Re: [Red5] onMetaData for vod
Hi Dan, Thanks alot for your very clear and concise explanation. It will help us all so much as we do our tests. Also, it will now be much easier to help out with patches where I can manage to understand the red5 streaming issues; it is very hard! Thanks once again. On 3/21/07, Dan Rossi [EMAIL PROTECTED] wrote: Hi, Hi Dan, You have been testing live streaming much longer than most of us so if I get you right, red5 has had the following issues for live streaming: Im not sure about longer my very first post here was july last year ?? :) 1. Heap space errors for large files. This is because of mina. Sorry i think I got you confused, there was java memory heap errors with red5 last november for videos larger than 46MB to be exact. This was fixed. The mina issue i mentioned about was suggested to me here http://jira.red5.org/browse/SN-1 Ive logged every single comparison test i can do here, it was painful and time consuming :\ the problem with vod buffering problems which red5 seems to have an issue with latency and is unable to handle the video data to push more bits down the connection to keep the buffer length in check which seems both FMS and wowza seem to do. So if you are a few k's from your server in a data centre you are fine and buffer lengths of 1 minute, if you are inter-continental across underwater cabling like I am then the problems start to arise i think. Well we've had issues with people broadcasting from brazil to our DC in NY, and from Sydney to NY. Its that bad we've had to move the short clips back to progressive download as i did in november, however we added scrub features into our player so was just a test to see how stable red5 is with it. So in terms of memory and resources stability is perfect maybe a little more cpu than i like (P4 3G 2G of ram) about 20% usage so far compared to IIS with the same traffic its pretty much 0% :) 2. Buffering issues. What do you think is causing buffering problems? See above. 3. Live delay. Audio/Video is out of sync? What do you think is causing this for live streaming? Well from the previous comments live delay shows up as 0 , i get the negative timestamp warnings, its streaming at 150k so no problem with buffering however a 180k vod still has ther same buffer problem on a 24MB adsl conn so its not that ! We've teried every config pososible it seems streaming at 80 video quality seems to make the frames stick less, streaming at 30fps seems to give alot of headroom to accomodate on playback so red5 is doing between 9 and 16fps so thats ok. The issue lots of us are experiencing as with my case is it will play the frames smooth for 3 seconds, stick, play frames smooth for another 3-5 seconds stick and over and over. Im just so hoping its related the the vod problem so if thats fixed so is the live issue. Back to the comparison, red5 streamed live better than fms which caused rebuffering, maybe its because of the bandwidth limits set but i got no error about it. Wowza rebuffered also on live streaming, but both wowza and fms was perfect with vod so cancels out bad route issues to servers. What other issues might you be having? I seem to be hitting the same issues you speak about. For example, when I begin loading files over 300MB, it sometimes fails to play. the only issue i have is on first load the buffer of 8 seconds is empty again so rebuffers and then when it goes to play it starts at 8 seconds not 0 seconds ! No major issue with 300MB files. ___ Red5 mailing list Red5@osflash.org http://osflash.org/mailman/listinfo/red5_osflash.org -- C is forever. ___ Red5 mailing list Red5@osflash.org http://osflash.org/mailman/listinfo/red5_osflash.org
Re: [Red5] onMetaData for vod
Ok cool as i said i have a server in a DC in Sydney Australia which may be perfect to check out latency issues. Im prob not so cool to put the ip in an archived list so i guess someone email me offlist for the url :\ joseph wamicha wrote: Hi Dan, Thanks alot for your very clear and concise explanation. It will help us all so much as we do our tests. Also, it will now be much easier to help out with patches where I can manage to understand the red5 streaming issues; it is very hard! Thanks once again. On 3/21/07, Dan Rossi [EMAIL PROTECTED] wrote: Hi, Hi Dan, You have been testing live streaming much longer than most of us so if I get you right, red5 has had the following issues for live streaming: Im not sure about longer my very first post here was july last year ?? :) 1. Heap space errors for large files. This is because of mina. Sorry i think I got you confused, there was java memory heap errors with red5 last november for videos larger than 46MB to be exact. This was fixed. The mina issue i mentioned about was suggested to me here http://jira.red5.org/browse/SN-1 Ive logged every single comparison test i can do here, it was painful and time consuming :\ the problem with vod buffering problems which red5 seems to have an issue with latency and is unable to handle the video data to push more bits down the connection to keep the buffer length in check which seems both FMS and wowza seem to do. So if you are a few k's from your server in a data centre you are fine and buffer lengths of 1 minute, if you are inter-continental across underwater cabling like I am then the problems start to arise i think. Well we've had issues with people broadcasting from brazil to our DC in NY, and from Sydney to NY. Its that bad we've had to move the short clips back to progressive download as i did in november, however we added scrub features into our player so was just a test to see how stable red5 is with it. So in terms of memory and resources stability is perfect maybe a little more cpu than i like (P4 3G 2G of ram) about 20% usage so far compared to IIS with the same traffic its pretty much 0% :) 2. Buffering issues. What do you think is causing buffering problems? See above. 3. Live delay. Audio/Video is out of sync? What do you think is causing this for live streaming? Well from the previous comments live delay shows up as 0 , i get the negative timestamp warnings, its streaming at 150k so no problem with buffering however a 180k vod still has ther same buffer problem on a 24MB adsl conn so its not that ! We've teried every config pososible it seems streaming at 80 video quality seems to make the frames stick less, streaming at 30fps seems to give alot of headroom to accomodate on playback so red5 is doing between 9 and 16fps so thats ok. The issue lots of us are experiencing as with my case is it will play the frames smooth for 3 seconds, stick, play frames smooth for another 3-5 seconds stick and over and over. Im just so hoping its related the the vod problem so if thats fixed so is the live issue. Back to the comparison, red5 streamed live better than fms which caused rebuffering, maybe its because of the bandwidth limits set but i got no error about it. Wowza rebuffered also on live streaming, but both wowza and fms was perfect with vod so cancels out bad route issues to servers. What other issues might you be having? I seem to be hitting the same issues you speak about. For example, when I begin loading files over 300MB, it sometimes fails to play. the only issue i have is on first load the buffer of 8 seconds is empty again so rebuffers and then when it goes to play it starts at 8 seconds not 0 seconds ! No major issue with 300MB files. ___ Red5 mailing list Red5@osflash.org http://osflash.org/mailman/listinfo/red5_osflash.org ___ Red5 mailing list Red5@osflash.org http://osflash.org/mailman/listinfo/red5_osflash.org ___ Red5 mailing list Red5@osflash.org http://osflash.org/mailman/listinfo/red5_osflash.org
Re: [Red5] onMetaData for vod
