[jitsi-dev] More serverside issues


#1

Here again I am attempting to use libjitsi on a server-side app. I just had
a crash just now and if anyone can help decipher the issue it would be
awesome. As an overview, I have no need of a decoder for vpx and I am only
setting up and using a vpx encoder and packetizer. So my question is this:
why are a vpx decoder and depacketizer being spawned?

Server log (show sysprops and fmj log entries): http://pastebin.com/dvvq8nMX
hs-err-pid log (top section): http://pastebin.com/BA2yGtUE

Other questions:
1. Why does setting 100 as my dynamic RTP type on my stream cause an
"ERROR"?
videoMediaStream.addDynamicRTPPayloadType((byte) 100, videoFormat);

2. Will the above error cause any issues I need to be aware of?

3. I used system prop examples from videobridge, but it would appear that
local video display is attempting to be configured or used by fmj or jitsi.
Are there additional properties that I need to set to force head-less
operation?

4. Is there a means for turning RTP stats off?

Thanks,
Paul

···

--
http://gregoire.org/
http://code.google.com/p/red5/


#2

I seem to be streaming data out now, I'll know for sure once I try to
consume it. I found that the VPX decoder seemed to be created because I had
my video MediaStream type set to VP8_RTP instead of just VP8; this was
confusing since I was informed that I should use OPUS_RTP on my audio
MediaStream. The opus jni encoder puts out the rtp type whereas the vpx
encoder just puts out vp8; thank goodness we have access to the sources. :slight_smile:

I remain a little concerned about the ERROR logged when i add the dynamic
payload type, but it doesn't seem to matter here.

Keep up the great work!

Paul

···

On Wed, Aug 27, 2014 at 2:26 PM, Mondain <mondain@gmail.com> wrote:

Here again I am attempting to use libjitsi on a server-side app. I just
had a crash just now and if anyone can help decipher the issue it would be
awesome. As an overview, I have no need of a decoder for vpx and I am only
setting up and using a vpx encoder and packetizer. So my question is this:
why are a vpx decoder and depacketizer being spawned?

Server log (show sysprops and fmj log entries):
http://pastebin.com/dvvq8nMX
hs-err-pid log (top section): http://pastebin.com/BA2yGtUE

Other questions:
1. Why does setting 100 as my dynamic RTP type on my stream cause an
"ERROR"?
videoMediaStream.addDynamicRTPPayloadType((byte) 100, videoFormat);

2. Will the above error cause any issues I need to be aware of?

3. I used system prop examples from videobridge, but it would appear that
local video display is attempting to be configured or used by fmj or jitsi.
Are there additional properties that I need to set to force head-less
operation?

4. Is there a means for turning RTP stats off?

Thanks,
Paul
--
http://gregoire.org/
http://code.google.com/p/red5/

--
http://gregoire.org/
http://code.google.com/p/red5/


#3

For reference here are the ERROR entries that I see, for both vp8 and
opus_rtp:

2014-08-28 15:45:11,830 [pool-9-thread-2] INFO fmj - Wrote registry file:
/root/.fmj.registry.xml
2014-08-28 15:45:12,810 [pool-9-thread-2] INFO
o.j.i.n.d.VideoMediaDeviceSession - video send resolution: 320x240
2014-08-28 15:45:12,810 [pool-9-thread-2] INFO
o.j.i.n.d.VideoMediaDeviceSession - video send FPS: default(no restriction)
2014-08-28 15:45:13,017 [pool-9-thread-2] INFO
o.j.impl.neomedia.MediaStreamImpl - video codec/freq: VP8/90000 Hz
2014-08-28 15:45:13,018 [pool-9-thread-2] INFO
o.j.impl.neomedia.MediaStreamImpl - video remote IP/port: my.ip.ad.dr/5000
2014-08-28 15:45:13,118 [Loop thread:
net.sf.fmj.media.parser.RawPullBufferParser$FrameTrack@40135757] INFO
o.j.i.n.codec.video.vp8.VPXEncoder - Setting new width/height: 320/240
2014-08-28 15:45:13,262 [RTP Forwarder:
org.jitsi.impl.neomedia.MediaStreamImpl$2@18db941e] ERROR
net.sf.fmj.media.Log - No format has been registered for RTP Payload type
100
2014-08-28 15:45:14,805 [pool-9-thread-2] INFO
o.j.impl.neomedia.MediaStreamImpl - audio codec/freq: opus/48000 Hz
2014-08-28 15:45:14,805 [pool-9-thread-2] INFO
o.j.impl.neomedia.MediaStreamImpl - audio remote IP/port: my.ip.ad.dr/5000
2014-08-28 15:45:14,857 [RTP Forwarder:
org.jitsi.impl.neomedia.MediaStreamImpl$2@4d764e80] ERROR
net.sf.fmj.media.Log - No format has been registered for RTP Payload type 9

