Sur d’autres sites (8590)

• ffmpeg stream chrome kiosk mode ubuntu 16.04 server

  21 décembre 2016, par Raul

  I have a weird out-of-sync issue while using ffmpeg to stream to youtube live a chrome browser from an ub untu 16.04 server.

  Issue : output video streamed to youtube has audio/video out of sync, sometimes with as much as 3s

  Current flow :

  1) start pulseaudio - we using something like this to start it :

  pulseaudio --start -vvv --disallow-exit --log-target=syslog --high-priority --exit-idle-time=-1 --daemonize

  2) start Xvfb

  Xvfb :0 -ac -screen 0 1920x1080x24

  3) start chrome linux in kiosk mode

  google-chrome --kiosk --disable-gpu --incognito --no-first-run --disable-java --disable-plugins --disable-translate --disk-cache-size=$((1024 * 1024)) --disk-cache-dir=/tmp/chrome/ --user-data-dir=/tmp/chrome/ --force-device-scale-factor=1 --window-size=1920,1080 --window-position=0,0 LOCATION_URL

  4) start ffmpeg

  ffmpeg -y \
  
    -thread_queue_size 8192 -rtbufsize 250M -f x11grab -video_size 1920x1080 -framerate 24 -i :0 \
  
    -thread_queue_size 8192 -channel_layout stereo -f alsa -i pulse \
  
    -c:v libx264 -pix_fmt yuv420p -c:v libx264 -g 48 -crf 24 -filter:v fps=24 -preset ultrafast -tune zerolatency \
  
    -c:a aac -strict -2 -channel_layout stereo -ab 96k -ac 2 -flags +global_header \
  
    -f flv YOUTUBE_LIVE_STREAMING_RTMP

  Note : this is running on an amazon ec2 instance, meaning there is no soundcard, so alsa and pulseaudio are creating a dummy audio card. However, the latency does not come from there. Logs :

  Nov 25 06:14:22 ip-172-31-29-8 pulseaudio[26602]: [pulseaudio] protocol-native.c: Adjust latency mode enabled, configuring sink latency to half of overall latency.
  
  Nov 25 06:14:22 ip-172-31-29-8 pulseaudio[26602]: [pulseaudio] protocol-native.c: Requested latency=23.22 ms, Received latency=23.22 ms
  
  Nov 25 06:14:22 ip-172-31-29-8 pulseaudio[26602]: [pulseaudio] protocol-native.c: Final latency 69.66 ms = 23.22 ms + 2*11.61 ms + 23.22 ms

  At this point, here’s what we observed :

  1. if we start ffmpeg exactly after issuing the command to start chrome, we see the DTS errors from ffmpeg. Audio is out of sync with the video and has delay of 3-5seconds AHEAD. We also noticed the out of sync remains the same for the full duration of the stream

  2. if we start ffmpeg after around 10seconds, audio and video are almost in sync. We then manually added a -itsoffset -0.125 to the ffmpeg command and everything is perfect.

  Questions :

  1. Why would ffmpeg have so much lag if it’s started right after chrome ?
  2. Is starting the ffmpeg after 10s or X seconds the expected behavior ? That is, is this because the system needs to wait for audio/video signals to be "ready" or something ?
  3. Is there a way to 100% calculate or know when Chrome is fully ready and start ffmpeg ? We found sometimes it takes 5s, sometimes 10. Depends on the URL we load.
  4. Besides the DTS error that ffmpeg throws, is there any other way to know if audio/video is out-of-sync ? as sometimes we have a delay of between 0.5 to 1s, but ffmpeg does not report anything. And a restart is required to "re-balance" the audio/video inputs and get them back in sync.
  5. Can pulseaudio be the problem in this scenario ?

  Thank you

  UPDATE Dec 20

  We were able to do some tricks to force chrome to start the audio on page load, and that will force connect to pulseaudio. Doing that, plus adding a 3s delay for ffmpeg to start, there is no more delay when ffmpeg starts.
  However, our app is a webRTC app, and we have a STRANGER thing happening : if we start the page with no webcam/audio, once the webcam/audio is enabled, ffmpeg (while showing no errors) has a delay of 2s or so. While keep talking, in about max 30s, that delay is GONE.

