Question

I’m working with the Twitter Gem and I’ve created a long running ruby task. I would like it to be able to handle common errors so I’m looking to build a list of those I should consider to protect against (for example the fail whale 500)

Here is the begin/end loop my code functions in:

Begin

# My (omitted) very long ruby task
# filled with Twitter API requests

rescue Errno::ENOENT
  sleep(5)
  logger.info "ENOENT error - attempting to retry"
  retry
rescue Errno::ETIMEDOUT
  sleep(5)
  logger.info " Operation timed out - attempting to retry"
  retry
rescue Errno::ECONNRESET
  sleep(5)
  logger.info "Connection reset by peer - attempting to retry"
  retry
end 

Can you think of any other Errors to protect and retry against? Is this a well structured way to handle errors? What are some design implementations I should consider?

Answer 1

Consider having a catch-all exception handler at the end that logs what kind of exception was encountered and re-raises it. Your script may fail the first time, but at least you’ll find out why.

begin

# My (omitted) very long ruby task
# filled with Twitter API requests

rescue Errno::ENOENT
  sleep(5)
  logger.info "ENOENT error - attempting to retry"
  retry
rescue Errno::ETIMEDOUT
  sleep(5)
  logger.info " Operation timed out - attempting to retry"
  retry
rescue Errno::ECONNRESET
  sleep(5)
  logger.info "Connection reset by peer - attempting to retry"
  retry
rescue # This rescues StandardError and its children
  sleep(5)
  # The next line is somewhat pseudocode, because I don't use logger
  logger.this_is_somewhat_bad "Somewhat bad exception #{$!.class} #{$!} happened - I'm giving up"
  raise
rescue Exception
  sleep(5)
  # The next line is somewhat pseudocode, because I don't use logger
  logger.omg_wtf_bbq "Really bad exception #{$!.class} #{$!} happened - I'm giving up"
  raise
end