Recherche avancée

Sur d’autres sites (9339)

• configure : fix CoreGraphics module name

  13 octobre 2017, par James Almer

  configure : fix CoreGraphics module name
  Signed-off-by : James Almer <jamrial@gmail.com>
  

  • [DH] configure

• Nginx rtmp module - on_publish fires multiple time instead of once

  29 juillet 2017, par Stephen Wright

  This is copy and pasted from the bug report I created on the rtmp-module by Arut, I am not completely sure if it is a bug or me not understanding how the module works, I have read the whole directives of module as from https://github.com/arut/nginx-rtmp-module/wiki/Directives

  Proper explanation, if code is not displayed properly I will edit and fix

  Hi, been using the module and finding it very very good !

  Think I have found a issue though, although it may be me misunderstanding the directives.

  Essentially I wish to fire a script (/usr/local/bin/make_thumbnail.sh) which creates a thumbnail automatically from a stream (using ffmpeg), the idea is to have this done for every stream as soon as it is published in order to create a function a bit like twitch tv where the streamer will not have to specify any thumbnail image, authenticated users simply start a stream (which will later be authenticated but is not yet) The script does also write data into the database however this stage works fine and I don’t believe the issue is related, if I comment out these lines then the thumbnail creation still works and my issue continues.

  Initially this was done using the "exec" command as I believe I mis-read the documentation and I believe the exec command doesn’t work for my problem as ". When publishing stops the process is terminated." does this mean it will continually execute until stream stops ?

  I have started using the exec_publish command to try and fix this issue however the same issue seems to occur. The entire script repeats approximately every 15-17 seconds, a new thumbnail is created and a new database entry is create with all the correct information.

  Below is the nginx.conf line. Please ignore if indentation is incorrect couldn’t see a way to indent blocks of code and it’s late here, assume all code is indented correctly unless you believe that could be the issue in which case I will post it indented as early as I can.

  application live {
  
  allow play all;
  
  live on;
  
  record all;
  
  record_path /var/stream/video_recordings/;
  
  record_unique on; 
  
  hls on;
  
  hls_nested on;
  
  hls_path /var/stream/HLS/live;
  
  hls_fragment 10s;
  
  
  
  #on publish create thumbnail using first second of stream and save in 
  
  /var/stream/video_recordings/thumbnails
  
  exec_publish usr/local/bin/make_thumbnail.sh $name;

  The rest can be pasted or attached if needed but is working nginx config for rtmp + website

  The most simple version of the make_thumbnail..sh is pasted below, I have omitted the variables that I have used for database entryys obviously but as the script works without fail from terminal I believe this to be an nginx issue (if I run the command manually under the nginx user e.g. sudo -u nginx /usr/local/bin/make_thumbnail.sh with a name the same as any running stream, it works and only executes once as would expect, all permissions in script are ok and tested.

  make_thumbnail.sh

  #!/bin/bash
  
  
  
  TIME=$(date +%s)
  
  NAME=$1 
  
  echo "time: "
  
  FILENAME=${TIME}_${NAME}
  
  
  
  ffmpeg -i rtmp://192.168.0.98:1935/live/$1 -vframes 1 -s 150x150 -ss 10 -
  
  strftime 1 /var/stream/video_recordings/thumbnails/"$FILENAME.jpg";
  
  
  
  #Writes path to video into database
  
  mysql --user=$DB_USER --password=$DB_PASSWD $DB_NAME << EOF
  
  INSERT INTO $TABLE3 (thumbnailfile) VALUES ('$FILENAME');
  
  set @last_id_in_thumbnails = LAST_INSERT_ID();
  
  INSERT INTO $TABLE (created_at, updated_at, thumnailID) VALUES 
  
  (NOW(),NOW(),@last_id_in_thumbnails);
  
  SET @last_id_in_livestreams = LAST_INSERT_ID();
  
  INSERT INTO $TABLE2 (created_at, updated_at, filename,liveID) VALUES 
  
  (NOW(),NOW(),'$FILENAME',@last_id_in_livestreams); 
  
  EOF

  I have not got the nginx rtmp logs installed, I can obviously do this however some of the logs appear in the nginx error.log, strangely the latest stream I tried did not update in the access log, however I think this is because I did not attempt to connect to it via any method. I don’t fully understand the error.log, in my stupidity I decided to use nginx with which I am quite inexperienced and I am finding it very difficult to troubleshoot this issue, it appears to me that as part of the RTMP protocol or my streaming software (OBS) is either directly pinging the rtmp stream or is being pinged by the server to ensure the connection is still there. And this ping is

  I have left a stream running from approx 4 minutes without interacting with the server, streaming software, computer running the stream, I have ensured the internet connection is constant as my first though was the connection dropped, however on inspecting the database the executing is done always after at least 11 seconds however usually this is 16, I can’t seem to figure out how to select the closest dates from the database however there has been at least a few 17 second differences (potentially when
  I am unsure if this is an issue or if it is intended behavior but I do require this to finish a university degree, I’m not asking for answers but if it is a legitimate issue then I would be happy to spend as much time I can commit to it if some insight into what is causing it, or if there is a workaround I believe it should be documented somewhere, I have googled into making any exec commands run only once on publish