There is software from On2 That works same way as FME, Not sure about licensing issues however price tag is high, But might help someone, It would be great to use it with Red5 (by the way it crashed just like FME on the trial version ) -Original Message- From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of Dan Rossi Sent: Viernes, 16 de Marzo de 2007 06:32 p.m. To: Red5@osflash.org Subject: Re: [Red5] onMetaData for vod Damn my apologies for the suggestion to support FME ? If it cant be supported, undoing this would be good and then i guess an open source project to build an equivalanet in flex or something, because the licencing required to purchase the C++ API ! joseph wamicha wrote: Looking at subversion logs; $svn log -r 1751:1761 -v http://svn1.cvsdude.com/osflash/red5/java/server/trunk I highly suspect this changes that added FME support are the cause: It talks about to correctly decode stream metadata. -- -- r1760 | jbauch | 2007-03-16 03:44:23 +0300 (Fri, 16 Mar 2007) | 5 lines Changed paths: M /doc/trunk/changelog.txt M /java/server/trunk/src/org/red5/io/amf/Output.java A /java/server/trunk/src/org/red5/io/object/ICustomSerializable.java A /java/server/trunk/src/org/red5/io/object/ISerializerOptionAware.java M /java/server/trunk/src/org/red5/io/object/Serializer.java M /java/server/trunk/src/org/red5/server/net/rtmp/RTMPHandler.java M /java/server/trunk/src/org/red5/server/net/rtmp/codec/RTMPProtocolDeco der.java M /java/server/trunk/src/org/red5/server/net/rtmp/status/Status.java M /java/server/trunk/src/org/red5/server/net/rtmp/status/StatusObject.java M /java/server/trunk/src/org/red5/server/stream/ClientBroadcastStream.ja va added interface to allow objects to serialize themselves added interface for objects that know about their serializer options send publish/record start events before the first packet is received correctly decode published stream metadata - Now Red5 supports being connected to from FME! -- -- r1761 | jbauch | 2007-03-16 04:05:26 +0300 (Fri, 16 Mar 2007) | 1 line Changed paths: M /java/server/trunk/src/org/red5/server/adapter/ApplicationAdapter.java M /java/server/trunk/src/org/red5/server/api/stream/IStreamService.java M /java/server/trunk/src/org/red5/server/net/rtmp/RTMPHandler.java M /java/server/trunk/src/org/red5/server/net/rtmp/message/Constants.java M /java/server/trunk/src/org/red5/server/stream/StreamService.java added FME specific methods -- -- On 3/17/07, Dan Rossi [EMAIL PROTECTED] wrote: Shit you are right its had to keep checking things to get something else to work. metadata is busted again. - --- ___ Red5 mailing list Red5@osflash.org http://osflash.org/mailman/listinfo/red5_osflash.org ___ Red5 mailing list Red5@osflash.org http://osflash.org/mailman/listinfo/red5_osflash.org ___ Red5 mailing list Red5@osflash.org http://osflash.org/mailman/listinfo/red5_osflash.org
Re: [Red5] onMetaData for vod
Hi, joseph wamicha schrieb: Problem: In revision 1761, the onMetaData method isn't being triggered or called. Thus when streaming vod currently, I can't seek and the video.height and video.width dimensions can not be set using metadata information in onMetaData function. I just commited a fix for the onMetaData problem - sorry for me breaking this! ;) Joachim ___ Red5 mailing list Red5@osflash.org http://osflash.org/mailman/listinfo/red5_osflash.org
Re: [Red5] onMetaData for vod
Hi Joachim! Thanks a million! If I may ask, where is onMetaData sent from in red5? I was really curious to know as it would be great to have onMetaData data for VOD within a live tv stream created through serverstream (simpleplaylist). Thanks once again. On 3/21/07, Joachim Bauch [EMAIL PROTECTED] wrote: Hi, joseph wamicha schrieb: Problem: In revision 1761, the onMetaData method isn't being triggered or called. Thus when streaming vod currently, I can't seek and the video.height and video.width dimensions can not be set using metadata information in onMetaData function. I just commited a fix for the onMetaData problem - sorry for me breaking this! ;) Joachim ___ Red5 mailing list Red5@osflash.org http://osflash.org/mailman/listinfo/red5_osflash.org -- C is forever. ___ Red5 mailing list Red5@osflash.org http://osflash.org/mailman/listinfo/red5_osflash.org
Re: [Red5] onMetaData for vod
Hi Joseph, joseph wamicha schrieb: If I may ask, where is onMetaData sent from in red5? I was really curious to know as it would be great to have onMetaData data for VOD within a live tv stream created through serverstream (simpleplaylist). the onMetaData usually is the first tag inside a FLV file and that is sent by PlaylistSubscriberStream.play(IPlayItem). Joachim ___ Red5 mailing list Red5@osflash.org http://osflash.org/mailman/listinfo/red5_osflash.org
Re: [Red5] onMetaData for vod
Ok updating now, what about the metakeyframe errors ? Joachim Bauch wrote: Hi, joseph wamicha schrieb: Problem: In revision 1761, the onMetaData method isn't being triggered or called. Thus when streaming vod currently, I can't seek and the video.height and video.width dimensions can not be set using metadata information in onMetaData function. I just commited a fix for the onMetaData problem - sorry for me breaking this! ;) Joachim ___ Red5 mailing list Red5@osflash.org http://osflash.org/mailman/listinfo/red5_osflash.org ___ Red5 mailing list Red5@osflash.org http://osflash.org/mailman/listinfo/red5_osflash.org
Re: [Red5] onMetaData for vod
Hi Steve, Yes it was working previously. On 3/17/07, Dan Rossi [EMAIL PROTECTED] wrote: How do i fix this we're getting complaints coz of video sizing. Dan Rossi wrote: the update to 1761 killed onMetadata. the other wierd thing is if i comment out the write methods in MetaService its still writing a metakey file so its happening elsewhere. Steven Gong wrote: Joseph, Do you mean the it works on 1751? On 3/17/07, joseph wamicha [EMAIL PROTECTED] wrote: Looking at subversion logs; $svn log -r 1751:1761 -v http://svn1.cvsdude.com/osflash/red5/java/server/trunk I highly suspect this changes that added FME support are the cause: It talks about to correctly decode stream metadata. r1760 | jbauch | 2007-03-16 03:44:23 +0300 (Fri, 16 Mar 2007) | 5 lines Changed paths: M /doc/trunk/changelog.txt M /java/server/trunk/src/org/red5/io/amf/Output.java A /java/server/trunk/src/org/red5/io/object/ICustomSerializable.java A /java/server/trunk/src/org/red5/io/object/ISerializerOptionAware.java M /java/server/trunk/src/org/red5/io/object/Serializer.java M /java/server/trunk/src/org/red5/server/net/rtmp/RTMPHandler.java M /java/server/trunk/src/org/red5/server/net/rtmp/codec/RTMPProtocolDecoder.java M /java/server/trunk/src/org/red5/server/net/rtmp/status/Status.java M /java/server/trunk/src/org/red5/server/net/rtmp/status/StatusObject.java M /java/server/trunk/src/org/red5/server/stream/ClientBroadcastStream.java added interface to allow objects to serialize themselves added interface for objects that know about their serializer options send publish/record start events before the first packet is received correctly decode published stream metadata - Now Red5 supports being connected to from FME! r1761 | jbauch | 2007-03-16 04:05:26 +0300 (Fri, 16 Mar 2007) | 1 line Changed paths: M /java/server/trunk/src/org/red5/server/adapter/ApplicationAdapter.java M /java/server/trunk/src/org/red5/server/api/stream/IStreamService.java M /java/server/trunk/src/org/red5/server/net/rtmp/RTMPHandler.java M /java/server/trunk/src/org/red5/server/net/rtmp/message/Constants.java M /java/server/trunk/src/org/red5/server/stream/StreamService.java added FME specific methods On 3/17/07, Dan Rossi [EMAIL PROTECTED] wrote: Shit you are right its had to keep checking things to get something else to work. metadata is busted again. -- C is forever. ___ Red5 mailing list Red5@osflash.org http://osflash.org/mailman/listinfo/red5_osflash.org ___ Red5 mailing list Red5@osflash.org http://osflash.org/mailman/listinfo/red5_osflash.org ___ Red5 mailing list Red5@osflash.org http://osflash.org/mailman/listinfo/red5_osflash.org ___ Red5 mailing list Red5@osflash.org http://osflash.org/mailman/listinfo/red5_osflash.org -- C is forever. ___ Red5 mailing list Red5@osflash.org http://osflash.org/mailman/listinfo/red5_osflash.org