···

On Thu, Aug 28, 2014 at 10:51 AM, Mondain <mondain@gmail.com> wrote:

I seem to be streaming data out now, I'll know for sure once I try to
consume it. I found that the VPX decoder seemed to be created because I had
my video MediaStream type set to VP8_RTP instead of just VP8; this was
confusing since I was informed that I should use OPUS_RTP on my audio
MediaStream. The opus jni encoder puts out the rtp type whereas the vpx
encoder just puts out vp8; thank goodness we have access to the sources. :slight_smile:

I remain a little concerned about the ERROR logged when i add the dynamic
payload type, but it doesn't seem to matter here.

Keep up the great work!

Paul

On Wed, Aug 27, 2014 at 2:26 PM, Mondain <mondain@gmail.com> wrote:

Here again I am attempting to use libjitsi on a server-side app. I just
had a crash just now and if anyone can help decipher the issue it would be
awesome. As an overview, I have no need of a decoder for vpx and I am only
setting up and using a vpx encoder and packetizer. So my question is this:
why are a vpx decoder and depacketizer being spawned?

Server log (show sysprops and fmj log entries):
http://pastebin.com/dvvq8nMX
hs-err-pid log (top section): http://pastebin.com/BA2yGtUE

Other questions:
1. Why does setting 100 as my dynamic RTP type on my stream cause an
"ERROR"?
videoMediaStream.addDynamicRTPPayloadType((byte) 100, videoFormat);

2. Will the above error cause any issues I need to be aware of?

3. I used system prop examples from videobridge, but it would appear that
local video display is attempting to be configured or used by fmj or jitsi.
Are there additional properties that I need to set to force head-less
operation?

4. Is there a means for turning RTP stats off?

Thanks,
Paul
--
http://gregoire.org/
http://code.google.com/p/red5/

--
http://gregoire.org/
http://code.google.com/p/red5/

--
http://gregoire.org/
http://code.google.com/p/red5/


#4

I seem to be streaming data out now, I'll know for sure once I try to
consume it. I found that the VPX decoder seemed to be created because I
had
my video MediaStream type set to VP8_RTP instead of just VP8; this was
confusing since I was informed that I should use OPUS_RTP on my audio
MediaStream. The opus jni encoder puts out the rtp type whereas the vpx
encoder just puts out vp8; thank goodness we have access to the
sources. :slight_smile:

Exactly, "VP8" is the output from the encoder. And what you need for streaming over RTP is "VP8_RTP", which is the output of the packetizer.

Boris

···

On August 28, 2014 5:51:56 PM EEST, Mondain <mondain@gmail.com> wrote:

I remain a little concerned about the ERROR logged when i add the
dynamic
payload type, but it doesn't seem to matter here.

Keep up the great work!

Paul

On Wed, Aug 27, 2014 at 2:26 PM, Mondain <mondain@gmail.com> wrote:

Here again I am attempting to use libjitsi on a server-side app. I

just

had a crash just now and if anyone can help decipher the issue it

would be

awesome. As an overview, I have no need of a decoder for vpx and I am

only

setting up and using a vpx encoder and packetizer. So my question is

this:

why are a vpx decoder and depacketizer being spawned?

Server log (show sysprops and fmj log entries):
http://pastebin.com/dvvq8nMX
hs-err-pid log (top section): http://pastebin.com/BA2yGtUE

Other questions:
1. Why does setting 100 as my dynamic RTP type on my stream cause an
"ERROR"?
videoMediaStream.addDynamicRTPPayloadType((byte) 100, videoFormat);

2. Will the above error cause any issues I need to be aware of?

3. I used system prop examples from videobridge, but it would appear

that

local video display is attempting to be configured or used by fmj or

jitsi.

Are there additional properties that I need to set to force head-less
operation?

4. Is there a means for turning RTP stats off?

Thanks,
Paul
--
http://gregoire.org/
http://code.google.com/p/red5/

--
Sent from a mobile. Please excuse my brevity.