  So the new questions are :

  1. Besides the DTS error that ffmpeg throws, is there any other way to know if audio/video is out-of-sync ? as sometimes we have a delay of between 0.5 to 1s, but ffmpeg does not report anything.
  2. What could cause the initial audio/video out of sync issue and then catching up ?

• ffmpeg stream chrome kiosk mode ubuntu 16.04 server

  15 février 2021, par Raul

  I have a weird out-of-sync issue while using ffmpeg to stream to youtube live a chrome browser from an ub untu 16.04 server.

  



  Issue : output video streamed to youtube has audio/video out of sync, sometimes with as much as 3s

  



  Current flow :

  



  1) start pulseaudio - we using something like this to start it :

  



  pulseaudio --start -vvv --disallow-exit --log-target=syslog --high-priority --exit-idle-time=-1 --daemonize


  



  2) start Xvfb

  



  Xvfb :0 -ac -screen 0 1920x1080x24


  



  3) start chrome linux in kiosk mode

  



  google-chrome --kiosk --disable-gpu --incognito --no-first-run --disable-java --disable-plugins --disable-translate --disk-cache-size=$((1024 * 1024)) --disk-cache-dir=/tmp/chrome/ --user-data-dir=/tmp/chrome/ --force-device-scale-factor=1 --window-size=1920,1080 --window-position=0,0 LOCATION_URL


  



  4) start ffmpeg

  



  ffmpeg -y \
  -thread_queue_size 8192 -rtbufsize 250M -f x11grab -video_size 1920x1080 -framerate 24 -i :0 \
  -thread_queue_size 8192 -channel_layout stereo -f alsa -i pulse \
  -c:v libx264 -pix_fmt yuv420p -c:v libx264 -g 48 -crf 24 -filter:v fps=24 -preset ultrafast -tune zerolatency \
  -c:a aac -strict -2 -channel_layout stereo -ab 96k -ac 2 -flags +global_header \
  -f flv YOUTUBE_LIVE_STREAMING_RTMP


  



  Note : this is running on an amazon ec2 instance, meaning there is no soundcard, so alsa and pulseaudio are creating a dummy audio card. However, the latency does not come from there. Logs :

  



  Nov 25 06:14:22 ip-172-31-29-8 pulseaudio[26602]: [pulseaudio] protocol-native.c: Adjust latency mode enabled, configuring sink latency to half of overall latency.
Nov 25 06:14:22 ip-172-31-29-8 pulseaudio[26602]: [pulseaudio] protocol-native.c: Requested latency=23.22 ms, Received latency=23.22 ms
Nov 25 06:14:22 ip-172-31-29-8 pulseaudio[26602]: [pulseaudio] protocol-native.c: Final latency 69.66 ms = 23.22 ms + 2*11.61 ms + 23.22 ms


  



  At this point, here's what we observed :

  



  


  1. if we start ffmpeg exactly after issuing the command to start chrome, we see the DTS errors from ffmpeg. Audio is out of sync with the video and has delay of 3-5seconds AHEAD. We also noticed the out of sync remains the same for the full duration of the stream

  


  2. if we start ffmpeg after around 10seconds, audio and video are almost in sync. We then manually added a -itsoffset -0.125 to the ffmpeg command and everything is perfect.

  


  



  Questions :

  



  

  1. Why would ffmpeg have so much lag if it's started right after chrome ?

  


  2. Is starting the ffmpeg after 10s or X seconds the expected behavior ? That is, is this because the system needs to wait for audio/video signals to be "ready" or something ?

  


  3. Is there a way to 100% calculate or know when Chrome is fully ready and start ffmpeg ? We found sometimes it takes 5s, sometimes 10. Depends on the URL we load.

  


  4. Besides the DTS error that ffmpeg throws, is there any other way to know if audio/video is out-of-sync ? as sometimes we have a delay of between 0.5 to 1s, but ffmpeg does not report anything. And a restart is required to "re-balance" the audio/video inputs and get them back in sync.

