Commits on Nov 22, 2021

1. Various changes of RTP timeout handling. …

   - introduced timeout-mode parameter to control mode of operation off/any/all during session
     off - don't monitor timeouts / turn monitoring off
     any - react if a timeout occur on any RTP stream
     all - react if timeout occur on all streams (old behavior)
   - Don't monitor RTCP streams for timeout (to prevent false alarms if RTP timeout is very low)
   
   This patch introduces primary the possibility to react on RTP timeouts of a single stream.
   
   During call establishment / early-media, there is often only one stream direction active. To prevent false alarms
   for this case, the timeout-mode parameter for the call-interface was added. The logic is, that a call starts with
   a disabled timeout monitoring and when the session is established in both directions (for example 200OK Reply for
   an answer) the timeout will be activated.
   
   When activated, timeout_activated element of struct call will be set to earliest possible timeout occurence
   (activation time + timeout time). This should prevent false alarms immediatliy after session establishment
   
   There is still an unresolved problem at the moment. Even timeout_activated is used, it sometimes happens
   that on first check a timeout is detected. This occurs even, when timeout value (and timeout_activated
   in succession) is increased.
   To circumvent this, missed_packet_counter of struct packet_stream is used, so that the checks fails on the third
   missed package in a row.
   
   Signed-off-by: Arnd Schmitter <[email protected]>

   

   arnd-s committed Nov 22, 2021
   Configuration menu

   • View commit details
   • Copy full SHA for bffbea4
   • Browse repository at this point

   Copy the full SHA

   bffbea4 View commit details

   Browse the repository at this point in the history