Silence healthcheck requests from the log

[dhh]

The default Rails health check is intended to be hit every second or so by Kamal to ensure that the application is still or has just become responsive. This clutters up the production logs and makes it hard to follow what's actually going on. This adds a new middleware called Rails::Rack::SilenceRequest that is used by a new config config.silence_healthcheck = path to prevent that.

[ghiculescu]

@dhh did you mean to delete this test?

[dhh]

I did not. Not sure what happened there. Please do PR to restore!

[ghiculescu]

#52812

[AnalyzePlatypus]

Will this ship in Rails 8, or in an earlier point release?

[dhh]

8. We don't back port features.

[cristianbica]

@dhh how do you feel abstracting this to a config.silence_requests_paths = ['/up', '/others']?
I'm thinking at use-cases like an endpoint exporting some metrics, people running rack-mini-profiler in production or people just filtering out logging for some endpoints which just adds noise to logging.

[dhh]

@cristianbica Are any of those examples you mention problems you've actually hit in production? At what intervals are those requests running? I don't want to blanket rule out making this more generic, but I find that you're better off with something specific, unless you've proven without a doubt the need for something generic.

[adampope]

@dhh In our app we have an async request for red dot / badge counts that happens on most page loads. I recently wrote a very similar middleware to filter those out of the development logs as I got bored of scrolling back up the tail past the badge request to see details of the main request. Having something configurable like @cristianbica is suggesting, and being able to be different in different environments would certainly have been great for our real use case.

[cristianbica]

@cristianbica Are any of those examples you mention problems you've actually hit in production? At what intervals are those requests running? I don't want to blanket rule out making this more generic, but I find that you're better off with something specific, unless you've proven without a doubt the need for something generic.

@dhh on my current projects I do not have a valid use-case to prove the need to generalize this functionality.

In the past projects I've implemented different solution to reduce logging (e.g. silencer gem for requests). The purpose was to reduce clutter in the logs and reduce costs. From what I remember:

• we had an endpoint exposing sidekiq metrics called minutely
• we had many webhook-like requests - those were recorded and processed in the background; logging was too verbose so there was no benefit to log them

Anyway, looking at the implementation in this PR, one could easy filter out logging for other paths by adding to application.rb multiple lines like this

config.middleware.insert_before ::Rails::Rack::Logger, :Rails::Rack::SilenceRequest, path: '/sidekiq_stats'

[morgoth]

@dhh With thruster gem included by default, the "/up" requests are showing up in the logs anyway. Is there any way to silence that as well, as I don't see such possibility in thruster project yet?