  


  5. Can pulseaudio be the problem in this scenario ?

  


  



  Thank you

  



  UPDATE Dec 20

  



  We were able to do some tricks to force chrome to start the audio on page load, and that will force connect to pulseaudio. Doing that, plus adding a 3s delay for ffmpeg to start, there is no more delay when ffmpeg starts.
However, our app is a webRTC app, and we have a STRANGER thing happening : if we start the page with no webcam/audio, once the webcam/audio is enabled, ffmpeg (while showing no errors) has a delay of 2s or so. While keep talking, in about max 30s, that delay is GONE.

  



  So the new questions are :

  



  

  1. Besides the DTS error that ffmpeg throws, is there any other way to know if audio/video is out-of-sync ? as sometimes we have a delay of between 0.5 to 1s, but ffmpeg does not report anything.

  


  2. What could cause the initial audio/video out of sync issue and then catching up ?

  


  


• Installing ffmpeg, librosa nad pydub in Apache Spark container

  17 avril 2023, par yaviens

  I'm working on a Python spark streaming project, that needs to use pydub and librosa to process audio, this libraries require having installed ffmpeg library. I'm having trubles to build the spark containers with this libraries, I don't know how to solve this problem.

  


  I use a docker-compose.yml to buil the images, define ports, etc of the spark master and workers.
docker-compose :

  


  version: "3.3"
services:
  spark-master:
    build:
      context: ./
      dockerfile: Dockerfile
    #image: docker.io/bitnami/spark:3.3
    ports:
      - "9090:8080"
      - "7077:7077"
    volumes:
       - ./apps:/opt/spark-apps
       - ./data:/opt/spark-data
       - ./data:/data
       - ./src:/src
       - ./output:/output
    environment:
      - SPARK_LOCAL_IP=spark-master
      - SPARK_WORKLOAD=master
  spark-worker-a:
    build:
      context: ./
      dockerfile: Dockerfile
    #image: docker.io/bitnami/spark:3.3
    ports:
      - "9091:8080"
      - "7000:7000"
    depends_on:
      - spark-master
    environment:
      - SPARK_MASTER=spark://spark-master:7077
      - SPARK_WORKER_CORES=1
      - SPARK_WORKER_MEMORY=1G
      - SPARK_DRIVER_MEMORY=1G
      - SPARK_EXECUTOR_MEMORY=1G
      - SPARK_WORKLOAD=worker
      - SPARK_LOCAL_IP=spark-worker-a
    volumes:
       - ./apps:/opt/spark-apps
       - ./data:/opt/spark-data
       - ./data:/data
       - ./src:/src
       - ./output:/output       
  spark-worker-b:
    build:
      context: ./
      dockerfile: Dockerfile
    #image: docker.io/bitnami/spark:3.3
    ports:
      - "9092:8080"
      - "7001:7000"
    depends_on:
      - spark-master
    environment:
      - SPARK_MASTER=spark://spark-master:7077
      - SPARK_WORKER_CORES=1
      - SPARK_WORKER_MEMORY=1G
      - SPARK_DRIVER_MEMORY=1G
      - SPARK_EXECUTOR_MEMORY=1G
      - SPARK_WORKLOAD=worker
      - SPARK_LOCAL_IP=spark-worker-b
    volumes:
        - ./apps:/opt/spark-apps
        - ./data:/opt/spark-data
        - ./data:/data
        - ./src:/src
        - ./output:/output 


  


  In the same path of the docker-compose.yml is the Dockerfile I'm using to build the image :

  


  # builder step used to download and configure spark environment
FROM openjdk:11.0.11-jre-slim-buster as builder

# Add Dependencies for PySpark
RUN apt-get update && apt-get install -y curl vim wget software-properties-common ssh net-tools ca-certificates python3 python3-pip python3-numpy python3-matplotlib python3-scipy python3-pandas python3-simpy

RUN update-alternatives --install "/usr/bin/python" "python" "$(which python3)" 1

# Fix the value of PYTHONHASHSEED
# Note: this is needed when you use Python 3.3 or greater
ENV SPARK_VERSION=3.0.2 \
HADOOP_VERSION=3.2 \
SPARK_HOME=/opt/spark \
PYTHONHASHSEED=1

