This PR updates some of the internal config to make it much easier to track the logs from our application, and aggregate them in our centralised logging server.
More specifically, it creates files like so, that are easy to pick up and read with our log forwarders
We rely on gunicorn's native handling of multiple processes rather than supervisor' support.
This means that supervisor watches one main gunicorn process, which forks other workers, and all their logs are aggregated and fed into the following two files.
For the background worker processes
Logs for the workers using dramatiq go to the following files:
This PR updates some of the internal config to make it much easier to track the logs from our application, and aggregate them in our centralised logging server.
More specifically, it creates files like so, that are easy to pick up and read with our log forwarders
For the web servers
Logs go to the following files.
/var/log/web_prod_00.log /var/log/web_prod_00.error.log
We rely on gunicorn's native handling of multiple processes rather than supervisor' support.
This means that supervisor watches one main gunicorn process, which forks other workers, and all their logs are aggregated and fed into the following two files.
For the background worker processes
Logs for the workers using dramatiq go to the following files:
/var/log/worker_00.log /var/log/worker_00.error.log
Just like gunicorn, we really on dramatiq's own handling of subprocesses, that rather than supervisord.