RE: OT: Question about FW packetand NAT traversal

N95 is not based on the helix and it uses the microcore.
As far as the poke packet is concerned, if I remember
correctly, N95
based player does send the poke packet.

Ashish

-----Original Message-----
From: porting-symbian-bounceshelixcommunity.org
[mailto:porting-symbian-bounceshelixcommunity.org] On
Behalf Of ext
Greg Wright
Sent: Thursday, January 03, 2008 11:58 AM
To: Yusuf Mayet
Cc: porting-symbianhelixcommunity.org; protocol-devhelixcommunity.org
Subject: [Porting-symbian] Re: [Protocol-dev] OT: Question
about FW
packetand NAT traversal

Adding the Symbian list for people familiar with the N95.

If the phone allows it, you can always download a Symbian
RealPlayer
from the Helix site and install it. That way you would know
for sure.

--greg.


Yusuf Mayet wrote:
> Hi,
> 
> thanks for the responses.  Yes, the phone I used is
quite old, I will 
> try a newer one.  Will the Nokia N95 be Helix-based,
and will it 
> therefore have NAT poking?
> 
> On 1/3/08, Greg Wright <gwrightreal.com> wrote:
>> Henry Ping wrote:
>>> The poke-port code was added in mid 2004 and
should be in the most 
>>> recent RealPlayer release.
>>>
>>> We use "Symmetric RTP" and fallback
to "TCP Tunneling" as described 
>>> in the reference link you provided. It's been
working well for us.
>>>
>>> In addition, we also provided RTSP
proxy(Application Level Gateway),
>> though
>>> I think it's being phased out by Real.
>> If the phone is old, it will have a micro-core
based player on it and

>> not Helix based. In that case it is very old and
may not have NAT 
>> poking.
>>
>> --greg.
>>> Henry
>>>
>>>> -----Original Message-----
>>>> From: Eric Hyche [mailto:ehychereal.com]
>>>> Sent: Wednesday, January 02, 2008 9:46 AM
>>>> To: 'Yusuf Mayet'; protocol-devhelixcommunity.org
>>>> Cc: Henry Ping
>>>> Subject: RE: [Protocol-dev] OT: Question
about FW packet and NAT
>> traversal
>>>>
>>>> Poke-port code was added fairly recently
(within the last year 
>>>> perhaps?) to the Helix code base, so it is
no surprise to me that 
>>>> it might not be in the RealPlayer version
that you tried, given 
>>>> that RealPlayer tends to lag behind the
latest Helix release.
>>>>
>>>> As far as what type of NAT traversal
techniques are used, Henry 
>>>> added this code, I believe, and he would
know better how to answer.

>>>> I'm copying him on this reply.
>>>>
>>>> Eric
>>>>
>>>>
=============================================
>>>> Eric Hyche (ehychereal.com)
>>>> Technical Lead
>>>> RealNetworks, Inc.
>>>>
>>>>> -----Original Message-----
>>>>> From: protocol-dev-bounceshelixcommunity.org
>>>>> [mailto:protocol-dev-bounceshelixcommunity.org] On Behalf Of 
>>>>> Yusuf Mayet
>>>>> Sent: Friday, December 28, 2007 4:33
AM
>>>>> To: protocol-devhelixcommunity.org
>>>>> Subject: [Protocol-dev] OT: Question
about FW packet and NAT 
>>>>> traversal
>>>>>
>>>>> Hi,
>>>>>
>>>>> This might be OT.
>>>>>
>>>>> When looking at RTSP clients,
especially on cellphone handsets, I 
>>>>> noticed that Pvplayer, when behind a
NAT/FW, sends a "FW packet" 
>>>>> to the server, to poke a hole through
the NAT, so the UDP port 
>>>>> mappings agreed in the SDP are opened. 
However, RealPlayer (not 
>>>>> sure of version number) does not send
this, and can lead to dead
sessions.
>>>>> So my question is rather: when type of
NAT traversal techniques 
>>>>> are currently implemented, or planned
to be implemented in 
>>>>> clients, such as STUN or RTSP-over-TCP,
etc?
>>>>>
>>>>> I'm looking for discussion and ideas.
>>>>>
>>>>> Reference:
>>>>> http://ietf.org/internet-drafts/draft-ietf-mmusic-
rtsp-nat-eva
>>>>> luation-00.txt
>>>>>
>>>>>
>>>>> --
>>>>>
>>>>> thanks,
>>>>> Yusuf
>>>>>
>>>>>
_______________________________________________
>>>>> Protocol-dev mailing list
>>>>> Protocol-devhelixcommunity.org
>>>>> http://lists.helixcommunity.org/mailman/listinfo/pr
otocol-dev
>>>>>
>>>
>>>
_______________________________________________
>>> Protocol-dev mailing list
>>> Protocol-devhelixcommunity.org
>>> http://lists.helixcommunity.org/mailman/listinfo/pr
otocol-dev
>>>
>>
> 
> 


