[Freeswitch-users] Eavesdrop results in rewriting RTP sequence numbers on a bridge.

[Richard Brady]

"Dobtou" think it's possible that FS inserts silence when a bug is attached
but not when not?


On 12 September 2013 16:26, Anthony Minessale
<anthony.minessale at gmail.com>wrote:

> My phone seems to think dobtou is a valid word, sorry =D
>
>
>
> On Thu, Sep 12, 2013 at 10:22 AM, Richard Brady <rnbrady at gmail.com> wrote:
>
>> Yes, my bad. Misinterpretation of Wireshark graph:
>>
>>
>> https://dl.dropboxusercontent.com/u/51603927/graph_of_media_bug_impact_on_RTP_seqnum.png
>>
>> Something definitely changes when the media bug is added. I will work out
>> what but I think FS is inserting silence frames when the media bug is on,
>> and not inserting them when it is off.
>>
>> Thanks for setting me straight.
>>
>> On 12 September 2013 13:58, Anthony Minessale <
>> anthony.minessale at gmail.com> wrote:
>>
>>> Seq numbers never are preserved. Again dobtou mean timestamps?
>>> On Sep 12, 2013 7:31 AM, "Richard Brady" <rnbrady at gmail.com> wrote:
>>>
>>>> Thanks.
>>>>
>>>> I'm not talking about the new stream for the eavesdrop. I'm talking
>>>> about the streams on the A and B legs of the bridge.
>>>>
>>>> I don't see why FS would rewrite sequence numbers on an RTP stream
>>>> which it is simply bridging (no transcoding, no muxing, etc). Sequence
>>>> numbers allow packet loss to be detected and compensated for. If FS is not
>>>> running a JB or PLC then it should leave discontinuous sequence numbers as
>>>> is so that a downstream JB / PLC can use them.
>>>>
>>>> I'll send some Wireshark screenshots shortly. And will double check re
>>>> seq num vs timestamps.
>>>>
>>>> On 11 September 2013 20:19, Anthony Minessale <
>>>> anthony.minessale at gmail.com> wrote:
>>>>
>>>>> Seq numbers are always rewritten.  Maybe you mean timestamps?
>>>>> The muxing and recreation of a new media stream requires a newly
>>>>> generated stream.  There is not a way around that.
>>>>>
>>>>>
>>>>>
>>>>> On Wed, Sep 11, 2013 at 9:29 AM, Richard Brady <rnbrady at gmail.com>wrote:
>>>>>
>>>>>> Hi guys
>>>>>>
>>>>>> Seems than running eavesdrop (and potentially other media bugs) on a
>>>>>> bridged channel results in the rewriting of sequence numbers on RTP
>>>>>> streams, potentially hiding packet loss when it occurs and creating havoc
>>>>>> with downstream jitterbuffers.
>>>>>>
>>>>>> Is there a way to disable this behaviour?
>>>>>>
>>>>>> Richard
>>>>>>
>>>>>> --
>>>>>> Richard Brady
>>>>>> M: +44 (0)7771 623 348
>>>>>> T:  +44 (0)20 8144 8160
>>>>>> E: rnbrady at gmail.com
>>>>>>
>>>>>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.freeswitch.org/pipermail/freeswitch-users/attachments/20130912/f37bd974/attachment.html