Re: [Red5] onMetaData for vod
The one horrible thing about svn u cant revert the entire source onces its updated. Unfortunately in my case i have to keep up with svn after testing constantly of course, but it was to fix a problem with live streaming. I tried to run this svn merge -r1761:1758 /www/red5 but it seems to break things. Dan Rossi wrote: How do i fix this we're getting complaints coz of video sizing. Dan Rossi wrote: the update to 1761 killed onMetadata. the other wierd thing is if i comment out the write methods in MetaService its still writing a metakey file so its happening elsewhere. Steven Gong wrote: Joseph, Do you mean the it works on 1751? On 3/17/07, joseph wamicha [EMAIL PROTECTED] wrote: Looking at subversion logs; $svn log -r 1751:1761 -v http://svn1.cvsdude.com/osflash/red5/java/server/trunk I highly suspect this changes that added FME support are the cause: It talks about to correctly decode stream metadata. r1760 | jbauch | 2007-03-16 03:44:23 +0300 (Fri, 16 Mar 2007) | 5 lines Changed paths: M /doc/trunk/changelog.txt M /java/server/trunk/src/org/red5/io/amf/Output.java A /java/server/trunk/src/org/red5/io/object/ICustomSerializable.java A /java/server/trunk/src/org/red5/io/object/ISerializerOptionAware.java M /java/server/trunk/src/org/red5/io/object/Serializer.java M /java/server/trunk/src/org/red5/server/net/rtmp/RTMPHandler.java M /java/server/trunk/src/org/red5/server/net/rtmp/codec/RTMPProtocolDecoder.java M /java/server/trunk/src/org/red5/server/net/rtmp/status/Status.java M /java/server/trunk/src/org/red5/server/net/rtmp/status/StatusObject.java M /java/server/trunk/src/org/red5/server/stream/ClientBroadcastStream.java added interface to allow objects to serialize themselves added interface for objects that know about their serializer options send publish/record start events before the first packet is received correctly decode published stream metadata - Now Red5 supports being connected to from FME! r1761 | jbauch | 2007-03-16 04:05:26 +0300 (Fri, 16 Mar 2007) | 1 line Changed paths: M /java/server/trunk/src/org/red5/server/adapter/ApplicationAdapter.java M /java/server/trunk/src/org/red5/server/api/stream/IStreamService.java M /java/server/trunk/src/org/red5/server/net/rtmp/RTMPHandler.java M /java/server/trunk/src/org/red5/server/net/rtmp/message/Constants.java M /java/server/trunk/src/org/red5/server/stream/StreamService.java added FME specific methods On 3/17/07, Dan Rossi [EMAIL PROTECTED] wrote: Shit you are right its had to keep checking things to get something else to work. metadata is busted again. -- C is forever. ___ Red5 mailing list Red5@osflash.org http://osflash.org/mailman/listinfo/red5_osflash.org ___ Red5 mailing list Red5@osflash.org http://osflash.org/mailman/listinfo/red5_osflash.org ___ Red5 mailing list Red5@osflash.org http://osflash.org/mailman/listinfo/red5_osflash.org ___ Red5 mailing list Red5@osflash.org http://osflash.org/mailman/listinfo/red5_osflash.org ___ Red5 mailing list Red5@osflash.org http://osflash.org/mailman/listinfo/red5_osflash.org
Re: [Red5] onMetaData for vod
the last rev that works is 1755. To revert back simply remove the conf and src directories, choose the 1755 revision and choose svn checkout back into the current source. Worked for me nothing else will do it. There was a break from 1756 , which wouldnt compile anyway which was fixed in 1758. Dan Rossi wrote: The one horrible thing about svn u cant revert the entire source onces its updated. Unfortunately in my case i have to keep up with svn after testing constantly of course, but it was to fix a problem with live streaming. I tried to run this svn merge -r1761:1758 /www/red5 but it seems to break things. Dan Rossi wrote: How do i fix this we're getting complaints coz of video sizing. Dan Rossi wrote: the update to 1761 killed onMetadata. the other wierd thing is if i comment out the write methods in MetaService its still writing a metakey file so its happening elsewhere. Steven Gong wrote: Joseph, Do you mean the it works on 1751? On 3/17/07, joseph wamicha [EMAIL PROTECTED] wrote: Looking at subversion logs; $svn log -r 1751:1761 -v http://svn1.cvsdude.com/osflash/red5/java/server/trunk I highly suspect this changes that added FME support are the cause: It talks about to correctly decode stream metadata. r1760 | jbauch | 2007-03-16 03:44:23 +0300 (Fri, 16 Mar 2007) | 5 lines Changed paths: M /doc/trunk/changelog.txt M /java/server/trunk/src/org/red5/io/amf/Output.java A /java/server/trunk/src/org/red5/io/object/ICustomSerializable.java A /java/server/trunk/src/org/red5/io/object/ISerializerOptionAware.java M /java/server/trunk/src/org/red5/io/object/Serializer.java M /java/server/trunk/src/org/red5/server/net/rtmp/RTMPHandler.java M /java/server/trunk/src/org/red5/server/net/rtmp/codec/RTMPProtocolDecoder.java M /java/server/trunk/src/org/red5/server/net/rtmp/status/Status.java M /java/server/trunk/src/org/red5/server/net/rtmp/status/StatusObject.java M /java/server/trunk/src/org/red5/server/stream/ClientBroadcastStream.java added interface to allow objects to serialize themselves added interface for objects that know about their serializer options send publish/record start events before the first packet is received correctly decode published stream metadata - Now Red5 supports being connected to from FME! r1761 | jbauch | 2007-03-16 04:05:26 +0300 (Fri, 16 Mar 2007) | 1 line Changed paths: M /java/server/trunk/src/org/red5/server/adapter/ApplicationAdapter.java M /java/server/trunk/src/org/red5/server/api/stream/IStreamService.java M /java/server/trunk/src/org/red5/server/net/rtmp/RTMPHandler.java M /java/server/trunk/src/org/red5/server/net/rtmp/message/Constants.java M /java/server/trunk/src/org/red5/server/stream/StreamService.java added FME specific methods On 3/17/07, Dan Rossi [EMAIL PROTECTED] wrote: Shit you are right its had to keep checking things to get something else to work. metadata is busted again. -- C is forever. ___ Red5 mailing list Red5@osflash.org http://osflash.org/mailman/listinfo/red5_osflash.org ___ Red5 mailing list Red5@osflash.org http://osflash.org/mailman/listinfo/red5_osflash.org ___ Red5 mailing list Red5@osflash.org http://osflash.org/mailman/listinfo/red5_osflash.org ___ Red5 mailing list Red5@osflash.org http://osflash.org/mailman/listinfo/red5_osflash.org ___ Red5 mailing list Red5@osflash.org http://osflash.org/mailman/listinfo/red5_osflash.org ___ Red5 mailing list Red5@osflash.org http://osflash.org/mailman/listinfo/red5_osflash.org
Re: [Red5] onMetaData for vod