_______________________________________________
Porting-symbian mailing list
Porting-symbianhelixcommunity.org
http://lists.helixcommunity.org/mailman/listinfo
/porting-symbian

_______________________________________________
Protocol-dev mailing list
Protocol-devhelixcommunity.org
http://lists.helixcommunity.org/mailman/listinfo/pr
otocol-dev

Thanks again.

I started comparing the FW packets of pvlayer and RealPlayer
on N95
(User-Agent: RealMedia Player/mc.30.19.01 (s60;
epoc_av30_armv5)).

Pvplayer:
Ethereal reports it as UDP 8 bytes long. Decoding it as RTP
just has
"Unknown RTP version 0"

Real:
Ethereal reports it as UDP 28 bytes long. Decoding it as RTP
just has
some RTP headers, and a payload.

Can anyone shed more light on the matter.  What else can/do
I look out
for.  The problem I am having, is that the RTSP server seems
to accept
pvplayer's FW poke packet, and therefore the stream is
successful,
while it rejects the N95's one, causing the stream to fail.

On 1/3/08, Chris Newman <chrisfloor51.com> wrote:
> Ashish.As.Guptanokia.com wrote:
> > N95 is not based on the helix and it uses the
microcore.
>
> Will the main Atlas Helix be in the default player on
the next S60
> release v3.2?
>
> > As far as the poke packet is concerned, if I
remember correctly, N95
> > based player does send the poke packet.
>
> Fairly sure it does. Think the N70 (fairly old, S60v2
Symbian OS 8.0a)
> does as well (unless this was the one used for
testing). Maybe the
> 6600(OS7.0s)/6630(OS8.0a)/6680(OS8.0a) don't.
>
> chris
>
> >
> > Ashish
> >
> > -----Original Message-----
> > From: porting-symbian-bounceshelixcommunity.org
> > [mailto:porting-symbian-bounceshelixcommunity.org] On Behalf Of ext
> > Greg Wright
> > Sent: Thursday, January 03, 2008 11:58 AM
> > To: Yusuf Mayet
> > Cc: porting-symbianhelixcommunity.org;
protocol-devhelixcommunity.org
> > Subject: [Porting-symbian] Re: [Protocol-dev] OT:
Question about FW
> > packetand NAT traversal
> >
> > Adding the Symbian list for people familiar with
the N95.
> >
> > If the phone allows it, you can always download a
Symbian RealPlayer
> > from the Helix site and install it. That way you
would know for sure.
> >
> > --greg.
> >
> >
> > Yusuf Mayet wrote:
> >> Hi,
> >>
> >> thanks for the responses.  Yes, the phone I
used is quite old, I will
> >> try a newer one.  Will the Nokia N95 be
Helix-based, and will it
> >> therefore have NAT poking?
> >>
> >> On 1/3/08, Greg Wright <gwrightreal.com> wrote:
> >>> Henry Ping wrote:
> >>>> The poke-port code was added in mid
2004 and should be in the most
> >>>> recent RealPlayer release.
> >>>>
> >>>> We use "Symmetric RTP" and
fallback to "TCP Tunneling" as described
> >>>> in the reference link you provided.
It's been working well for us.
> >>>>
> >>>> In addition, we also provided RTSP
proxy(Application Level Gateway),
> >>> though
> >>>> I think it's being phased out by
Real.
> >>> If the phone is old, it will have a
micro-core based player on it and
> >
> >>> not Helix based. In that case it is very
old and may not have NAT
> >>> poking.
> >>>
> >>> --greg.
> >>>> Henry
> >>>>
> >>>>> -----Original Message-----
> >>>>> From: Eric Hyche
[mailto:ehychereal.com]
> >>>>> Sent: Wednesday, January 02, 2008
9:46 AM
> >>>>> To: 'Yusuf Mayet';
protocol-devhelixcommunity.org
> >>>>> Cc: Henry Ping
> >>>>> Subject: RE: [Protocol-dev] OT:
Question about FW packet and NAT
> >>> traversal
> >>>>> Poke-port code was added fairly
recently (within the last year
> >>>>> perhaps?) to the Helix code base,
so it is no surprise to me that
> >>>>> it might not be in the RealPlayer
version that you tried, given
> >>>>> that RealPlayer tends to lag
behind the latest Helix release.
> >>>>>
> >>>>> As far as what type of NAT
traversal techniques are used, Henry
> >>>>> added this code, I believe, and he
