM, Fischer, Matt wrote:
> Our keystone database is clustered across regions, so we have this job
> running on node1 in each site on alternating hours. I don’t think you’d
> want a bunch of cron jobs firing off all at once to cleanup tokens on
> multiple clustered nodes. That’s one reason I know not to put this in
> the code.
>
> Are there other reasons that an operator might like to keep old tokens?
> Auditing?

Well, without knowing keystone code (yay great start to the email) I 
would imagine this would be like other periodic tasks, where the task 
gets generated and one of the workers picks it up, wherever that worker 
may be. But maybe that's wishful thinking.

-- 
-jlk