Its cool, and very painful , doing work on the weekends ! oseph wamicha wrote: Thanks alot Dan! On 3/17/07, Dan Rossi [EMAIL PROTECTED] wrote: the last rev that works is 1755. To revert back simply remove the conf and src directories, choose the 1755 revision and choose svn checkout back into the current source. Worked for me nothing else will do it. There was a break from 1756 , which wouldnt compile anyway which was fixed in 1758. Dan Rossi wrote: The one horrible thing about svn u cant revert the entire source onces its updated. Unfortunately in my case i have to keep up with svn after testing constantly of course, but it was to fix a problem with live streaming. I tried to run this svn merge -r1761:1758 /www/red5 but it seems to break things. Dan Rossi wrote: How do i fix this we're getting complaints coz of video sizing. Dan Rossi wrote: the update to 1761 killed onMetadata. the other wierd thing is if i comment out the write methods in MetaService its still writing a metakey file so its happening elsewhere. Steven Gong wrote: Joseph, Do you mean the it works on 1751? On 3/17/07, joseph wamicha [EMAIL PROTECTED] wrote: Looking at subversion logs; $svn log -r 1751:1761 -v http://svn1.cvsdude.com/osflash/red5/java/server/trunk I highly suspect this changes that added FME support are the cause: It talks about to correctly decode stream metadata. r1760 | jbauch | 2007-03-16 03:44:23 +0300 (Fri, 16 Mar 2007) | 5 lines Changed paths: M /doc/trunk/changelog.txt M /java/server/trunk/src/org/red5/io/amf/Output.java A /java/server/trunk/src/org/red5/io/object/ICustomSerializable.java A /java/server/trunk/src/org/red5/io/object/ISerializerOptionAware.java M /java/server/trunk/src/org/red5/io/object/Serializer.java M /java/server/trunk/src/org/red5/server/net/rtmp/RTMPHandler.java M /java/server/trunk/src/org/red5/server/net/rtmp/codec/RTMPProtocolDecoder.java M /java/server/trunk/src/org/red5/server/net/rtmp/status/Status.java M /java/server/trunk/src/org/red5/server/net/rtmp/status/StatusObject.java M /java/server/trunk/src/org/red5/server/stream/ClientBroadcastStream.java added interface to allow objects to serialize themselves added interface for objects that know about their serializer options send publish/record start events before the first packet is received correctly decode published stream metadata - Now Red5 supports being connected to from FME! r1761 | jbauch | 2007-03-16 04:05:26 +0300 (Fri, 16 Mar 2007) | 1 line Changed paths: M /java/server/trunk/src/org/red5/server/adapter/ApplicationAdapter.java M /java/server/trunk/src/org/red5/server/api/stream/IStreamService.java M /java/server/trunk/src/org/red5/server/net/rtmp/RTMPHandler.java M /java/server/trunk/src/org/red5/server/net/rtmp/message/Constants.java M /java/server/trunk/src/org/red5/server/stream/StreamService.java added FME specific methods On 3/17/07, Dan Rossi [EMAIL PROTECTED] wrote: Shit you are right its had to keep checking things to get something else to work. metadata is busted again. -- C is forever. ___ Red5 mailing list Red5@osflash.org http://osflash.org/mailman/listinfo/red5_osflash.org ___ Red5 mailing list Red5@osflash.org http://osflash.org/mailman/listinfo/red5_osflash.org ___ Red5 mailing list Red5@osflash.org http://osflash.org/mailman/listinfo/red5_osflash.org ___ Red5 mailing list Red5@osflash.org http://osflash.org/mailman/listinfo/red5_osflash.org ___ Red5 mailing list Red5@osflash.org http://osflash.org/mailman/listinfo/red5_osflash.org ___ Red5 mailing list Red5@osflash.org http://osflash.org/mailman/listinfo/red5_osflash.org ___ Red5 mailing list Red5@osflash.org http://osflash.org/mailman/listinfo/red5_osflash.org ___ Red5 mailing list Red5@osflash.org http://osflash.org/mailman/listinfo/red5_osflash.org
Re: [Red5] onMetaData for vod
Yap working weekends is a bore...but I love red5 at least makes it worth it ;) On 3/17/07, Dan Rossi [EMAIL PROTECTED] wrote: Its cool, and very painful , doing work on the weekends ! -- C is forever. ___ Red5 mailing list Red5@osflash.org http://osflash.org/mailman/listinfo/red5_osflash.org
Re: [Red5] onMetaData for vod
The diff file for the changes is attached. svn diff -r 1755:1761 http://svn1.cvsdude.com/osflash/red5/java/server/trunk FME.diff We could try to go through it to see what change could be causing the breakage. On 3/17/07, joseph wamicha [EMAIL PROTECTED] wrote: Yap working weekends is a bore...but I love red5 at least makes it worth it ;) On 3/17/07, Dan Rossi [EMAIL PROTECTED] wrote: Its cool, and very painful , doing work on the weekends ! -- C is forever. -- C is forever. Index: src/org/red5/server/adapter/ApplicationAdapter.java === --- src/org/red5/server/adapter/ApplicationAdapter.java (revision 1755) +++ src/org/red5/server/adapter/ApplicationAdapter.java (revision 1761) @@ -1025,4 +1025,21 @@ } } } + +/** + * Notification method that is sent by FME just before publishing starts. + * + * @param streamName Name of stream that is about to be published. + */ +public void FCPublish(String streamName) { + // Override if necessary. +} + +/** + * Notification method that is sent by FME when publishing of a stream ends. + */ +public void FCUnpublish() { + // Override if necessary. +} + } Index: src/org/red5/server/stream/ServerStream.java === --- src/org/red5/server/stream/ServerStream.java (revision 1755) +++ src/org/red5/server/stream/ServerStream.java (revision 1761) @@ -662,7 +662,12 @@ IMessage message; do { // Pull message from message input object... -message = msgIn.pullMessage(); + try { +message = msgIn.pullMessage(); + } catch (IOException err) { +log.error(Error while pulling message., err); +message = null; + } // If message is null then return null if (message == null) { return null; Index: src/org/red5/server/stream/PlaylistSubscriberStream.java === --- src/org/red5/server/stream/PlaylistSubscriberStream.java (revision 1755) +++ src/org/red5/server/stream/PlaylistSubscriberStream.java (revision 1761) @@ -19,6 +19,7 @@ * 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA */ +import java.io.IOException; import java.util.ArrayList; import java.util.HashMap; import java.util.List; @@ -160,7 +161,7 @@ } /** [EMAIL PROTECTED] */ -public void play() { +public void play() throws IOException { synchronized (items) { // Return if playlist is empty if (items.size() == 0) { @@ -298,6 +299,15 @@ try { engine.play(item); break; +} catch (IOException err) { + log.error(Error while starting to play item, moving to next., err); + // go for next item + moveToPrevious(); + if (currentItemIndex == -1) { + // we reaches the end. + break; + } + item = items.get(currentItemIndex); } catch (StreamNotFoundException e) { // go for next item moveToPrevious(); @@ -353,6 +363,15 @@ } } break; +} catch (IOException err) { + log.error(Error while starting to play item, moving to next., err); + // go for next item + moveToNext(); + if (currentItemIndex == -1) { + // we reaches the end. + break; + } + item = items.get(currentItemIndex); } catch (StreamNotFoundException e) { // go for next item moveToNext(); @@ -380,6 +399,8 @@ IPlayItem item = items.get(currentItemIndex); try { engine.play(item); + } catch (IOException e) { +log.error(setItem caught a IOException, e); } catch (StreamNotFoundException e) { // let the engine retain the STOPPED state // and wait for control from outside @@ -475,7 +496,11 @@ * @param message Message that has been written */ public void written(Object message) { - engine.pullAndPush(); + try { + engine.pullAndPush(); + } catch (Throwable err) { + log.error(Error while pulling message., err); + } } /** @@ -726,9 +751,10 @@ * @param item Playlist item * @throws StreamNotFoundException Stream not found * @throws IllegalStateException Stream is in stopped state + * @throws IOException */ public synchronized void play(IPlayItem item) -throws StreamNotFoundException, IllegalStateException { +throws StreamNotFoundException, IllegalStateException, IOException { // Can't play if state is stopped if (state != State.STOPPED) { throw new IllegalStateException(); @@ -858,11 +884,6 @@ sendStreamNotFoundStatus(currentItem); throw new StreamNotFoundException(item.getName()); } - if (sendNotifications) { -sendReset(); -sendResetStatus(item); -sendStartStatus(item); - } state = State.PLAYING; if (decision == 1)
Re: [Red5] onMetaData for vod
Its getting there the latency buffering issue is making me lose sleep though lol. The live thing isnt suffering the latency apart from the error causing disconnections and we can only get 1frame per second playback, the biggest issue is vod right now ;) The odd thing is wowza plays back vod fine but live was unstreamable. Whereas red5 was doable with live streaming but vod playback is unstreamable, both on the same machine too :D joseph wamicha wrote: Yap working weekends is a bore...but I love red5 at least makes it worth it ;) On 3/17/07, Dan Rossi [EMAIL PROTECTED] wrote: Its cool, and very painful , doing work on the weekends ! ___ Red5 mailing list Red5@osflash.org http://osflash.org/mailman/listinfo/red5_osflash.org ___ Red5 mailing list Red5@osflash.org http://osflash.org/mailman/listinfo/red5_osflash.org