#5

So I should assume that specifying OPUS_RTP is correct for the audio stream
and VP8 for the video stream? I am getting this output in my log now with
these changes: http://pastebin.com/JQZfPJSx
Just for a little more detail on my side, does a Processor or Transformer
know that it should packetize the VP8 for RTP on its own?

Paul

···

On Fri, Aug 29, 2014 at 2:10 AM, Boris Grozev <boris@jitsi.org> wrote:

On August 28, 2014 5:51:56 PM EEST, Mondain <mondain@gmail.com> wrote:
>I seem to be streaming data out now, I'll know for sure once I try to
>consume it. I found that the VPX decoder seemed to be created because I
>had
>my video MediaStream type set to VP8_RTP instead of just VP8; this was
>confusing since I was informed that I should use OPUS_RTP on my audio
>MediaStream. The opus jni encoder puts out the rtp type whereas the vpx
>encoder just puts out vp8; thank goodness we have access to the
>sources. :slight_smile:

Exactly, "VP8" is the output from the encoder. And what you need for
streaming over RTP is "VP8_RTP", which is the output of the packetizer.

Boris
>
>I remain a little concerned about the ERROR logged when i add the
>dynamic
>payload type, but it doesn't seem to matter here.
>
>Keep up the great work!
>
>Paul
>
>
>On Wed, Aug 27, 2014 at 2:26 PM, Mondain <mondain@gmail.com> wrote:
>
>> Here again I am attempting to use libjitsi on a server-side app. I
>just
>> had a crash just now and if anyone can help decipher the issue it
>would be
>> awesome. As an overview, I have no need of a decoder for vpx and I am
>only
>> setting up and using a vpx encoder and packetizer. So my question is
>this:
>> why are a vpx decoder and depacketizer being spawned?
>>
>> Server log (show sysprops and fmj log entries):
>> http://pastebin.com/dvvq8nMX
>> hs-err-pid log (top section): http://pastebin.com/BA2yGtUE
>>
>> Other questions:
>> 1. Why does setting 100 as my dynamic RTP type on my stream cause an
>> "ERROR"?
>> videoMediaStream.addDynamicRTPPayloadType((byte) 100, videoFormat);
>>
>> 2. Will the above error cause any issues I need to be aware of?
>>
>> 3. I used system prop examples from videobridge, but it would appear
>that
>> local video display is attempting to be configured or used by fmj or
>jitsi.
>> Are there additional properties that I need to set to force head-less
>> operation?
>>
>> 4. Is there a means for turning RTP stats off?
>>
>> Thanks,
>> Paul
>> --
>> http://gregoire.org/
>> http://code.google.com/p/red5/
>>

--
Sent from a mobile. Please excuse my brevity.

--
http://gregoire.org/
http://code.google.com/p/red5/


#6

So I should assume that specifying OPUS_RTP is correct for the audio
stream
and VP8 for the video stream? I am getting this output in my log now
with
these changes: http://pastebin.com/JQZfPJSx
Just for a little more detail on my side, does a Processor or
Transformer
know that it should packetize the VP8 for RTP on its own?

Depends on the output format - VP8 is not packetized, VP8_RTP is.

Boris

···

On August 29, 2014 9:09:12 PM EEST, Mondain <mondain@gmail.com> wrote:

Paul

On Fri, Aug 29, 2014 at 2:10 AM, Boris Grozev <boris@jitsi.org> wrote:

On August 28, 2014 5:51:56 PM EEST, Mondain <mondain@gmail.com> >wrote:
>I seem to be streaming data out now, I'll know for sure once I try

to

>consume it. I found that the VPX decoder seemed to be created

because I

>had
>my video MediaStream type set to VP8_RTP instead of just VP8; this

was

>confusing since I was informed that I should use OPUS_RTP on my

audio

>MediaStream. The opus jni encoder puts out the rtp type whereas the

vpx

>encoder just puts out vp8; thank goodness we have access to the
>sources. :slight_smile:

Exactly, "VP8" is the output from the encoder. And what you need for
streaming over RTP is "VP8_RTP", which is the output of the

packetizer.

