[Red5devs] [Fwd: RTMPS and the opening of multiple tcp ports, thus needing synchronization]

[Alexander Wichers]

Hi Daniel,

What I'm interested in is if you've seen different destination ports 
being opened in wireshark? So far I've only logged RTMPTE and that 
doesn't show any of that...

The FMS SSL configuration is not for the faint hearted (there are to 
many options for me) so I haven't looked at it just yet, and the thing 
you're describing sounds awful to debug...

Daniel Rossi wrote:
> As of now RTMPS is not functioning correctly it seems in Flash Player 10 
> / Flex 3, I had tested in Flash 9 and was functioning fine using the 
> self signed generated certificate. Its very odd, at one stage I was 
> getting NetConnection.Connect.CertificateUntrustedSigner errors, I 
> re-generate the cert and now it won't connect at all just hangs waiting 
> to connect. I have also noticed in the past rtmps will only work on port 
> 443 without the port added to the url or it needed the port not sure 
> tried both and still no luck now. There is documentation in flash player 
> 9.124 regards to not accepting  self signed certificates anymore.
> 
> Sorry if this is not what you are after just my observations.
> 
> On 30/10/2008, at 6:20 PM, Alexander Wichers wrote:
> 
>> Hi Core Developers,
>>
>> I have implemented the full protocol range (RTMPS/RTMPT/RTMPE, etc..)
>> within my own project (mammoth-project.org). And I see strange tcp port
>> behavior on RTMPTE and RTMPS, I know you guys have come across this same
>> issue as I see a mutex around every function in BaseRTMPTConnection.java.
>>
>> The 'problem' is that the client opens unneeded multiple tcp connections
>> to the server (on RTMPS and RTMPTE only!). Until now I haven't seen this
>> behavior in FMS and I actually observe port re-usage rather than opening
>> a new port for every new connection that is made after the old one.
>>
>> IMHO. it has to do with the RTMPT connection (/idle/mYemS4vRu2C2O91a/0
>> for ex.) string which the client checks to see if it's talking to a
>> 'real' server. I've tried to reverse engineer this string and it turns
>> out it's an encrypted set of ids that are converted to a modified base64
>> string, but.. that doesn't solve the problem.
>>
>> My question: Have you guys seen this behavior on FMS and why do you
>> think it happens?
>>
>> Thanks!
>> Alexander
>>
>> _______________________________________________
>> Red5devs mailing list
>> Red5devs at osflash.org
>> http://osflash.org/mailman/listinfo/red5devs_osflash.org
> 
> 
> _______________________________________________
> Red5devs mailing list
> Red5devs at osflash.org
> http://osflash.org/mailman/listinfo/red5devs_osflash.org