Re: [Red5] onMetaData for vod
Ok that makes it easier :) joseph wamicha wrote: The diff file for the changes is attached. svn diff -r 1755:1761 http://svn1.cvsdude.com/osflash/red5/java/server/trunk FME.diff We could try to go through it to see what change could be causing the breakage. On 3/17/07, joseph wamicha [EMAIL PROTECTED] wrote: Yap working weekends is a bore...but I love red5 at least makes it worth it ;) On 3/17/07, Dan Rossi [EMAIL PROTECTED] wrote: Its cool, and very painful , doing work on the weekends ! -- C is forever. Index: src/org/red5/server/adapter/ApplicationAdapter.java === --- src/org/red5/server/adapter/ApplicationAdapter.java (revision 1755) +++ src/org/red5/server/adapter/ApplicationAdapter.java (revision 1761) @@ -1025,4 +1025,21 @@ } } } + +/** + * Notification method that is sent by FME just before publishing starts. + * + * @param streamName Name of stream that is about to be published. + */ +public void FCPublish(String streamName) { + // Override if necessary. +} + +/** + * Notification method that is sent by FME when publishing of a stream ends. + */ +public void FCUnpublish() { + // Override if necessary. +} + } Index: src/org/red5/server/stream/ServerStream.java === --- src/org/red5/server/stream/ServerStream.java (revision 1755) +++ src/org/red5/server/stream/ServerStream.java (revision 1761) @@ -662,7 +662,12 @@ IMessage message; do { // Pull message from message input object... -message = msgIn.pullMessage(); + try { + message = msgIn.pullMessage(); + } catch (IOException err) { + log.error(Error while pulling message., err); + message = null; + } // If message is null then return null if (message == null) { return null; Index: src/org/red5/server/stream/PlaylistSubscriberStream.java === --- src/org/red5/server/stream/PlaylistSubscriberStream.java (revision 1755) +++ src/org/red5/server/stream/PlaylistSubscriberStream.java (revision 1761) @@ -19,6 +19,7 @@ * 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA */ +import java.io.IOException; import java.util.ArrayList; import java.util.HashMap; import java.util.List; @@ -160,7 +161,7 @@ } /** [EMAIL PROTECTED] */ -public void play() { +public void play() throws IOException { synchronized (items) { // Return if playlist is empty if (items.size() == 0) { @@ -298,6 +299,15 @@ try { engine.play(item); break; + } catch (IOException err) { + log.error(Error while starting to play item, moving to next., err); + // go for next item + moveToPrevious(); + if (currentItemIndex == -1) { + // we reaches the end. + break; + } + item = items.get(currentItemIndex); } catch (StreamNotFoundException e) { // go for next item moveToPrevious(); @@ -353,6 +363,15 @@ } } break; + } catch (IOException err) { + log.error(Error while starting to play item, moving to next., err); + // go for next item + moveToNext(); + if (currentItemIndex == -1) { + // we reaches the end. + break; + } + item = items.get(currentItemIndex); } catch (StreamNotFoundException e) { // go for next item moveToNext(); @@ -380,6 +399,8 @@ IPlayItem item
Re: [Red5] onMetaData for vod
Glancing at the diff file, I don't see anything that should be breaking onMetaData (but then again I'm still learning red5 so could be wrong). In which part of red5 is onMetaData triggered? On 3/17/07, Dan Rossi [EMAIL PROTECTED] wrote: Its getting there the latency buffering issue is making me lose sleep though lol. The live thing isnt suffering the latency apart from the error causing disconnections and we can only get 1frame per second playback, the biggest issue is vod right now ;) The odd thing is wowza plays back vod fine but live was unstreamable. Whereas red5 was doable with live streaming but vod playback is unstreamable, both on the same machine too :D joseph wamicha wrote: Yap working weekends is a bore...but I love red5 at least makes it worth it ;) On 3/17/07, Dan Rossi [EMAIL PROTECTED] wrote: Its cool, and very painful , doing work on the weekends ! ___ Red5 mailing list Red5@osflash.org http://osflash.org/mailman/listinfo/red5_osflash.org ___ Red5 mailing list Red5@osflash.org http://osflash.org/mailman/listinfo/red5_osflash.org -- C is forever. ___ Red5 mailing list Red5@osflash.org http://osflash.org/mailman/listinfo/red5_osflash.org
Re: [Red5] onMetaData for vod
I think it's src/org/red5/server/stream/PlaylistSubscriberStream.java; almost there... On 3/17/07, joseph wamicha [EMAIL PROTECTED] wrote: Glancing at the diff file, I don't see anything that should be breaking onMetaData (but then again I'm still learning red5 so could be wrong). In which part of red5 is onMetaData triggered? On 3/17/07, Dan Rossi [EMAIL PROTECTED] wrote: Its getting there the latency buffering issue is making me lose sleep though lol. The live thing isnt suffering the latency apart from the error causing disconnections and we can only get 1frame per second playback, the biggest issue is vod right now ;) The odd thing is wowza plays back vod fine but live was unstreamable. Whereas red5 was doable with live streaming but vod playback is unstreamable, both on the same machine too :D joseph wamicha wrote: Yap working weekends is a bore...but I love red5 at least makes it worth it ;) On 3/17/07, Dan Rossi [EMAIL PROTECTED] wrote: Its cool, and very painful , doing work on the weekends ! ___ Red5 mailing list Red5@osflash.org http://osflash.org/mailman/listinfo/red5_osflash.org ___ Red5 mailing list Red5@osflash.org http://osflash.org/mailman/listinfo/red5_osflash.org -- C is forever. -- C is forever. ___ Red5 mailing list Red5@osflash.org http://osflash.org/mailman/listinfo/red5_osflash.org
Re: [Red5] onMetaData for vod
Thats been throwing exceptions i think. joseph wamicha wrote: I think it's src/org/red5/server/stream/PlaylistSubscriberStream.java; almost there... On 3/17/07, joseph wamicha [EMAIL PROTECTED] wrote: Glancing at the diff file, I don't see anything that should be breaking onMetaData (but then again I'm still learning red5 so could be wrong). In which part of red5 is onMetaData triggered? On 3/17/07, Dan Rossi [EMAIL PROTECTED] wrote: Its getting there the latency buffering issue is making me lose sleep though lol. The live thing isnt suffering the latency apart from the error causing disconnections and we can only get 1frame per second playback, the biggest issue is vod right now ;) The odd thing is wowza plays back vod fine but live was unstreamable. Whereas red5 was doable with live streaming but vod playback is unstreamable, both on the same machine too :D joseph wamicha wrote: Yap working weekends is a bore...but I love red5 at least makes it worth it ;) On 3/17/07, Dan Rossi [EMAIL PROTECTED] wrote: Its cool, and very painful , doing work on the weekends ! ___ Red5 mailing list Red5@osflash.org http://osflash.org/mailman/listinfo/red5_osflash.org ___ Red5 mailing list Red5@osflash.org http://osflash.org/mailman/listinfo/red5_osflash.org -- C is forever. ___ Red5 mailing list Red5@osflash.org http://osflash.org/mailman/listinfo/red5_osflash.org ___ Red5 mailing list Red5@osflash.org http://osflash.org/mailman/listinfo/red5_osflash.org
Re: [Red5] onMetaData for vod