Boris
>
>I remain a little concerned about the ERROR logged when i add the
>dynamic
>payload type, but it doesn't seem to matter here.
>
>Keep up the great work!
>
>Paul
>
>
>On Wed, Aug 27, 2014 at 2:26 PM, Mondain <mondain@gmail.com> wrote:
>
>> Here again I am attempting to use libjitsi on a server-side app. I
>just
>> had a crash just now and if anyone can help decipher the issue it
>would be
>> awesome. As an overview, I have no need of a decoder for vpx and I

am

>only
>> setting up and using a vpx encoder and packetizer. So my question

is

>this:
>> why are a vpx decoder and depacketizer being spawned?
>>
>> Server log (show sysprops and fmj log entries):
>> http://pastebin.com/dvvq8nMX
>> hs-err-pid log (top section): http://pastebin.com/BA2yGtUE
>>
>> Other questions:
>> 1. Why does setting 100 as my dynamic RTP type on my stream cause

an

>> "ERROR"?
>> videoMediaStream.addDynamicRTPPayloadType((byte) 100,

videoFormat);

>>
>> 2. Will the above error cause any issues I need to be aware of?
>>
>> 3. I used system prop examples from videobridge, but it would

appear

>that
>> local video display is attempting to be configured or used by fmj

or

>jitsi.
>> Are there additional properties that I need to set to force

head-less

>> operation?
>>
>> 4. Is there a means for turning RTP stats off?
>>
>> Thanks,
>> Paul
>> --
>> http://gregoire.org/
>> http://code.google.com/p/red5/
>>

--
Sent from a mobile. Please excuse my brevity.

--
Sent from a mobile. Please excuse my brevity.


#7

My output is to be consumed by a WebRTC client, so does VP8 need to be
packetized in that case?

···

On Sat, Aug 30, 2014 at 2:53 AM, Boris Grozev <boris@jitsi.org> wrote:

On August 29, 2014 9:09:12 PM EEST, Mondain <mondain@gmail.com> wrote:
>So I should assume that specifying OPUS_RTP is correct for the audio
>stream
>and VP8 for the video stream? I am getting this output in my log now
>with
>these changes: http://pastebin.com/JQZfPJSx
>Just for a little more detail on my side, does a Processor or
>Transformer
>know that it should packetize the VP8 for RTP on its own?

Depends on the output format - VP8 is not packetized, VP8_RTP is.

Boris
>
>Paul
>
>
>On Fri, Aug 29, 2014 at 2:10 AM, Boris Grozev <boris@jitsi.org> wrote:
>
>> On August 28, 2014 5:51:56 PM EEST, Mondain <mondain@gmail.com> > >wrote:
>> >I seem to be streaming data out now, I'll know for sure once I try
>to
>> >consume it. I found that the VPX decoder seemed to be created
>because I
>> >had
>> >my video MediaStream type set to VP8_RTP instead of just VP8; this
>was
>> >confusing since I was informed that I should use OPUS_RTP on my
>audio
>> >MediaStream. The opus jni encoder puts out the rtp type whereas the
>vpx
>> >encoder just puts out vp8; thank goodness we have access to the
>> >sources. :slight_smile:
>>
>> Exactly, "VP8" is the output from the encoder. And what you need for
>> streaming over RTP is "VP8_RTP", which is the output of the
>packetizer.
>>
>>
>> Boris
>> >
>> >I remain a little concerned about the ERROR logged when i add the
>> >dynamic
>> >payload type, but it doesn't seem to matter here.
>> >
>> >Keep up the great work!
>> >
>> >Paul
>> >
>> >
>> >On Wed, Aug 27, 2014 at 2:26 PM, Mondain <mondain@gmail.com> wrote:
>> >
>> >> Here again I am attempting to use libjitsi on a server-side app. I
>> >just
>> >> had a crash just now and if anyone can help decipher the issue it
>> >would be
>> >> awesome. As an overview, I have no need of a decoder for vpx and I
>am
>> >only
>> >> setting up and using a vpx encoder and packetizer. So my question
>is
>> >this:
>> >> why are a vpx decoder and depacketizer being spawned?
>> >>
>> >> Server log (show sysprops and fmj log entries):
>> >> http://pastebin.com/dvvq8nMX
>> >> hs-err-pid log (top section): http://pastebin.com/BA2yGtUE
>> >>
>> >> Other questions:
>> >> 1. Why does setting 100 as my dynamic RTP type on my stream cause
>an
>> >> "ERROR"?
>> >> videoMediaStream.addDynamicRTPPayloadType((byte) 100,
>videoFormat);
>> >>
>> >> 2. Will the above error cause any issues I need to be aware of?
>> >>
>> >> 3. I used system prop examples from videobridge, but it would
>appear
>> >that
>> >> local video display is attempting to be configured or used by fmj
>or
>> >jitsi.
>> >> Are there additional properties that I need to set to force
>head-less
>> >> operation?
>> >>
>> >> 4. Is there a means for turning RTP stats off?
>> >>
>> >> Thanks,
>> >> Paul
>> >> --
>> >> http://gregoire.org/
>> >> http://code.google.com/p/red5/
>> >>
>>
>>
>> --
>> Sent from a mobile. Please excuse my brevity.
>>

--
Sent from a mobile. Please excuse my brevity.

