#httplog

HttpLog is a super useful GEM that I use in many of my Rails projects to visualize the HTTP requests made by the code and debug issues.

A good configuration to use with its last version (v0.99) is the following

HttpLog.configure do |config| # Enable logging in Prod Console and everywhere in Dev and Test. Also enable if environment variable is set. config.enabled = not(Rails.env.production?) || ENV.has_key?('HTTPLOG_URL_WHITELIST_PATTERN') || defined?(Rails::Console) # Assign minimum log severity to show entries in logs config.severity = Rails.logger.level # Tweak which parts of the HTTP cycle to log... config.log_connect = false config.log_request = true config.log_headers = true config.log_data = true config.log_status = true config.log_response = true config.log_benchmark = true config.compact_log = false # Log everything if environment variable is not specified config.url_whitelist_pattern = ENV['HTTPLOG_URL_WHITELIST_PATTERN'] || /.*/ config.url_blacklist_pattern = nil end

I recommend to put this in your config/initializers/01_httplog.rb. Make sure (e.g., with 01_) that it is one of the first initializers so you don't pollute your logs during deployment and server initialization.

Sometimes at Tumblr, I have to do some work with long running web requests, and sometimes catching them in the logs can be difficult. We run HAProxy with httplog turned on, and as you can guess, we get quite a bit of traffic running through these.

So I just wrote a small awk one liner to help me jump through these logs and quickly find requests that were greater than 1 second.

Check it out.