[asterisk-dev] [Code Review] 4536: iax2_poke_noanswer expiration timer too short

Y Ateya reviewboard at asterisk.org
Fri Apr 3 14:58:54 CDT 2015


-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviewboard.asterisk.org/r/4536/
-----------------------------------------------------------

(Updated April 3, 2015, 7:58 p.m.)


Review request for Asterisk Developers and rnewton.


Changes
-------

- Restored the deleted comment line.
- Updated poke timeout calculations:
  * In case of peer unreachable (peer->lastms < 0): Use time less than pokefreqnotok to prevent duplicate POKE. As poke will be retried after pokefreqnotok; so make POKE timeout less than POKE next cycle.
  * In case of peer reachable: Allow time for POKE retries and take qualify time and MAX_RETRY_TIME into consideration. Also take into consideration that POKE timeout should be less than pokefreqok. The following table shows time of POKE retrials and timeout values based on this patch (Qualify: peer qualify time, 1st: time of first retry, 2nd: Time of second rety starting from 1st rety, 3rd: Time of third retry starting from 2nd retry).

Qualify	1st 	2nd 	3rd 	Timeout
500	125	1250	10000	20500
2000	500	5000	10000	22000
10000	2500	10000	10000	30000


Bugs: ASTERISK-24894
    https://issues.asterisk.org/jira/browse/ASTERISK-24894


Repository: Asterisk


Description
-------

Increase POKE retry window from DEFAULT_MAXMS * 2 (4 seconds) to bigger number (derived from qualify time; which control POKE retry time).


Diffs (updated)
-----

  trunk/channels/chan_iax2.c 432806 

Diff: https://reviewboard.asterisk.org/r/4536/diff/


Testing
-------

- Tried test with multiple qualify values (2 and 10 seconds).
- Tried test with 100% packets loss to ensure that when a POKE packet is dropped it will be retried couple of time before declaring client disconnected.


Thanks,

Y Ateya

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.digium.com/pipermail/asterisk-dev/attachments/20150403/d9769eee/attachment-0001.html>


More information about the asterisk-dev mailing list