would know better how to answer.
> >
> >>>>> I'm copying him on this reply.
> >>>>>
> >>>>> Eric
> >>>>>
> >>>>>
=============================================
> >>>>> Eric Hyche (ehychereal.com)
> >>>>> Technical Lead
> >>>>> RealNetworks, Inc.
> >>>>>
> >>>>>> -----Original Message-----
> >>>>>> From: protocol-dev-bounceshelixcommunity.org
> >>>>>>
[mailto:protocol-dev-bounceshelixcommunity.org] On
Behalf Of
> >>>>>> Yusuf Mayet
> >>>>>> Sent: Friday, December 28,
2007 4:33 AM
> >>>>>> To: protocol-devhelixcommunity.org
> >>>>>> Subject: [Protocol-dev] OT:
Question about FW packet and NAT
> >>>>>> traversal
> >>>>>>
> >>>>>> Hi,
> >>>>>>
> >>>>>> This might be OT.
> >>>>>>
> >>>>>> When looking at RTSP clients,
especially on cellphone handsets, I
> >>>>>> noticed that Pvplayer, when
behind a NAT/FW, sends a "FW packet"
> >>>>>> to the server, to poke a hole
through the NAT, so the UDP port
> >>>>>> mappings agreed in the SDP are
opened.  However, RealPlayer (not
> >>>>>> sure of version number) does
not send this, and can lead to dead
> > sessions.
> >>>>>> So my question is rather: when
type of NAT traversal techniques
> >>>>>> are currently implemented, or
planned to be implemented in
> >>>>>> clients, such as STUN or
RTSP-over-TCP, etc?
> >>>>>>
> >>>>>> I'm looking for discussion and
ideas.
> >>>>>>
> >>>>>> Reference:
> >>>>>> http://ietf.org/internet-drafts/draft-ietf-mmusic-
rtsp-nat-eva
> >>>>>> luation-00.txt
> >>>>>>
> >>>>>>
> >>>>>> --
> >>>>>>
> >>>>>> thanks,
> >>>>>> Yusuf
> >>>>>>
> >>>>>>
_______________________________________________
> >>>>>> Protocol-dev mailing list
> >>>>>> Protocol-devhelixcommunity.org
> >>>>>> http://lists.helixcommunity.org/mailman/listinfo/pr
otocol-dev
> >>>>>>
> >>>>
_______________________________________________
> >>>> Protocol-dev mailing list
> >>>> Protocol-devhelixcommunity.org
> >>>> http://lists.helixcommunity.org/mailman/listinfo/pr
otocol-dev
> >>>>
> >>
> >
> >
> > _______________________________________________
> > Porting-symbian mailing list
> > Porting-symbianhelixcommunity.org
> > http://lists.helixcommunity.org/mailman/listinfo
/porting-symbian
> >
> > _______________________________________________
> > Porting-symbian mailing list
> > Porting-symbianhelixcommunity.org
> > http://lists.helixcommunity.org/mailman/listinfo
/porting-symbian
> >
> >
>
>


-- 

thanks,
Yusuf

_______________________________________________
Protocol-dev mailing list
Protocol-devhelixcommunity.org
http://lists.helixcommunity.org/mailman/listinfo/pr
otocol-dev

Hi,

inline.

On 1/4/08, Chris Newman <chrisfloor51.com> wrote:
> Hi Yusuf,
>      thoughts inlined below:
>
> Yusuf Mayet wrote:
> > Thanks again.
> >
> > I started comparing the FW packets of pvlayer and
RealPlayer on N95
> > (User-Agent: RealMedia Player/mc.30.19.01 (s60;
epoc_av30_armv5)).
> >
> > Pvplayer:
> > Ethereal reports it as UDP 8 bytes long. Decoding
it as RTP just has
> > "Unknown RTP version 0"
> >
> > Real:
> > Ethereal reports it as UDP 28 bytes long. Decoding
it as RTP just has
> > some RTP headers, and a payload.
> >
> > Can anyone shed more light on the matter.  What
else can/do I look out
> > for.  The problem I am having, is that the RTSP
server seems to accept
> > pvplayer's FW poke packet, and therefore the
stream is successful,
> > while it rejects the N95's one, causing the stream
to fail.
>
> Are you sure the pvPlayer is UDP streaming and not
dropping back to TCP
> streaming? In my experience the microcore player never
drops back so if
> there are networking glitches with UDP it times out. If
you build from
> the Helix codebase you can configure the TCP drop back
timeout which works.
>

pvPlayer stays on UDP the whole time, and does not fail-back
to TCP.
So I know from the networking side that all is good.

