PreparePaths(), WriteInfo(), DownloadUrlFile(control), WgetDownload( wpull_args, accept_on_exit_code=AcceptAny(), env={ 'ITEM_IDENT': ItemInterpolation('%(ident)s'), 'LOG_KEY': ItemInterpolation('%(log_key)s'), 'REDIS_URL': REDIS_URL, 'PATH': os.environ['PATH'] } ), RelabelIfAborted(control), WriteInfo(), MoveFiles(), LimitConcurrent(2, RsyncUpload( target = RSYNC_URL, target_source_path = ItemInterpolation("%(data_dir)s"), files=ItemValue("all_target_files"), extra_args = [ '--partial', '--partial-dir', '.rsync-tmp' ] ) ), StopHeartbeat(), MarkItemAsDone(control, EXPIRE_TIME) )
pipeline = Pipeline( CheckIP(), GetItemFromQueue(control, pipeline_id, downloader, ao_only=env.get('AO_ONLY'), large=env.get('LARGE')), StartHeartbeat(control), SetFetchDepth(), PreparePaths(), WriteInfo(), DownloadUrlFile(control), WgetDownload(wpull_args, accept_on_exit_code=AcceptAny(), env={ 'ITEM_IDENT': ItemInterpolation('%(ident)s'), 'LOG_KEY': ItemInterpolation('%(log_key)s'), 'REDIS_URL': REDIS_URL, 'PATH': os.environ['PATH'] }), RelabelIfAborted(control), WriteInfo(), MoveFiles(), LimitConcurrent( 2, RsyncUpload(target=RSYNC_URL, target_source_path=ItemInterpolation("%(data_dir)s"), files=ItemValue("all_target_files"), extra_args=['--partial', '--partial-dir', '.rsync-tmp'])), StopHeartbeat(), MarkItemAsDone(control, EXPIRE_TIME)) def stop_control(): #control.flag_logging_thread_for_termination() control.unregister_pipeline(pipeline_id) pipeline.on_cleanup += stop_control
pipeline_id, downloader, ao_only=env.get('AO_ONLY'), large=env.get('LARGE'), version_check=(VERSION, pipeline_version)), StartHeartbeat(control), SetFetchDepth(), PreparePaths(), WriteInfo(), DownloadUrlFile(control), Wpull(wpull_args, accept_on_exit_code=AcceptAny(), env={ 'ITEM_IDENT': ItemInterpolation('%(ident)s'), 'LOG_KEY': ItemInterpolation('%(log_key)s'), 'REDIS_URL': REDIS_URL, 'PATH': os.environ['PATH'] }), RelabelIfAborted(control), CompressLogIfFailed(), WriteInfo(), MoveFiles(target_directory=os.environ["FINISHED_WARCS_DIR"]), StopHeartbeat(), MarkItemAsDone(control, EXPIRE_TIME)) def stop_control(): #control.flag_logging_thread_for_termination() control.unregister_pipeline(pipeline_id) pipeline.on_cleanup += stop_control pipeline.running_status = "Running" def status_running(): pipeline.running_status = "Running"