In PlaylistSubscriberStream.java, I suspect it is this method that triggers onMetaData to be called. private void sendVODInitCM(IMessageInput msgIn, IPlayItem item) { OOBControlMessage oobCtrlMsg = new OOBControlMessage(); oobCtrlMsg.setTarget(IPassive.KEY); oobCtrlMsg.setServiceName(init); MapObject, Object paramMap = new HashMapObject, Object(); paramMap.put(startTS, (int) item.getStart()); oobCtrlMsg.setServiceParamMap(paramMap); msgIn.sendOOBControlMessage(this, oobCtrlMsg); } It seems sendOOBControlMessage not working? On 3/17/07, joseph wamicha [EMAIL PROTECTED] wrote: I think it's src/org/red5/server/stream/PlaylistSubscriberStream.java; almost there... On 3/17/07, joseph wamicha [EMAIL PROTECTED] wrote: Glancing at the diff file, I don't see anything that should be breaking onMetaData (but then again I'm still learning red5 so could be wrong). In which part of red5 is onMetaData triggered? On 3/17/07, Dan Rossi [EMAIL PROTECTED] wrote: Its getting there the latency buffering issue is making me lose sleep though lol. The live thing isnt suffering the latency apart from the error causing disconnections and we can only get 1frame per second playback, the biggest issue is vod right now ;) The odd thing is wowza plays back vod fine but live was unstreamable. Whereas red5 was doable with live streaming but vod playback is unstreamable, both on the same machine too :D joseph wamicha wrote: Yap working weekends is a bore...but I love red5 at least makes it worth it ;) On 3/17/07, Dan Rossi [EMAIL PROTECTED] wrote: Its cool, and very painful , doing work on the weekends ! ___ Red5 mailing list Red5@osflash.org http://osflash.org/mailman/listinfo/red5_osflash.org ___ Red5 mailing list Red5@osflash.org http://osflash.org/mailman/listinfo/red5_osflash.org -- C is forever. -- C is forever. -- C is forever. ___ Red5 mailing list Red5@osflash.org http://osflash.org/mailman/listinfo/red5_osflash.org
Re: [Red5] onMetaData for vod
PlaylistSubscriberStream seems ok. I can't figure it out; it seems like something about the RTMP control messages is not working. On 3/17/07, Dan Rossi [EMAIL PROTECTED] wrote: Thats been throwing exceptions i think. joseph wamicha wrote: I think it's src/org/red5/server/stream/PlaylistSubscriberStream.java; almost there... On 3/17/07, joseph wamicha [EMAIL PROTECTED] wrote: Glancing at the diff file, I don't see anything that should be breaking onMetaData (but then again I'm still learning red5 so could be wrong). In which part of red5 is onMetaData triggered? On 3/17/07, Dan Rossi [EMAIL PROTECTED] wrote: Its getting there the latency buffering issue is making me lose sleep though lol. The live thing isnt suffering the latency apart from the error causing disconnections and we can only get 1frame per second playback, the biggest issue is vod right now ;) The odd thing is wowza plays back vod fine but live was unstreamable. Whereas red5 was doable with live streaming but vod playback is unstreamable, both on the same machine too :D joseph wamicha wrote: Yap working weekends is a bore...but I love red5 at least makes it worth it ;) On 3/17/07, Dan Rossi [EMAIL PROTECTED] wrote: Its cool, and very painful , doing work on the weekends ! ___ Red5 mailing list Red5@osflash.org http://osflash.org/mailman/listinfo/red5_osflash.org ___ Red5 mailing list Red5@osflash.org http://osflash.org/mailman/listinfo/red5_osflash.org -- C is forever. ___ Red5 mailing list Red5@osflash.org http://osflash.org/mailman/listinfo/red5_osflash.org ___ Red5 mailing list Red5@osflash.org http://osflash.org/mailman/listinfo/red5_osflash.org -- C is forever. ___ Red5 mailing list Red5@osflash.org http://osflash.org/mailman/listinfo/red5_osflash.org
Re: [Red5] onMetaData for vod
play(stream, 0); the other flags i was using was default so were not neccesary. the metadata is called before the first frame. joseph wamicha wrote: Hello, Problem: In revision 1761, the onMetaData method isn't being triggered or called. Thus when streaming vod currently, I can't seek and the video.heightand video.width dimensions can not be set using metadata information in onMetaData function. I seem to notice a pattern with red5. The first time a vod item is played, a .meta file is generated which contains keyframes and their positions and timestamps. I think it is this file that is used when you do a seek from Flash End. Right now seek does not seem to be working when using John's flex2 player even with this .meta file in place. Is anyone else experiencing the same problems up of revision 1761 for onMetaData? I'd be grateful for any help. ___ Red5 mailing list Red5@osflash.org http://osflash.org/mailman/listinfo/red5_osflash.org ___ Red5 mailing list Red5@osflash.org http://osflash.org/mailman/listinfo/red5_osflash.org
Re: [Red5] onMetaData for vod
I don't know what has changed since revision 1761 but even this option does not trigger onMetaData function. How is onMetaData function triggered from red5? It seems to have broken. I have not changed anything since updating to this revision. On 3/17/07, Dan Rossi [EMAIL PROTECTED] wrote: play(stream, 0); the other flags i was using was default so were not neccesary. the metadata is called before the first frame. joseph wamicha wrote: Hello, Problem: In revision 1761, the onMetaData method isn't being triggered or called. Thus when streaming vod currently, I can't seek and the video.heightand video.width dimensions can not be set using metadata information in onMetaData function. I seem to notice a pattern with red5. The first time a vod item is played, a .meta file is generated which contains keyframes and their positions and timestamps. I think it is this file that is used when you do a seek from Flash End. Right now seek does not seem to be working when using John's flex2 player even with this .meta file in place. Is anyone else experiencing the same problems up of revision 1761 for onMetaData? I'd be grateful for any help. ___ Red5 mailing list Red5@osflash.org http://osflash.org/mailman/listinfo/red5_osflash.org ___ Red5 mailing list Red5@osflash.org http://osflash.org/mailman/listinfo/red5_osflash.org -- C is forever. ___ Red5 mailing list Red5@osflash.org http://osflash.org/mailman/listinfo/red5_osflash.org
Re: [Red5] onMetaData for vod
Shit you are right its had to keep checking things to get something else to work. metadata is busted again. joseph wamicha wrote: I don't know what has changed since revision 1761 but even this option does not trigger onMetaData function. How is onMetaData function triggered from red5? It seems to have broken. I have not changed anything since updating to this revision. On 3/17/07, Dan Rossi [EMAIL PROTECTED] wrote: play(stream, 0); the other flags i was using was default so were not neccesary. the metadata is called before the first frame. joseph wamicha wrote: Hello, Problem: In revision 1761, the onMetaData method isn't being triggered or called. Thus when streaming vod currently, I can't seek and the video.heightand video.width dimensions can not be set using metadata information in onMetaData function. I seem to notice a pattern with red5. The first time a vod item is played, a .meta file is generated which contains keyframes and their positions and timestamps. I think it is this file that is used when you do a seek from Flash End. Right now seek does not seem to be working when using John's flex2 player even with this .meta file in place. Is anyone else experiencing the same problems up of revision 1761 for onMetaData? I'd be grateful for any help. ___ Red5 mailing list Red5@osflash.org http://osflash.org/mailman/listinfo/red5_osflash.org ___ Red5 mailing list Red5@osflash.org http://osflash.org/mailman/listinfo/red5_osflash.org ___ Red5 mailing list Red5@osflash.org http://osflash.org/mailman/listinfo/red5_osflash.org ___ Red5 mailing list Red5@osflash.org http://osflash.org/mailman/listinfo/red5_osflash.org
Re: [Red5] onMetaData for vod
