Backgroundrb Usage in Rails App

Rails
1

Hey Guys, ’

I am developing a rails application for a client. Suffice it to say I
have to do some long running tasks so of course I used my favorite
plugin backgroundrb. Unfortunately I have a problem. I recently put some
new work into a background task that requires the use of ruburple (a gem
for connecting to IM protocols). Since then, I noticed that the worker
that uses it sometimes just stops running with no explanation.

I checked the logs and I assume the relevant code is below:

libnm_glib_nm_state_cb: dbus returned an error.
(org.freedesktop.DBus.Error.ServiceUnknown) The name
org.freedesktop.NetworkManager was not provided by any .service files

** (process:13909): CRITICAL **: xmlnode_set_attrib: assertion `value !=
NULL’ failed

** (process:13909): CRITICAL **: xmlnode_set_attrib: assertion `value !=
NULL’ failed

** (process:13909): CRITICAL **: xmlnode_set_attrib: assertion `value !=
NULL’ failed
#{my_rails_app}/releases/…/vendor/plugins/backgroundrb/framework/core.rb:149:
[BUG] Segmentation fault
ruby 1.8.6 (2007-06-07) [i486-linux]

My question is two-fold:

  1. what’s the best way to go about fixing this? How can I get more
    detailed information regarding the problem here. I assume the
    combination of the CRITICAL and the always fun segmentation fault is
    killing the thread. How can I find and eliminate the problem or
    implement exception handling so it ignores the issue.

  2. How can I just force the thread back up if it dies. There has to be a
    way to have backgroundrb recreate the thread if it dies, right? I mean
    sure, that segmentation fault shouldn’t happen but if it does, I need a
    way to bring the thread back online.

So if you could help with that I would appreciate it. I know ezra pokes
around here every once in a while.

2

BTW the app is running on

ruby 1.8.6 (2007-06-07) [i486-linux]
rails 2.0.1
ubuntu 7.10
gems are updated

and the weird part is that the “segmentation fault” doesn’t happen every
time. Most of the time it works just fine. However 1/5 times the fault
will occur and the thread dies.

3

BTW, the line that it references for the segmentation fault:

backgroundrb/framework/core.rb:149

method starts event loop in the process

  def start_reactor
    Signal.trap("TERM") { terminate_me }
    Signal.trap("INT") { shutdown }
    loop do
      check_for_timer_events
      user_thread_window #=> let user level threads run for a while
      ready_fds = select(@read_ios,@write_ios,nil,0.005) # HERE IT

FAILS!!!
#next if ready_fds.blank?

      next if !ready_fds or ready_fds.empty?

      ready_fds = ready_fds.flatten.compact
      ready_fds.each do |t_sock|
        if t_sock.is_a? UNIXSocket
          handle_internal_messages(t_sock)
        else
          handle_external_messages(t_sock)
        end
      end
    end
  end
4

On Fri, Jan 25, 2008 at 5:18 PM, Nathan E.
[email protected] wrote:

    Signal.trap("INT") { shutdown }
      ready_fds.each do |t_sock|
        if t_sock.is_a? UNIXSocket
          handle_internal_messages(t_sock)
        else
          handle_external_messages(t_sock)
        end
      end
    end
  end

By any chance, can you run this thing with gdb? Also segmentation
fault is one error from which process itself can’t recover and must be
restarted from outside. You can have a monit script for that.

–
Let them talk of their oriental summer climes of everlasting
conservatories; give me the privilege of making my own summer with my
own coals.

http://gnufied.org