Comment #7 on issue 1872 by jean.deruelle: Fix the Timer recovery and
subsequent requests on diff nodes
http://code.google.com/p/mobicents/issues/detail?id=1872
This issue was updated by revision r15685.
gravitate the timers where the LB decided to fail over the subsequent
request for a sipapplication session which would work better with any kind
of Load Balancer but it is done only if the concurrency control mode is
SipApplicationSession.basically the timer is cancelled remotely and
rescheduled locally. the most recent node that recreate a sip application
session will move the timers locally in recreating the sip application
session
if the sip application session is present on multiple nodes because there
is no affinity then the timers will sit where the sipappsession was
recreated last (note recreated, not updated).
it would stay at the node where it was recreated last it the sip
application session flip flop due to LB flip flopping to different nodes if
there is no affinity then there is no guarantees anyway wrt to concurrency
control
so no need to flip flop the timers
