Disable access logging or change level of logs for the health endpoints

[moritzheiber]

I would like to be able to disable access logging for the health endpoints, so that I can make sense of my application logs again. Inside Kubernetes, the health endpoints of Keto are checked regularly, with each checks equal to two lines in its logfile.

I would propose either of these two solutions:

1. Disable access logging for the health endpoints, akin to what Hydra already has implemented.
2. Shift the level of severity for the logs from info to debug to avoid logging them in a "production" context.

As an example, here are two lines in my logs that I encounter on a regular basis (i.e. every couple of seconds):

{"msg": "started handling request", "request": "/health/ready", "method": "GET", "level": "info", "time": "2020-01-22T11:51:21Z", "remote": "127.0.0.1:50224"}
{"msg": "completed handling request", "text_status": "OK", "request": "/health/alive", "took": 78385,  "measure#keto": {"latency": 78385}, "method": "GET", "level": "info", "time": "2020-01-22T11:52:42Z", "remote": "127.0.0.1:37770", "status": 200}

Note: Not the same event.

Thanks!

[aeneasr]

Good idea, I think it would make sense to do what we've already done in ORY Oathkeeper:

https://github.com/ory/oathkeeper/blob/master/cmd/server/server.go#L82

What do you think? If you're up for it, we'd appreaciate a PR!

[moritzheiber]

See #151