Two offers in response to one discovery

John Wobus jw354 at cornell.edu
Fri Dec 16 16:52:46 UTC 2011


On Dec 15, 2011, at 5:16 PM, John Miller wrote:
> OK, my bad. They actually do check for config changes and only  
> restart then.
> When there is a lot of activity it just seems like once per minute.
>
> So you all know, Bradford killed the Nov. 8th process and everything  
> seems
> to be working just fine now. The only unanswered questions I have  
> now are
> how the 2 processes were able to bind to the same port and what  
> caused the
> issue in the first place.

Interesting.  We do a similar thing to what you describe (periodically  
check for
config changes and kill/restart) so I like to hear about such  
experiences.  I think
the ISC server refuses to run if another instance is already  
listening, but I find
it easy to  imagine now that that is very specific code to catch such  
a problem
and there may well be ways to trick it.

Bradford's code to stop and restart the server could be written to do  
its own
check.  Obviously, the more checking, or rather the more effective their
checking process, the less likely this is to happen.

John Wobus
Cornell University



More information about the dhcp-users mailing list