On the webs start the rack application as how you usualy do. On the worker containers or server start with $ rack-app workers start
or if you want run as a service in the background use the -d flag $ rack-app workers start -d
to stop the daemonized process $ rack-app workers stop
The workers use rabbitmq as message broker so you can easily scale horizontally the workers as how your pocket allow this. By default the worker will scale vertically on the server as much as possible if the load requires.
You can configurate the behavior with Linux variables:
this is required for connecting to rabbitmq
this will manage how much message should be prefetched to a consumer. The default value is 50.
This variable manage the time delay between a new consumer creation or remove based on queue size demand
This is used for limiting the maximum vertical of worker process count use
This is used to set the log output level for the extensions log messages
If this configure, the background process will pipe all stdout io actions to the given output.
Same as WORKER_STDOUT but for ERR
If you plan using multiple application on the same container/server than you should use worker namespace to add application specific uniq part to the queues.
this is an optional env variable, this defines the workers cluster such as 'backup' or 'secondary'. This is useful when you have two seperated database and you want populate both at once, but you only want one public web api which accept the incoming content.