I’ve developed a distributed ffmpeg google app engine solution.
The solution uses a combination of publish subscribe and redis queues for distributed communication.
https://github.com/dmzoneill/appengine-ffmpeg
Its composed of 2 services which scale horizontally (default and worker).
Coordinator (default service and human interface)
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 | #!/usr/bin/python from flask import Flask import os import sys import glob import string import random import redis import logging from gcloud import storage, pubsub from google.cloud import logging PROJECT_ID = 'transcode-159215' TOPIC = 'projects/{}/topics/message'.format(PROJECT_ID) logclient = logging.Client() logger = logclient.logger( "ffmpeg-pool" ) app = Flask(__name__) app.config[ "SECRET_KEY" ] = "test" app.debug = True def publish( msg ): pubsub_client = pubsub.Client( PROJECT_ID ) topic = pubsub_client.topic( "ffmpeg-pool" ) if not topic.exists(): topic.create() topic.publish( msg ) @app.route( "/readlog" ) def readLog(): msg = "" try: for entry in logger.list_entries(): msg = msg + entry.payload + "<br>" logger.delete() except: msg = "" return msg @app.route( "/cleantopic" ) def cleanTopics(): client = pubsub.Client( PROJECT_ID ) topic = client.topic( "ffmpeg-pool" ) topic.delete() topic.create() return "Cleaned topic" @app.route( "/split" ) def split(): publish( "split" ) return "File queued for spliting" @app.route( "/transcode" ) def transcode(): publish( "transcode" ) return "Job queued for transcoding" @app.route( "/combine" ) def combine(): publish( "combine" ) return "Job queued for combining" @app.route( "/" ) def home(): return "<a href='/split'>/split</a> | <a href='/transcode'>/transcode</a> | <a href='/combine'>/combine</a> | <a href='/cleantopic'>/cleantopic</a> | <a href='/readlog'>/readlog</a>" if __name__ == '__main__': app.run(host='127.0.0.1', port=8080, debug=True) |
Worker
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 | import os from gcloud import storage, pubsub, logging import sys import socket import time import redis import glob from google.cloud import logging logclient = logging.Client() logger = logclient.logger( "ffmpeg-pool" ) PROJECT_ID = 'transcode-159215' TOPIC = 'projects/{}/topics/message'.format(PROJECT_ID) psclient = None pstopic = None pssub = None class RedisQueue(object): def __init__( self, name, namespace = 'queue' ): self.__db = redis.Redis( host = "redis-11670.c10.us-east-1-4.ec2.cloud.redislabs.com", port=11670 ) self.key = '%s:%s' %(namespace, name) def qsize( self ): return self.__db.llen( self.key ) def empty( self ): return self.qsize() == 0 def put( self, item ): self.__db.rpush( self.key, item ) def get( self, block=True, timeout=None ): if block: item = self.__db.blpop( self.key, timeout=timeout ) else: item = self.__db.lpop( self.key ) if item: item = item[1] return item def get_nowait( self ): return self.get( False ) def download( rfile ): client = storage.Client( PROJECT_ID ) bucket = client.bucket( PROJECT_ID + ".appspot.com" ) blob = bucket.blob( rfile ) with open( "/tmp/" + rfile, 'w' ) as f: blob.download_to_file( f ) logger.log_text( "Worker: Downloaded: /tmp/" + rfile ) def upload( rfile ): client = storage.Client( PROJECT_ID ) bucket = client.bucket( PROJECT_ID + ".appspot.com" ) blob = bucket.blob( rfile ) blob = bucket.blob( rfile ) blob.upload_from_file( open( "/tmp/" + rfile ) ) logger.log_text( "Worker: Uploaded /tmp/" + rfile ) def transcode( rfile ): download( rfile ) os.system( "rm /tmp/output*" ) ret = os.system( "ffmpeg -i /tmp/" + rfile + " -c:v libx265 -preset medium -crf 28 -c:a aac -b:a 128k -strict -2 /tmp/output-" + rfile + ".mkv" ) if ret: logger.log_text( "Worker: convert failed : " + rfile + " - " + str( ret ).encode( 'utf-8' ) ) return upload( "output-" + rfile + ".mkv" ) def split(): rqueue = RedisQueue( "test" ) download( "sample.mp4" ) os.system( "rm -f /tmp/chunk*" ) ret = os.system( "ffmpeg -i /tmp/sample.mp4 -map 0:a -map 0:v -codec copy -f segment -segment_time 10 -segment_format matroska -v error '/tmp/chunk-%03d.orig'" ) if ret: return "Failed" for rfile in glob.glob( "/tmp/chunk*" ): basename = os.path.basename( rfile ) upload( basename ) rqueue.put( basename ) def combine(): client = storage.Client( PROJECT_ID ) bucket = client.bucket( PROJECT_ID + ".appspot.com" ) blobs = bucket.list_blobs() os.system( "rm /tmp/*" ) names = [] for blob in blobs: if "output" in blob.name: names.append( blob.name.encode( 'utf-8' ) ) names.sort() with open( '/tmp/combine.lst', 'w' ) as f1: for name in names: f1.write( "file '/tmp/" + name + "'\n" ) download( name ) logger.log_text( "Worker: created combine list: /tmp/combine.lst" ) ret = os.system( "ffmpeg -f concat -safe 0 -i /tmp/combine.lst -c copy /tmp/combined.mkv" ) if ret: logger.log_text( "Worker: combine failed: /tmp/combine.mkv - " + str(ret).encode( 'utf-8' ) ) return upload( "combined.mkv" ) def subscribe(): global psclient, pstopic, pssub psclient = pubsub.Client( PROJECT_ID ) pstopic = psclient.topic( "ffmpeg-pool" ) if not pstopic.exists(): pstopic.create() pssub = pstopic.subscription( "ffmpeg-worker-" + socket.gethostname() ) if not pssub.exists(): pssub.create() def handlemessages(): global psclient, pstopic, pssub rqueue = RedisQueue( 'test' ) subscribe() while True: messages = pssub.pull( return_immediately=False, max_messages=110 ) for ack_id, message in messages: payload = message.data.encode( 'utf-8' ).replace( u"\u2018", "'" ).replace( u"\u2019", "'" ) logger.log_text( "Worker: Received message: " + payload ) try: pssub.acknowledge( [ack_id] ) if payload == "combine": combine() elif payload == "split": split() else: rfile = rqueue.get() basename = os.path.basename( rfile ) logger.log_text( "Worker: Redis popped: " + basename ) while basename != "None": transcode( basename ) rfile = rqueue.get() basename = os.path.basename( rfile ) logger.log_text( "Worker: Redis popped: " + rfile ) except Exception as e: logger.log_text( "Worker: Error: " + e.message ) sys.stderr.write( e.message ) subscribe() time.sleep( 1 ) if __name__ == '__main__': handlemessages() |
