Backgroundrb devel - Mar 2008 - Cleaing up after workers in backgroundrb

Hi there--

I just updated to the latest build of Backgroundrb and am hitting up
against a memory leak because my workers aren''t cleaning up.  My code
concurrently pulls a large number of images using RMagick.  I thought
the problem was with RMagick but after putting in some garbage
collection code when pulling an image, my code runs well from irb.
When wrapped with backgroundrb, it eventually hangs.

In the old backgroundrb I had a self.delete at the end of do_work to
clean up.  Now my setup is a little different:

MiddleMan.worker(:context_worker).process_context(id)

and then

class ContextWorker < BackgrounDRb::MetaWorker

  set_worker_name :context_worker
  pool_size 1


  def create(args = nil)

  end

  def process_context(id)

     do_stuff()

  end

end

How do I clean up after process_context().

thanks


-- 
Mike Bukhin
Interactive Telecommunications Program - ''07

On Tue, 2008-03-04 at 16:27 -0500, mike bukhin wrote:
> Hi there--
> 
> I just updated to the latest build of Backgroundrb and am hitting up
> against a memory leak because my workers aren''t cleaning up.  My
code
> concurrently pulls a large number of images using RMagick.  I thought
> the problem was with RMagick but after putting in some garbage
> collection code when pulling an image, my code runs well from irb.
> When wrapped with backgroundrb, it eventually hangs.
> 
> In the old backgroundrb I had a self.delete at the end of do_work to
> clean up.  Now my setup is a little different:
> 
> MiddleMan.worker(:context_worker).process_context(id)
> 
> and then
> 
> class ContextWorker < BackgrounDRb::MetaWorker
> 
>   set_worker_name :context_worker
>   pool_size 1
> 
> 
>   def create(args = nil)
> 
>   end
> 
>   def process_context(id)
> 
>      do_stuff()
> 
>   end
> 
> end
> 
> How do I clean up after process_context().
> 
You can still call, ''exit'' at the end of worker to finish up
execution
of task.

And what you mean by ''cleanup'' actually? RMagick is one heck
of a
library to work with, but if it works at irb prompt, should work in
BackgrounDRb worker too. Send us the worker code, and we will see whats
going wrong.

On Thu, 2008-03-06 at 13:00 +0530, hemant kumar wrote:
> On Tue, 2008-03-04 at 16:27 -0500, mike bukhin wrote:
> > Hi there--
> > 
> > I just updated to the latest build of Backgroundrb and am hitting up
> > against a memory leak because my workers aren''t cleaning up. 
My code
> > concurrently pulls a large number of images using RMagick.  I thought
> > the problem was with RMagick but after putting in some garbage
> > collection code when pulling an image, my code runs well from irb.
> > When wrapped with backgroundrb, it eventually hangs.
> > 
> > In the old backgroundrb I had a self.delete at the end of do_work to
> > clean up.  Now my setup is a little different:
> > 
> > MiddleMan.worker(:context_worker).process_context(id)
> > 
> > and then
> > 
> > class ContextWorker < BackgrounDRb::MetaWorker
> > 
> >   set_worker_name :context_worker
> >   pool_size 1
> > 
> > 
> >   def create(args = nil)
> > 
> >   end
> > 
> >   def process_context(id)
> > 
> >      do_stuff()
> > 
> >   end
> > 
> > end
> > 
> > How do I clean up after process_context().
> > 
> 
> You can still call, ''exit'' at the end of worker to finish
up execution
> of task.
> 
> And what you mean by ''cleanup'' actually? RMagick is one
heck of a
> library to work with, but if it works at irb prompt, should work in
> BackgrounDRb worker too. Send us the worker code, and we will see whats
> going wrong.
> 
> 
> _______________________________________________
> Backgroundrb-devel mailing list
> Backgroundrb-devel at rubyforge.org
> http://rubyforge.org/mailman/listinfo/backgroundrb-devel
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
http://rubyforge.org/pipermail/backgroundrb-devel/attachments/20080306/95941546/attachment.html

I just remember that in the older version of backgroundrb I was having
memory/performance issues until I added self.delete to do_work.  Then
those problems went away.  My worker is very simple:

class ContextWorker < BackgrounDRb::MetaWorker

  set_worker_name :context_worker
  pool_size 5


  def create(args = nil)

  end

  def process_context(context_id)
    require ''context_update''
    @r_update = ContextUpdate.new(Time.now)
    @r_update.load_public_context(context_id)
  end

end


The bulk of the code is in libraries.  So you''re saying I self.delete
is the same as putting exit after
@r_update.load_public_context(context_id)?

thanks!

On Thu, Mar 6, 2008 at 2:30 AM, hemant kumar <gethemant at gmail.com>
wrote:
>
>
>  On Tue, 2008-03-04 at 16:27 -0500, mike bukhin wrote:
>  > Hi there--
>  >
>  > I just updated to the latest build of Backgroundrb and am hitting up
>  > against a memory leak because my workers aren''t cleaning up.
My code
>  > concurrently pulls a large number of images using RMagick.  I thought
>  > the problem was with RMagick but after putting in some garbage
>  > collection code when pulling an image, my code runs well from irb.
>  > When wrapped with backgroundrb, it eventually hangs.
>  >
>  > In the old backgroundrb I had a self.delete at the end of do_work to
>  > clean up.  Now my setup is a little different:
>  >
>  > MiddleMan.worker(:context_worker).process_context(id)
>  >
>  > and then
>  >
>  > class ContextWorker < BackgrounDRb::MetaWorker
>  >
>  >   set_worker_name :context_worker
>  >   pool_size 1
>  >
>  >
>  >   def create(args = nil)
>  >
>  >   end
>  >
>  >   def process_context(id)
>  >
>  >      do_stuff()
>  >
>  >   end
>  >
>  > end
>  >
>  > How do I clean up after process_context().
>  >
>
>  You can still call, ''exit'' at the end of worker to
finish up execution
>  of task.
>
>  And what you mean by ''cleanup'' actually? RMagick is one
heck of a
>  library to work with, but if it works at irb prompt, should work in
>  BackgrounDRb worker too. Send us the worker code, and we will see whats
>  going wrong.
>
>
>


-- 
Mike Bukhin
Interactive Telecommunications Program - ''07

On Thu, 2008-03-06 at 08:43 -0500, mike bukhin wrote:
> I just remember that in the older version of backgroundrb I was having
> memory/performance issues until I added self.delete to do_work.  Then
> those problems went away.  My worker is very simple:
> 
> class ContextWorker < BackgrounDRb::MetaWorker
> 
>   set_worker_name :context_worker
>   pool_size 5
> 
> 
>   def create(args = nil)
> 
>   end
> 
>   def process_context(context_id)
>     require ''context_update''
>     @r_update = ContextUpdate.new(Time.now)
>     @r_update.load_public_context(context_id)
>   end
> 
> end
> 
> 
> The bulk of the code is in libraries.  So you''re saying I
self.delete
> is the same as putting exit after
> @r_update.load_public_context(context_id)?
> 
As far as I remember self.delete used to delete/exit the current worker.
"exit" does the same thing in newer version.