On 2010-07-07 21:57, Wen Jun wrote: > Well, that protection mechanism to avoid surf registration load to SIPX is > understood. That seems a pretty wise fence. > > My another question is how&when SIPX knows to give phones by random > registration timers ? It just always randomizes the returned expiration value.
While the power-on case Dale described is the most easily understood way that registrations can become synchronized, it's interesting to note that we found that in large systems it tends to happen even without anything that dramatic. If you normally have several phones registering every second, and there is a momentary network problem that delays some refreshes, then when the network comes back there is a 'spike' - if the spike is large enough, then the processing of those requests takes long enough that more are delayed, and so the spike actually grows with each refresh cycle - eventually becoming large enough to cause serious problems. Randomizing refresh times completely prevented this. _______________________________________________ sipx-users mailing list [email protected] List Archive: http://list.sipfoundry.org/archive/sipx-users Unsubscribe: http://list.sipfoundry.org/mailman/listinfo/sipx-users sipXecs IP PBX -- http://www.sipfoundry.org/
