[rabbitmq-discuss] regarding supervisor2

Matthew Sackman matthew at rabbitmq.com
Tue Jan 25 11:00:41 GMT 2011


Howdy,

On Mon, Jan 24, 2011 at 04:56:55PM -0800, Arun Suresh wrote:
> Ive been playing around with supervisor2 which you guys ship with rabbitmq
> server. The delayed restart is seriously a life saver.

Glad you like it.

> The only minor issue ive had with it is:
> Once the process has exceeded MaxT and MaxR, you guys spawn a timer thread
> to restart the process after the specified delay.. During this period, the
> process is actually dead. but supervisor:which_children still returns a
> stale Pid. Any reason why inside the do_restart method, once the restart1
> method returns a {terminate, NState} , state_del_child() is NOT called ?

>From memory, no reason at all, and I think it would be an improvement
for it to do the del_child sooner rather than later and thus not return
the stale pid. Slight complexities with simple_* supervisors possibly -
can't remember - it's been a while since I looked at that code. I'll
take a look and most likely file a bug.

Best wishes,

Matthew


More information about the rabbitmq-discuss mailing list