Reputation: 12732
I currently have a Rails application that has multiple processes: the web serving processes and the background workers, that are triggered by Redis.
Problem is sometimes is hard to check the log files and determine where a given behavior happened - was it on the Web portion or on the Resque workers?
Is there a way to include the process name or even process id or something that allows me to differentiate each log entry by process?
Upvotes: 5
Views: 3574
Reputation: 8247
If you need process ID outside of a controller context (e.g. delayed job) you can put this in an initializer:
class ActiveSupport::BufferedLogger
def formatter=(formatter)
@log.formatter = formatter
end
end
class Formatter
SEVERITY_TO_COLOR_MAP = {'DEBUG'=>'0;37', 'INFO'=>'32', 'WARN'=>'33', 'ERROR'=>'31', 'FATAL'=>'31', 'UNKNOWN'=>'37'}
def call(severity, time, progname, msg)
formatted_severity = sprintf("%-5s","#{severity}")
formatted_time = time.strftime("%Y-%m-%d %H:%M:%S.") << time.usec.to_s[0..6].ljust(6)
color = SEVERITY_TO_COLOR_MAP[severity]
"\033[0;37m#{formatted_time} (pid:#{$$})\033[0m [\033[#{color}m#{formatted_severity}\033[0m] #{msg.strip}\n"
end
end
Rails.logger.formatter = Formatter.new
More here: http://www.software-thoughts.com/2013/08/adding-process-id-and-timestamps-to.html
Original post here: http://cbpowell.wordpress.com/2012/04/05/beautiful-logging-for-ruby-on-rails-3-2/
Upvotes: 2
Reputation: 8247
It looks like there are a few options out there for this:
config.log_tags = [:subdomain, :uuid, :remote_ip, Proc.new { "PID-%.5d" % $$ }]
(which the previous link says is slow)config.log_tags = [Proc.new { "PID: %.5d" % Process.pid }]
Here is a related SO article: - Rails 3.2.2 log files unordered, requests intertwined
The best bet to me seems to be to use :uuid instead. It conveys the same information to let you distinguish between requests when you have multiple processes logging to the same file.
Upvotes: 5