Bug #9115

Logger traps all exceptions; breaks Timeout

Added by Chris Phoenix over 2 years ago. Updated about 1 year ago.

ruby -v:
ruby 2.0.0p247 (2013-06-27) [i386-mingw32]


Line 577-579 of logger.rb

  rescue Exception => ignored
    warn("log writing failed. #{ignored}")

Thus, when the system times out in the middle of writing a log message, it warns "log writing failed. execution expired" and just keeps right on running.

This is true in 1.9.3 as well. I haven't looked at older versions.

Pardon me while I go grep "rescue Exception" in the entire Ruby codebase, and see whether I can reliably use Timeout at all...

OK, you might check out C:\Ruby200\lib\ruby\gems\2.0.0\gems\activerecord-3.2.13\lib\active_record\railties\databases.rake

All the other "rescue Exception" seem to re-raise it, except maybe C:\Ruby200\lib\ruby\2.0.0\xmlrpc\server.rb and C:\Ruby200\lib\ruby\gems\2.0.0\gems\activesupport-3.2.13\lib\active_support\callbacks.rb


#1 [ruby-core:58385] Updated by Chris Phoenix over 2 years ago

...I guess my search was broken. I did another and found a whole lot of "rescue exception" that don't seem to always re-raise. But the logger is a big one because it's supposed to be usable everywhere, and someone who logs a lot will run into this bug.

#2 [ruby-core:64346] Updated by Aaron Stone almost 2 years ago

The behavior has changed in Ruby 2.1, and the default rescue no longer rescues Timeout unless Timeout was given an explicit exception class to raise. Could you try Ruby 2.1 and see if it behaves according to your expectation here?

#3 [ruby-core:64753] Updated by Bill Paulson almost 2 years ago

The source code for 2.1.2 doesn't seem to have changed: the LogDevice::write method still does a rescue Exception and ignores the result.
Our application ran into this problem when one thread raises an exception in another thread that's doing a LogDevice::write(). The exception is supposed to be handled at a higher level, so write() rescuing it and ignoring it causes our application to hang.
A write() method without this problem looks like

  def write(message)
    @mutex.synchronize do
      if @shift_age and @dev.respond_to?(:stat)
        rescue IOError
          warn("log shifting failed. #{$!}")
      rescue IOError
        warn("log writing failed. #{$!}")

#4 [ruby-core:69940] Updated by Stefano Tortarolo about 1 year ago

I confirm that this bug is reproducible up to 2.2 and it's pretty easy to trigger.

require 'timeout'
require 'logger'

class MyExc < RuntimeError; end

def long_call(logger)
  Timeout.timeout(3, MyExc) do
    a = 1
    loop do
      a += 1
      logger.warn "Test #{a}"

Aaron Stone is right though that using 2.1+ it would work if you don't provide a custom exception or if your exception inherits from TimeoutError.

Beware that it must inherit from Object::TimeoutError and not Timeout::Error.

Having said that, it'd be nice to have a proper fix at logger.rb level.

@Bill, I don't think your code would fix it though, because the underlying device might throw a different exception.

#5 [ruby-core:69979] Updated by Stefano Tortarolo about 1 year ago

Correction: providing an exception to Timeout#timeout would break it also in trunk.

It seems that by design inner blocks are able to rescue the exception provided to that method.

Also available in: Atom PDF