# Download and uncompress spark from the apache archive
RUN wget --no-verbose -O apache-spark.tgz "https://archive.apache.org/dist/spark/spark-${SPARK_VERSION}/spark-${SPARK_VERSION}-bin-hadoop${HADOOP_VERSION}.tgz" \
&& mkdir -p /opt/spark \
&& tar -xf apache-spark.tgz -C /opt/spark --strip-components=1 \
&& rm apache-spark.tgz


# Apache spark environment
FROM builder as apache-spark

WORKDIR /opt/spark

ENV SPARK_MASTER_PORT=7077 \
SPARK_MASTER_WEBUI_PORT=8080 \
SPARK_LOG_DIR=/opt/spark/logs \
SPARK_MASTER_LOG=/opt/spark/logs/spark-master.out \
SPARK_WORKER_LOG=/opt/spark/logs/spark-worker.out \
SPARK_WORKER_WEBUI_PORT=8080 \
SPARK_WORKER_PORT=7000 \
SPARK_MASTER="spark://spark-master:7077" \
SPARK_WORKLOAD="master"

EXPOSE 8080 7077 6066

RUN mkdir -p $SPARK_LOG_DIR && \
touch $SPARK_MASTER_LOG && \
touch $SPARK_WORKER_LOG && \
ln -sf /dev/stdout $SPARK_MASTER_LOG && \
ln -sf /dev/stdout $SPARK_WORKER_LOG

# Install ffmpeg lib
RUN apt-get -y update
RUN apt-get -y upgrade
RUN apt-get install -y ffmpeg
RUN apt-get -y install apt-utils gcc libpq-dev libsndfile-dev

#RUN apt-get update \
#&& apt-get upgrade -y \
#&& apt-get install -y \
#&& apt-get -y install apt-utils gcc libpq-dev libsndfile-dev

# Install required python libs
COPY requirements.txt .
RUN pip3 install -r requirements.txt

COPY start-spark.sh /

CMD ["/bin/bash", "https://net.cloudinfrastructureservices.co.uk/start-spark.sh"]


  


  The start-spark.sh :

  


  #!/bin/bash
. "https://net.cloudinfrastructureservices.co.uk/opt/spark/bin/load-spark-env.sh"
# When the spark work_load is master run class org.apache.spark.deploy.master.Master
if [ "$SPARK_WORKLOAD" == "master" ];
then

export SPARK_MASTER_HOST=`hostname`

cd /opt/spark/bin && ./spark-class org.apache.spark.deploy.master.Master --ip $SPARK_MASTER_HOST --port $SPARK_MASTER_PORT --webui-port $SPARK_MASTER_WEBUI_PORT >> $SPARK_MASTER_LOG

elif [ "$SPARK_WORKLOAD" == "worker" ];
then
# When the spark work_load is worker run class org.apache.spark.deploy.master.Worker
cd /opt/spark/bin && ./spark-class org.apache.spark.deploy.worker.Worker --webui-port $SPARK_WORKER_WEBUI_PORT $SPARK_MASTER >> $SPARK_WORKER_LOG

elif [ "$SPARK_WORKLOAD" == "submit" ];
then
    echo "SPARK SUBMIT"
else
    echo "Undefined Workload Type $SPARK_WORKLOAD, must specify: master, worker, submit"
fi


  


  When I execute : "docker-compose up" I have the next error message :

  


  [+] Running 4/4
 - Network spark_container_default             Created                                                                                0.9s 
 - Container spark_container-spark-master-1    Created                                                                                0.3s 
 - Container spark_container-spark-worker-a-1  Created                                                                                0.4s 
 - Container spark_container-spark-worker-b-1  Created                                                                                0.4s 
Attaching to spark_container-spark-master-1, spark_container-spark-worker-a-1, spark_container-spark-worker-b-1
spark_container-spark-master-1    | /bin/bash: https://net.cloudinfrastructureservices.co.uk/start-spark.sh: No such file or directory     
spark_container-spark-master-1 exited with code 127
spark_container-spark-worker-a-1  | /bin/bash: https://net.cloudinfrastructureservices.co.uk/start-spark.sh: No such file or directory     
spark_container-spark-worker-b-1  | /bin/bash: https://net.cloudinfrastructureservices.co.uk/start-spark.sh: No such file or directory     
spark_container-spark-worker-a-1 exited with code 127
spark_container-spark-worker-b-1 exited with code 127