Re: a question about the RTP packet-lost from gnuGK when H.460.18/19 enabled

Bo Xu <[email protected]>
Newsgroups gmane.comp.telephony.openh323gk.user
Message-ID <CAFgLERkesHxDgoRDhk_meN24r6XeWu-NNNytz4x+Ync=1poFYA@mail.gmail.com>
Hi,

I am continuing to investigate our gnuGK3.5 for the rtp-packet-lost issue,
the following is the gatekeeper.ini we are using now to enable H.460.18/19,
could you help me to check if it is right? do I need to put more parameters
to get better performance on my windows-& PC? Thanks!


[Gatekeeper::Main]
Name=4601819GK
TimeToLive=60
Home=10.1.11.210
Bind=10.1.11.210

[GkStatus::Auth]
rule=password
admin=sA3MW7H3VbE=

[RoutedMode]
GKRouted=1
H245Routed=1
CallSignalPort=1721

AcceptUnregisteredCalls=1
supportNATedEndpoints=1
AlwaysRewriteSourceCallSignalAddress=0

EnableH46018=1
H46018NoNat=1

CallSignalHandlerNumber=5
RtpHandlerNumber=20

[Proxy]
;Enable=0
;ProxyAlways=1
ProxyForNAT=0
ProxyForSameNAT=0



Regards,
Bo Xu





On Thu, Apr 3, 2014 at 8:28 AM, Bo Xu <[email protected]> wrote:

> > Hi,
>
> > I think GnuGk would run slightly better on Linux, but I don't think it
> > should make much of a difference in your case. Your GnuGk should have no
> > problem handling 5 720P streams.
>
> > 5 years ago I estimated that a GnuGk on a dual core could roughly handle
> > 50 proxied calls with 384 kbps video or about 15 calls with 2 Mbit
> > video.
>
> > Todays servers are faster, GnuGk changed in many ways, maybe somebody on
> > the list has some current numbers to share ?
>
> > Regards,
> > Jan
>
> > --
> > Jan Willamowius, Founder of the GNU Gatekeeper Project
> > EMail  : jan@...
> > Website: http://www.gnugk.org
> > Support: http://www.willamowius.com/gnugk-support.html
>
>
>
> Hi Jan,
>
> Thanks!
>
> Regards,
> Bo Xu
>
>
>
>
>
>
>
> On Wed, Apr 2, 2014 at 11:13 AM, Bo Xu <[email protected]> wrote:
>
>> Hi,
>>
>> several weeks ago I reported a RTP-packet-lost issue and got fixed by
>> Simon's help and that is because we didn't handle the KeepAlive-packet's
>> SequenceNumber well, after that our soft-terminal works well with
>> H.460.18/19 enabled by gnuGK. now in the multi-participants testing we got
>> another issue: if there are more than 5 participants are sending 720P-H.264
>> video stream, then we find that gnuGK lost packets again, normally there is
>> 1 packet-lost in about 1 minute for 5 participants(720P-H264); this cause
>> our MCU to keep on sending FastUpdate request and cause video-frozen or
>> video quality down. we are running gnuGK on a windows-box, will it be
>> better if we run gnuGK on a linux-box? and do you have any performance
>> testing document for how many 720P-H.264 video-stream could be support by
>> one gnuGK? Thanks!
>>
>> the gnuGK we are running now is:
>> 3.5 GNUGK
>> the PC on which our gnuGK is running is:
>> Windows 7 Professional 64 bit
>> I7 @ 3.4GHz
>> 4GB Ram
>>
>> Regards,
>> Bo Xu
>>
>>
>>
>> On Mon, Mar 3, 2014 at 3:18 PM, Bo Xu <[email protected]> wrote:
>>
>>> Hello Simon,
>>>
>>> Thanks a lot!    yes as you mentioned our issue does relate with
>>> KeepAlive, the details are:  our terminal use same series of RTP
>>> SequenceNumber for both video packet and KeepAlive packet, so when gnuGK
>>> relay this RTP channel the KeepAlive packet will not be send to our MCU and
>>> thus the SequenceNumber received by MCU becomes not continued, in my
>>> previous email I thought this is because gnuGK lost packet, actully gnuGK
>>> doesn't lost packet.   we are updating our terminal now to separate the
>>> SequenceNumber of video packet and KeepAlive packet so the SequenceNumber
>>> of video packet will always be continued, Thanks for help us find the
>>> reason!
>>>
>>> Regards,
>>> Bo Xu
>>>
>>>
>>>
>>>
>>>
>>>
>>>
>>>
>>> Message: 6
>>> Date: Tue, 4 Mar 2014 00:43:07 +1000
>>> From: "Simon Horne" <[email protected]>
>>> Subject: Re: [Openh323gk-users] a question about the RTP packet-lost
>>>         from    gnuGK   when H.460.18/29 enabled
>>> To: "'GNU Gatekeeper Users'" <[email protected]>
>>> Message-ID: <[email protected]>
>>> Content-Type: text/plain; charset="us-ascii"
>>>
>>> Bo
>>>
>>> I think this might be a red herring. I think the keepAlive packet which
>>> is
>>> send 1 packet every 20 sec (which should be discarded) is being counted.
>>>
>>> Simon
>>>
>>>
>>> From: Bo Xu [mailto:[email protected]]
>>> Sent: 04 March 2014 00:30
>>> To: openh323gk-users
>>> Subject: [Openh323gk-users] a question about the RTP packet-lost from
>>> gnuGK
>>> when H.460.18/29 enabled
>>> Hi,
>>>
>>> we are using gnuGK with H.460.18/19 enabled to connect our MCU and our
>>> software H.323 terminal, our gnuGK/MCU/terminal are in the same subnet in
>>> our testing.  we found gnuGK can relay the video RTP packets from MCU
>>> well,
>>> but about every 20 second gnuGK will lost one packet from our terminal.
>>>  I
>>> have tested both 720P video and CIF video and they are same.
>>>
>>> we already increased the RtpHandlerNumber parameter(we ever tried to set
>>> it
>>> up to 20), but it seems the packet-lost rate is always same(about every
>>> 20
>>> seconds lost one packet).
>>>
>>> I am working on this and I still didn't figure out why the packets from
>>> MCU
>>> are never lost, only the packets from terminal are lost?  I ever guess it
>>> might be because that every 20 seconds the buffer is full, but I cannot
>>> explain why both 720P and CIF video has the same packet-loast rate?  ?
>>> which
>>> part of the gnuGK code I need to read? could you give me some clues for
>>> this? Thanks!
>>>
>>> Regards,
>>> Bo Xu
>>>
>>
>>
>

------------------------------------------------------------------------------

_______________________________________________________

Posting: mailto:[email protected]
Archive: http://sourceforge.net/mailarchive/forum.php?forum_name=openh323gk-users
Unsubscribe: http://lists.sourceforge.net/lists/listinfo/openh323gk-users
Homepage: http://www.gnugk.org/
lmpx.com only provides a reader for public news (NNTP) servers. It is not affiliated with the servers or forums shown here and is not responsible for the content of articles, which is written by their respective authors.