  I can’t seem to pinpoint where in the log the issue is happening however think it is something to do with the below exceprts I would attach the file but can’t seem to select all lines after the timestamp upon starting a stream

  2017/07/26 18:17:35 [info] 1451#0: *2229 exec: starting managed child 
  
  'ffmpeg', client: 192.168.0.78, server: 0.0.0.0:1935
  
  2017/07/26 18:17:35 [info] 1451#0: *2412 client connected '192.168.0.98'
  
  2017/07/26 18:17:35 [info] 1451#0: *2412 connect: app='live' args='' 
  
  flashver='LNX 9,0,124,2' swf_url='' tc_url='rtmp://192.168.0.98:1935/live' 
  
  page_url='' acodecs=4071 vcodecs=252 object_encoding=0, client: 
  
  192.168.0.98, server: 0.0.0.0:1935
  
  2017/07/26 18:17:35 [info] 1451#0: *2412 createStream, client: 192.168.0.98, 
  
  server: 0.0.0.0:1935
  
  2017/07/26 18:17:35 [info] 1451#0: *2412 play: name='newname' args='' 
  
  start=-2000 duration=0 reset=0 silent=0, client: 192.168.0.98, server:  
  
  0.0.0.0:1935
  
  2017/07/26 18:17:36 [info] 1451#0: *2410 recv() failed (104: Connection  
  
  reset by peer), client: 192.168.0.98, server: 0.0.0.0:1935
  
  2017/07/26 18:17:36 [info] 1451#0: *2410 disconnect, client: 192.168.0.98, 
  
  server: 0.0.0.0:1935
  
  2017/07/26 18:17:36 [info] 1451#0: *2410 deleteStream, client: 192.168.0.98, 
  
  server: 0.0.0.0:1935
  
  2017/07/26 18:17:36 [notice] 1451#0: signal 17 (SIGCHLD) received
  
  2017/07/26 18:17:36 [notice] 1451#0: unknown process 10487 exited with code 
  
  0
  
  2017/07/26 18:17:36 [info] 1451#0: *2229 exec: child 10487 exited; ignoring, 
  
  client: 192.168.0.78, server: 0.0.0.0:1935
  
  
  
  ver: 0.0.0.0:1935
  
  2017/07/26 18:17:41 [info] 1451#0: *2229 exec: starting managed child 
  
  'usr/local/bin/make_thumbnail.sh', client: 192.168.0.78, server:  
  
  0.0.0.0:1935
  
  2017/07/26 18:17:41 [info] 1451#0: *2413 client connected '192.168.0.98'
  
  2017/07/26 18:17:41 [info] 1451#0: *2413 connect: app='live' args='' 
  
  flashver='LNX 9,0,124,2' swf_url='' tc_url='rtmp://192.168.0.98:1935/live' 
  
  page_url='' acodecs=4071 vcodecs=252 object_encoding=0, client: 
  
  192.168.0.98, 
  
  server: 0.0.0.0:1935
  
  2017/07/26 18:17:41 [info] 1451#0: *2413 createStream, client: 192.168.0.98, 
  
  server: 0.0.0.0:1935
  
  2017/07/26 18:17:41 [info] 1451#0: *2413 play: name='newname' args='' 
  
  start=-2000 duration=0 reset=0 silent=0, client: 192.168.0.98, server: 
  
  0.0.0.0:1935
  
  2017/07/26 18:17:43 [info] 1451#0: *2229 exec: starting managed child 
  
  'ffmpeg', 
  
  client: 192.168.0.78, server: 0.0.0.0:1935
  
  2017/07/26 18:17:43 [info] 1451#0: *2414 client connected '192.168.0.98'
  
  2017/07/26 18:17:43 [info] 1451#0: *2414 connect: app='live' args='' 
  
  flashver='LNX 9,0,124,2' swf_url='' tc_url='rtmp://192.168.0.98:1935/live' 
  
  page_url='' acodecs=4071 vcodecs=252 object_encoding=0, client: 
  
  192.168.0.98, 
  
  server: 0.0.0.0:1935
  
  @

• Evolution #3229 (Nouveau) : Utiliser une méthode standard dès que l’on a besoin d’un module JS

  14 juin 2014, par RastaPopoulos ♥

  Actuellement, on a tendance :
  - soit à charger du JS sur toutes les pages
  - soit à le charger au moment où on en a besoin
  - soit, mieux mais fastidieux, à gérer soi-même à la main un mécanisme de chargement unique d’un module (cf GIS dernièrement)

  Il semblerait pourtant qu’il existe depuis plusieurs années enfin des méthodes pour appeler un ou plusieurs modules uniquement lorsque nécessaire, et en pouvant les appeler plusieurs fois sans risque. Bref, l’équivalent des inclusions PHP.

  Pour jQuery de base je ne sais pas, mais ce serait au moins utile pour tel ou tel module de jQueryUI, pour le JS de datation de SPIP, (inc-dateur), pour Leaflet, etc.

  Doit-on vraiment réinventer la roue à chaque fois qu’on veut n’inclure un JS qu’une seule fois ?

  http://requirejs.org/docs/whyamd.html
  http://developer.telerik.com/featured/jquery-using-only-what-you-need/

  Qu’en pensent les experts en cache et en performance ? :)