--
http://gregoire.org/
http://code.google.com/p/red5/


#8

Yes

Boris

···

On August 31, 2014 8:56:21 PM EEST, Mondain <mondain@gmail.com> wrote:

My output is to be consumed by a WebRTC client, so does VP8 need to be
packetized in that case?

On Sat, Aug 30, 2014 at 2:53 AM, Boris Grozev <boris@jitsi.org> wrote:

On August 29, 2014 9:09:12 PM EEST, Mondain <mondain@gmail.com> >wrote:
>So I should assume that specifying OPUS_RTP is correct for the audio
>stream
>and VP8 for the video stream? I am getting this output in my log now
>with
>these changes: http://pastebin.com/JQZfPJSx
>Just for a little more detail on my side, does a Processor or
>Transformer
>know that it should packetize the VP8 for RTP on its own?

Depends on the output format - VP8 is not packetized, VP8_RTP is.

Boris
>
>Paul
>
>
>On Fri, Aug 29, 2014 at 2:10 AM, Boris Grozev <boris@jitsi.org> >wrote:
>
>> On August 28, 2014 5:51:56 PM EEST, Mondain <mondain@gmail.com> >> >wrote:
>> >I seem to be streaming data out now, I'll know for sure once I

try

>to
>> >consume it. I found that the VPX decoder seemed to be created
>because I
>> >had
>> >my video MediaStream type set to VP8_RTP instead of just VP8;

this

>was
>> >confusing since I was informed that I should use OPUS_RTP on my
>audio
>> >MediaStream. The opus jni encoder puts out the rtp type whereas

the

>vpx
>> >encoder just puts out vp8; thank goodness we have access to the
>> >sources. :slight_smile:
>>
>> Exactly, "VP8" is the output from the encoder. And what you need

for

>> streaming over RTP is "VP8_RTP", which is the output of the
>packetizer.
>>
>>
>> Boris
>> >
>> >I remain a little concerned about the ERROR logged when i add the
>> >dynamic
>> >payload type, but it doesn't seem to matter here.
>> >
>> >Keep up the great work!
>> >
>> >Paul
>> >
>> >
>> >On Wed, Aug 27, 2014 at 2:26 PM, Mondain <mondain@gmail.com> >wrote:
>> >
>> >> Here again I am attempting to use libjitsi on a server-side

app. I

>> >just
>> >> had a crash just now and if anyone can help decipher the issue

it

>> >would be
>> >> awesome. As an overview, I have no need of a decoder for vpx

and I

>am
>> >only
>> >> setting up and using a vpx encoder and packetizer. So my

question

>is
>> >this:
>> >> why are a vpx decoder and depacketizer being spawned?
>> >>
>> >> Server log (show sysprops and fmj log entries):
>> >> http://pastebin.com/dvvq8nMX
>> >> hs-err-pid log (top section): http://pastebin.com/BA2yGtUE
>> >>
>> >> Other questions:
>> >> 1. Why does setting 100 as my dynamic RTP type on my stream

cause

>an
>> >> "ERROR"?
>> >> videoMediaStream.addDynamicRTPPayloadType((byte) 100,
>videoFormat);
>> >>
>> >> 2. Will the above error cause any issues I need to be aware of?
>> >>
>> >> 3. I used system prop examples from videobridge, but it would
>appear
>> >that
>> >> local video display is attempting to be configured or used by

fmj

>or
>> >jitsi.
>> >> Are there additional properties that I need to set to force
>head-less
>> >> operation?
>> >>
>> >> 4. Is there a means for turning RTP stats off?
>> >>
>> >> Thanks,
>> >> Paul
>> >> --
>> >> http://gregoire.org/
>> >> http://code.google.com/p/red5/
>> >>
>>
>>
>> --
>> Sent from a mobile. Please excuse my brevity.
>>

--
Sent from a mobile. Please excuse my brevity.

--
Sent from a mobile. Please excuse my brevity.