Originally created by: sf-overlords
Created by: jbeard
Created date: 2010-05-06 19:49:43.793000
Assigned to:nobody
David Burley: how hard would it be to allow the reactor_setup to be idempotent and safe for running in parallel ?
Richard Copeland: running in parallel -- pretty hard
David Burley: if it needs done every time then I need to fix it in this step here
Richard Copeland: it tears down all the amqp queues and rebuilds them according to the code configuration
David Burley: won't that risk amqp event loss
Richard Copeland: yes
David Burley: so that doesn't seem like a good thing to do
Richard Copeland: that's why it should really only be done after the queues have drained
David Burley: why not just modify the queue live -- on code startup see if its got the current queue setup and modify it live; no data loss and the user runnign it doesn't need to be quite as smrt
Richard Copeland: It's possible it could be rewritten, but that's not the way it's written now; I think we can make it a post-push dependency
David Burley: that's probably OK short term but long term that's not a good plan