Container logs can very large and should be compressed

[yanokwa]

Containers logs (particularly nginx and service) can be very large and take up a lot of space. Also, it isn't fast to docker logs container | grep foo them for issues.

We can't use logrotate inside the containers because they all report out on stdout/err, but it looks like we can configure the “local” logging driver and it does log-rotation by default and still works with docker logs, so we can do docker logs --since 10m <container_id>.

My bias is to do this for every container in the docker-compose.yml file.

[yanokwa]

I looked at the logs on server and here's what is taking up space since the server started in July 2021 (roughly 3 months) and has 550k submissions

nginx 637M
service  14M
enketo  27K
postgres  56K
pyxform  45K
redis  54K
redis 118K
smtp 1.2K

Given this, I propose we only deal with logging for nginx and service.