maybe it has something to do with the ticket that it keeps trying to write the metadata file on each request even though its there ? Im going to turn the damn thing off and see if it works. Dan Rossi wrote: Shit you are right its had to keep checking things to get something else to work. metadata is busted again. joseph wamicha wrote: I don't know what has changed since revision 1761 but even this option does not trigger onMetaData function. How is onMetaData function triggered from red5? It seems to have broken. I have not changed anything since updating to this revision. On 3/17/07, Dan Rossi [EMAIL PROTECTED] wrote: play(stream, 0); the other flags i was using was default so were not neccesary. the metadata is called before the first frame. joseph wamicha wrote: Hello, Problem: In revision 1761, the onMetaData method isn't being triggered or called. Thus when streaming vod currently, I can't seek and the video.heightand video.width dimensions can not be set using metadata information in onMetaData function. I seem to notice a pattern with red5. The first time a vod item is played, a .meta file is generated which contains keyframes and their positions and timestamps. I think it is this file that is used when you do a seek from Flash End. Right now seek does not seem to be working when using John's flex2 player even with this .meta file in place. Is anyone else experiencing the same problems up of revision 1761 for onMetaData? I'd be grateful for any help. ___ Red5 mailing list Red5@osflash.org http://osflash.org/mailman/listinfo/red5_osflash.org ___ Red5 mailing list Red5@osflash.org http://osflash.org/mailman/listinfo/red5_osflash.org ___ Red5 mailing list Red5@osflash.org http://osflash.org/mailman/listinfo/red5_osflash.org ___ Red5 mailing list Red5@osflash.org http://osflash.org/mailman/listinfo/red5_osflash.org ___ Red5 mailing list Red5@osflash.org http://osflash.org/mailman/listinfo/red5_osflash.org
Re: [Red5] onMetaData for vod
Looking at subversion logs; $svn log -r 1751:1761 -v http://svn1.cvsdude.com/osflash/red5/java/server/trunk I highly suspect this changes that added FME support are the cause: It talks about to correctly decode stream metadata. r1760 | jbauch | 2007-03-16 03:44:23 +0300 (Fri, 16 Mar 2007) | 5 lines Changed paths: M /doc/trunk/changelog.txt M /java/server/trunk/src/org/red5/io/amf/Output.java A /java/server/trunk/src/org/red5/io/object/ICustomSerializable.java A /java/server/trunk/src/org/red5/io/object/ISerializerOptionAware.java M /java/server/trunk/src/org/red5/io/object/Serializer.java M /java/server/trunk/src/org/red5/server/net/rtmp/RTMPHandler.java M /java/server/trunk/src/org/red5/server/net/rtmp/codec/RTMPProtocolDecoder.java M /java/server/trunk/src/org/red5/server/net/rtmp/status/Status.java M /java/server/trunk/src/org/red5/server/net/rtmp/status/StatusObject.java M /java/server/trunk/src/org/red5/server/stream/ClientBroadcastStream.java added interface to allow objects to serialize themselves added interface for objects that know about their serializer options send publish/record start events before the first packet is received correctly decode published stream metadata - Now Red5 supports being connected to from FME! r1761 | jbauch | 2007-03-16 04:05:26 +0300 (Fri, 16 Mar 2007) | 1 line Changed paths: M /java/server/trunk/src/org/red5/server/adapter/ApplicationAdapter.java M /java/server/trunk/src/org/red5/server/api/stream/IStreamService.java M /java/server/trunk/src/org/red5/server/net/rtmp/RTMPHandler.java M /java/server/trunk/src/org/red5/server/net/rtmp/message/Constants.java M /java/server/trunk/src/org/red5/server/stream/StreamService.java added FME specific methods On 3/17/07, Dan Rossi [EMAIL PROTECTED] wrote: Shit you are right its had to keep checking things to get something else to work. metadata is busted again. -- C is forever. ___ Red5 mailing list Red5@osflash.org http://osflash.org/mailman/listinfo/red5_osflash.org
Re: [Red5] onMetaData for vod
the metadata cache isnt the problem its something else. Dan Rossi wrote: maybe it has something to do with the ticket that it keeps trying to write the metadata file on each request even though its there ? Im going to turn the damn thing off and see if it works. Dan Rossi wrote: Shit you are right its had to keep checking things to get something else to work. metadata is busted again. joseph wamicha wrote: I don't know what has changed since revision 1761 but even this option does not trigger onMetaData function. How is onMetaData function triggered from red5? It seems to have broken. I have not changed anything since updating to this revision. On 3/17/07, Dan Rossi [EMAIL PROTECTED] wrote: play(stream, 0); the other flags i was using was default so were not neccesary. the metadata is called before the first frame. joseph wamicha wrote: Hello, Problem: In revision 1761, the onMetaData method isn't being triggered or called. Thus when streaming vod currently, I can't seek and the video.heightand video.width dimensions can not be set using metadata information in onMetaData function. I seem to notice a pattern with red5. The first time a vod item is played, a .meta file is generated which contains keyframes and their positions and timestamps. I think it is this file that is used when you do a seek from Flash End. Right now seek does not seem to be working when using John's flex2 player even with this .meta file in place. Is anyone else experiencing the same problems up of revision 1761 for onMetaData? I'd be grateful for any help. ___ Red5 mailing list Red5@osflash.org http://osflash.org/mailman/listinfo/red5_osflash.org ___ Red5 mailing list Red5@osflash.org http://osflash.org/mailman/listinfo/red5_osflash.org ___ Red5 mailing list Red5@osflash.org http://osflash.org/mailman/listinfo/red5_osflash.org ___ Red5 mailing list Red5@osflash.org http://osflash.org/mailman/listinfo/red5_osflash.org ___ Red5 mailing list Red5@osflash.org http://osflash.org/mailman/listinfo/red5_osflash.org ___ Red5 mailing list Red5@osflash.org http://osflash.org/mailman/listinfo/red5_osflash.org
Re: [Red5] onMetaData for vod
what do you think it could be that cause this. On 3/17/07, Dan Rossi [EMAIL PROTECTED] wrote: the metadata cache isnt the problem its something else. -- C is forever. ___ Red5 mailing list Red5@osflash.org http://osflash.org/mailman/listinfo/red5_osflash.org
Re: [Red5] onMetaData for vod
Damn my apologies for the suggestion to support FME ? If it cant be supported, undoing this would be good and then i guess an open source project to build an equivalanet in flex or something, because the licencing required to purchase the C++ API ! joseph wamicha wrote: Looking at subversion logs; $svn log -r 1751:1761 -v http://svn1.cvsdude.com/osflash/red5/java/server/trunk I highly suspect this changes that added FME support are the cause: It talks about to correctly decode stream metadata. r1760 | jbauch | 2007-03-16 03:44:23 +0300 (Fri, 16 Mar 2007) | 5 lines Changed paths: M /doc/trunk/changelog.txt M /java/server/trunk/src/org/red5/io/amf/Output.java A /java/server/trunk/src/org/red5/io/object/ICustomSerializable.java A /java/server/trunk/src/org/red5/io/object/ISerializerOptionAware.java M /java/server/trunk/src/org/red5/io/object/Serializer.java M /java/server/trunk/src/org/red5/server/net/rtmp/RTMPHandler.java M /java/server/trunk/src/org/red5/server/net/rtmp/codec/RTMPProtocolDecoder.java M /java/server/trunk/src/org/red5/server/net/rtmp/status/Status.java M /java/server/trunk/src/org/red5/server/net/rtmp/status/StatusObject.java M /java/server/trunk/src/org/red5/server/stream/ClientBroadcastStream.java added interface to allow objects to serialize themselves added interface for objects that know about their serializer options send publish/record start events before the first packet is received correctly decode published stream metadata - Now Red5 supports being connected to from FME! r1761 | jbauch | 2007-03-16 04:05:26 +0300 (Fri, 16 Mar 2007) | 1 line Changed paths: M /java/server/trunk/src/org/red5/server/adapter/ApplicationAdapter.java M /java/server/trunk/src/org/red5/server/api/stream/IStreamService.java M /java/server/trunk/src/org/red5/server/net/rtmp/RTMPHandler.java M /java/server/trunk/src/org/red5/server/net/rtmp/message/Constants.java M /java/server/trunk/src/org/red5/server/stream/StreamService.java added FME specific methods On 3/17/07, Dan Rossi [EMAIL PROTECTED] wrote: Shit you are right its had to keep checking things to get something else to work. metadata is busted again. ___ Red5 mailing list Red5@osflash.org http://osflash.org/mailman/listinfo/red5_osflash.org ___ Red5 mailing list Red5@osflash.org http://osflash.org/mailman/listinfo/red5_osflash.org