> The case where I came across port poking was a
correctly configured
> firewall (with applied RTSP patch for the UDP port
allocation and
> possibly port address translation) but a misconfigured
load balancer
> around the RTSP server. In this instance the port
poking from the
> Realplayers worked but the Sony Ericssons (non-Symbian
based) without it
> didn't. If the firewall and load balancer are
misconfigured though I
> don't think the port poking would make a difference and
so you would TCP
> stream on the pvPlayer but the microcore Realplayer
would time out.
>

pvPlayer (the one I have tested with) is UDP streaming only,
so no TCP
for media.
In this case there is no load-balancer in front of the RTSP
server.
Also, this server uses UDP only for streaming.  Server uses
TCP for
RTSP only.

>From the doc I referenced in the start of the mail,
there are two
types of FW poke packets:
1.  RTP packets (No-Op)
2. UDP packets with no RTP headers

Which one does RP on N95 use?  It would seem to that pv uses
.2, while
RP on N95 uses some sort of RTP packet.  When the server
gets this
packet it complains about the SSRC contained within it, and
maybe
therefore rejects it.

> chris
>
> >
> > On 1/3/08, Chris Newman <chrisfloor51.com> wrote:
> >> Ashish.As.Guptanokia.com wrote:
> >>> N95 is not based on the helix and it uses
the microcore.
> >> Will the main Atlas Helix be in the default
player on the next S60
> >> release v3.2?
> >>
> >>> As far as the poke packet is concerned, if
I remember correctly, N95
> >>> based player does send the poke packet.
> >> Fairly sure it does. Think the N70 (fairly
old, S60v2 Symbian OS 8.0a)
> >> does as well (unless this was the one used for
testing). Maybe the
> >> 6600(OS7.0s)/6630(OS8.0a)/6680(OS8.0a) don't.
> >>
> >> chris
> >>
> >>> Ashish
> >>>
> >>> -----Original Message-----
> >>> From: porting-symbian-bounceshelixcommunity.org
> >>> [mailto:porting-symbian-bounceshelixcommunity.org] On Behalf Of ext
> >>> Greg Wright
> >>> Sent: Thursday, January 03, 2008 11:58 AM
> >>> To: Yusuf Mayet
> >>> Cc: porting-symbianhelixcommunity.org;
protocol-devhelixcommunity.org
> >>> Subject: [Porting-symbian] Re:
[Protocol-dev] OT: Question about FW
> >>> packetand NAT traversal
> >>>
> >>> Adding the Symbian list for people
familiar with the N95.
> >>>
> >>> If the phone allows it, you can always
download a Symbian RealPlayer
> >>> from the Helix site and install it. That
way you would know for sure.
> >>>
> >>> --greg.
> >>>
> >>>
> >>> Yusuf Mayet wrote:
> >>>> Hi,
> >>>>
> >>>> thanks for the responses.  Yes, the
phone I used is quite old, I will
> >>>> try a newer one.  Will the Nokia N95
be Helix-based, and will it
> >>>> therefore have NAT poking?
> >>>>
> >>>> On 1/3/08, Greg Wright <gwrightreal.com> wrote:
> >>>>> Henry Ping wrote:
> >>>>>> The poke-port code was added
in mid 2004 and should be in the most
> >>>>>> recent RealPlayer release.
> >>>>>>
> >>>>>> We use "Symmetric
RTP" and fallback to "TCP Tunneling" as
described
> >>>>>> in the reference link you
provided. It's been working well for us.
> >>>>>>
> >>>>>> In addition, we also provided
RTSP proxy(Application Level Gateway),
> >>>>> though
> >>>>>> I think it's being phased out
by Real.
> >>>>> If the phone is old, it will have
a micro-core based player on it and
> >>>>> not Helix based. In that case it
is very old and may not have NAT
> >>>>> poking.
> >>>>>
> >>>>> --greg.
> >>>>>> Henry