Re: [Red5] onMetaData for vod
Joseph, Do you mean the it works on 1751? On 3/17/07, joseph wamicha [EMAIL PROTECTED] wrote: Looking at subversion logs; $svn log -r 1751:1761 -v http://svn1.cvsdude.com/osflash/red5/java/server/trunk I highly suspect this changes that added FME support are the cause: It talks about to correctly decode stream metadata. r1760 | jbauch | 2007-03-16 03:44:23 +0300 (Fri, 16 Mar 2007) | 5 lines Changed paths: M /doc/trunk/changelog.txt M /java/server/trunk/src/org/red5/io/amf/Output.java A /java/server/trunk/src/org/red5/io/object/ICustomSerializable.java A /java/server/trunk/src/org/red5/io/object/ISerializerOptionAware.java M /java/server/trunk/src/org/red5/io/object/Serializer.java M /java/server/trunk/src/org/red5/server/net/rtmp/RTMPHandler.java M /java/server/trunk/src/org/red5/server/net/rtmp/codec/RTMPProtocolDecoder.java M /java/server/trunk/src/org/red5/server/net/rtmp/status/Status.java M /java/server/trunk/src/org/red5/server/net/rtmp/status/StatusObject.java M /java/server/trunk/src/org/red5/server/stream/ClientBroadcastStream.java added interface to allow objects to serialize themselves added interface for objects that know about their serializer options send publish/record start events before the first packet is received correctly decode published stream metadata - Now Red5 supports being connected to from FME! r1761 | jbauch | 2007-03-16 04:05:26 +0300 (Fri, 16 Mar 2007) | 1 line Changed paths: M /java/server/trunk/src/org/red5/server/adapter/ApplicationAdapter.java M /java/server/trunk/src/org/red5/server/api/stream/IStreamService.java M /java/server/trunk/src/org/red5/server/net/rtmp/RTMPHandler.java M /java/server/trunk/src/org/red5/server/net/rtmp/message/Constants.java M /java/server/trunk/src/org/red5/server/stream/StreamService.java added FME specific methods On 3/17/07, Dan Rossi [EMAIL PROTECTED] wrote: Shit you are right its had to keep checking things to get something else to work. metadata is busted again. -- C is forever. ___ Red5 mailing list Red5@osflash.org http://osflash.org/mailman/listinfo/red5_osflash.org -- I cannot tell why this heart languishes in silence. It is for small needs it never asks, or knows or remembers. -- Tagore Best Regards Steven Gong ___ Red5 mailing list Red5@osflash.org http://osflash.org/mailman/listinfo/red5_osflash.org
Re: [Red5] onMetaData for vod
the update to 1761 killed onMetadata. the other wierd thing is if i comment out the write methods in MetaService its still writing a metakey file so its happening elsewhere. Steven Gong wrote: Joseph, Do you mean the it works on 1751? On 3/17/07, joseph wamicha [EMAIL PROTECTED] wrote: Looking at subversion logs; $svn log -r 1751:1761 -v http://svn1.cvsdude.com/osflash/red5/java/server/trunk I highly suspect this changes that added FME support are the cause: It talks about to correctly decode stream metadata. r1760 | jbauch | 2007-03-16 03:44:23 +0300 (Fri, 16 Mar 2007) | 5 lines Changed paths: M /doc/trunk/changelog.txt M /java/server/trunk/src/org/red5/io/amf/Output.java A /java/server/trunk/src/org/red5/io/object/ICustomSerializable.java A /java/server/trunk/src/org/red5/io/object/ISerializerOptionAware.java M /java/server/trunk/src/org/red5/io/object/Serializer.java M /java/server/trunk/src/org/red5/server/net/rtmp/RTMPHandler.java M /java/server/trunk/src/org/red5/server/net/rtmp/codec/RTMPProtocolDecoder.java M /java/server/trunk/src/org/red5/server/net/rtmp/status/Status.java M /java/server/trunk/src/org/red5/server/net/rtmp/status/StatusObject.java M /java/server/trunk/src/org/red5/server/stream/ClientBroadcastStream.java added interface to allow objects to serialize themselves added interface for objects that know about their serializer options send publish/record start events before the first packet is received correctly decode published stream metadata - Now Red5 supports being connected to from FME! r1761 | jbauch | 2007-03-16 04:05:26 +0300 (Fri, 16 Mar 2007) | 1 line Changed paths: M /java/server/trunk/src/org/red5/server/adapter/ApplicationAdapter.java M /java/server/trunk/src/org/red5/server/api/stream/IStreamService.java M /java/server/trunk/src/org/red5/server/net/rtmp/RTMPHandler.java M /java/server/trunk/src/org/red5/server/net/rtmp/message/Constants.java M /java/server/trunk/src/org/red5/server/stream/StreamService.java added FME specific methods On 3/17/07, Dan Rossi [EMAIL PROTECTED] wrote: Shit you are right its had to keep checking things to get something else to work. metadata is busted again. -- C is forever. ___ Red5 mailing list Red5@osflash.org http://osflash.org/mailman/listinfo/red5_osflash.org ___ Red5 mailing list Red5@osflash.org http://osflash.org/mailman/listinfo/red5_osflash.org ___ Red5 mailing list Red5@osflash.org http://osflash.org/mailman/listinfo/red5_osflash.org
Re: [Red5] onMetaData for vod
How do i fix this we're getting complaints coz of video sizing. Dan Rossi wrote: the update to 1761 killed onMetadata. the other wierd thing is if i comment out the write methods in MetaService its still writing a metakey file so its happening elsewhere. Steven Gong wrote: Joseph, Do you mean the it works on 1751? On 3/17/07, joseph wamicha [EMAIL PROTECTED] wrote: Looking at subversion logs; $svn log -r 1751:1761 -v http://svn1.cvsdude.com/osflash/red5/java/server/trunk I highly suspect this changes that added FME support are the cause: It talks about to correctly decode stream metadata. r1760 | jbauch | 2007-03-16 03:44:23 +0300 (Fri, 16 Mar 2007) | 5 lines Changed paths: M /doc/trunk/changelog.txt M /java/server/trunk/src/org/red5/io/amf/Output.java A /java/server/trunk/src/org/red5/io/object/ICustomSerializable.java A /java/server/trunk/src/org/red5/io/object/ISerializerOptionAware.java M /java/server/trunk/src/org/red5/io/object/Serializer.java M /java/server/trunk/src/org/red5/server/net/rtmp/RTMPHandler.java M /java/server/trunk/src/org/red5/server/net/rtmp/codec/RTMPProtocolDecoder.java M /java/server/trunk/src/org/red5/server/net/rtmp/status/Status.java M /java/server/trunk/src/org/red5/server/net/rtmp/status/StatusObject.java M /java/server/trunk/src/org/red5/server/stream/ClientBroadcastStream.java added interface to allow objects to serialize themselves added interface for objects that know about their serializer options send publish/record start events before the first packet is received correctly decode published stream metadata - Now Red5 supports being connected to from FME! r1761 | jbauch | 2007-03-16 04:05:26 +0300 (Fri, 16 Mar 2007) | 1 line Changed paths: M /java/server/trunk/src/org/red5/server/adapter/ApplicationAdapter.java M /java/server/trunk/src/org/red5/server/api/stream/IStreamService.java M /java/server/trunk/src/org/red5/server/net/rtmp/RTMPHandler.java M /java/server/trunk/src/org/red5/server/net/rtmp/message/Constants.java M /java/server/trunk/src/org/red5/server/stream/StreamService.java added FME specific methods On 3/17/07, Dan Rossi [EMAIL PROTECTED] wrote: Shit you are right its had to keep checking things to get something else to work. metadata is busted again. -- C is forever. ___ Red5 mailing list Red5@osflash.org http://osflash.org/mailman/listinfo/red5_osflash.org ___ Red5 mailing list Red5@osflash.org http://osflash.org/mailman/listinfo/red5_osflash.org ___ Red5 mailing list Red5@osflash.org http://osflash.org/mailman/listinfo/red5_osflash.org ___ Red5 mailing list Red5@osflash.org http://osflash.org/mailman/listinfo/red5_osflash.org