> >>>>>>
> >>>>>>> -----Original
Message-----
> >>>>>>> From: Eric Hyche
[mailto:ehychereal.com]
> >>>>>>> Sent: Wednesday, January
02, 2008 9:46 AM
> >>>>>>> To: 'Yusuf Mayet';
protocol-devhelixcommunity.org
> >>>>>>> Cc: Henry Ping
> >>>>>>> Subject: RE:
[Protocol-dev] OT: Question about FW packet and NAT
> >>>>> traversal
> >>>>>>> Poke-port code was added
fairly recently (within the last year
> >>>>>>> perhaps?) to the Helix
code base, so it is no surprise to me that
> >>>>>>> it might not be in the
RealPlayer version that you tried, given
> >>>>>>> that RealPlayer tends to
lag behind the latest Helix release.
> >>>>>>>
> >>>>>>> As far as what type of NAT
traversal techniques are used, Henry
> >>>>>>> added this code, I
believe, and he would know better how to answer.
> >>>>>>> I'm copying him on this
reply.
> >>>>>>>
> >>>>>>> Eric
> >>>>>>>
> >>>>>>>
=============================================
> >>>>>>> Eric Hyche (ehychereal.com)
> >>>>>>> Technical Lead
> >>>>>>> RealNetworks, Inc.
> >>>>>>>
> >>>>>>>> -----Original
Message-----
> >>>>>>>> From:
protocol-dev-bounceshelixcommunity.org
> >>>>>>>>
[mailto:protocol-dev-bounceshelixcommunity.org] On
Behalf Of
> >>>>>>>> Yusuf Mayet
> >>>>>>>> Sent: Friday, December
28, 2007 4:33 AM
> >>>>>>>> To: protocol-devhelixcommunity.org
> >>>>>>>> Subject:
[Protocol-dev] OT: Question about FW packet and NAT
> >>>>>>>> traversal
> >>>>>>>>
> >>>>>>>> Hi,
> >>>>>>>>
> >>>>>>>> This might be OT.
> >>>>>>>>
> >>>>>>>> When looking at RTSP
clients, especially on cellphone handsets, I
> >>>>>>>> noticed that Pvplayer,
when behind a NAT/FW, sends a "FW packet"
> >>>>>>>> to the server, to poke
a hole through the NAT, so the UDP port
> >>>>>>>> mappings agreed in the
SDP are opened.  However, RealPlayer (not
> >>>>>>>> sure of version
number) does not send this, and can lead to dead
> >>> sessions.
> >>>>>>>> So my question is
rather: when type of NAT traversal techniques
> >>>>>>>> are currently
implemented, or planned to be implemented in
> >>>>>>>> clients, such as STUN
or RTSP-over-TCP, etc?
> >>>>>>>>
> >>>>>>>> I'm looking for
discussion and ideas.
> >>>>>>>>
> >>>>>>>> Reference:
> >>>>>>>> http://ietf.org/internet-drafts/draft-ietf-mmusic-
rtsp-nat-eva
> >>>>>>>> luation-00.txt
> >>>>>>>>
> >>>>>>>>
> >>>>>>>> --
> >>>>>>>>
> >>>>>>>> thanks,
> >>>>>>>> Yusuf
> >>>>>>>>
> >>>>>>>>
_______________________________________________
> >>>>>>>> Protocol-dev mailing
list
> >>>>>>>> Protocol-devhelixcommunity.org
> >>>>>>>> http://lists.helixcommunity.org/mailman/listinfo/pr
otocol-dev
> >>>>>>>>
> >>>>>>
_______________________________________________
> >>>>>> Protocol-dev mailing list
> >>>>>> Protocol-devhelixcommunity.org
> >>>>>> http://lists.helixcommunity.org/mailman/listinfo/pr
otocol-dev
> >>>>>>
> >>>
> >>>
_______________________________________________
> >>> Porting-symbian mailing list
> >>> Porting-symbianhelixcommunity.org
> >>> http://lists.helixcommunity.org/mailman/listinfo
/porting-symbian
> >>>
> >>>
_______________________________________________
> >>> Porting-symbian mailing list
> >>> Porting-symbianhelixcommunity.org
> >>> http://lists.helixcommunity.org/mailman/listinfo
/porting-symbian
> >>>
> >>>
> >>
> >
> >
>
>


-- 

thanks,
Yusuf

_______________________________________________
Protocol-dev mailing list
Protocol-devhelixcommunity.org
http://lists.helixcommunity.org/mailman/listinfo/pr
otocol-dev