Backgroundrb devel - Dec 2007 - Get "some read error" on calls to worker

I''m running the latest from svn (rev 285). I''ve been having a
problem
launching a process repeatedly. What happens is that I get a debug
message "some read error" and then the worker refuses to run again.

 

I''ve stripped my worker down to just doing a puts and it still happens.
Like the other threads I want to launch my workers as needed, however I
get the same behavior whether use the auto_load and ask_work or use
new_worker with no_auto_load.  If I remove the job_key I can get it to
work, but I still get the "Some read error" message. I need the job to
keep one worker per id. I''d rather have separate processes than go the
thread route - for now.

 

class UpdateFacebookAccountInfoWorker < BackgrounDRb::MetaWorker

  set_worker_name :update_facebook_account_info_worker

  set_no_auto_load true

  

  require "facebook_rails_controller_extensions"

  include RFacebook  

  

  require ''activemessaging/processor''

  include ActiveMessaging::MessageSender

 

  def create(args)

    puts "create=>"+ args.inspect

    exit!

  end

end

 

and the call...

 

      MiddleMan.new_worker( 

        :worker => :update_facebook_account_info_worker,

        :job_key => @some.id, # Only run one worker per id

        :data => { 

          :session_key => @session_key,

          :requesting_fb_id => @organizer.facebook_id, 

          :facebook_ids => @fb_ids,

          :params => params

        }

      )

 

The output...

 

create=>{:params=>REDACTED]}

Some read error

Some read error

Some read error

 

Thanks

Rob

-------------- next part --------------
An HTML attachment was scrubbed...
URL:
http://rubyforge.org/pipermail/backgroundrb-devel/attachments/20071217/a2b14747/attachment-0001.html

Hi Rob,
On Mon, 2007-12-17 at 11:47 -0500, Rob Nielsen wrote:
> I?m running the latest from svn (rev 285). I?ve been having a problem
> launching a process repeatedly. What happens is that I get a debug
> message ?some read error? and then the worker refuses to run again.
> 
>  
> 
> I?ve stripped my worker down to just doing a puts and it still
> happens. Like the other threads I want to launch my workers as needed,
> however I get the same behavior whether use the auto_load and ask_work
> or use new_worker with no_auto_load.  If I remove the job_key I can
> get it to work, but I still get the ?Some read error? message. I need
> the job to keep one worker per id. I?d rather have separate processes
> than go the thread route ? for now.
> 
>  
> 
> class UpdateFacebookAccountInfoWorker < BackgrounDRb::MetaWorker
> 
>   set_worker_name :update_facebook_account_info_worker
> 
>   set_no_auto_load true
> 
>   
> 
>   require "facebook_rails_controller_extensions"
> 
>   include RFacebook  
> 
>   
> 
>   require ''activemessaging/processor''
> 
>   include ActiveMessaging::MessageSender
> 
>  
> 
>   def create(args)
> 
>     puts "create=>"+ args.inspect
> 
>     exit!
> 
>   end
> 
> end
> 
>  
> 
> and the call?
> 
>  
> 
>       MiddleMan.new_worker( 
> 
>         :worker => :update_facebook_account_info_worker,
> 
>         :job_key => @some.id, # Only run one worker per id
> 
>         :data => { 
> 
>           :session_key => @session_key,
> 
>           :requesting_fb_id => @organizer.facebook_id, 
> 
>           :facebook_ids => @fb_ids,
> 
>           :params => params
> 
>         }
> 
>       )
> 
>  
> 
> The output?
> 
>  
> 
> create=>{:params=>REDACTED]}
> 
> Some read error
> 
> Some read error
> 
> Some read error
> 
>  
Ok so, create method is getting invoked when you invoke:

MiddleMan.new_worker(...)

right?

Now, next i see an "exit!" statement there, which basically asks the
worker to quit. Now when a worker quits, master process gets an error
when reading from that socket, so thats why you are getting "some read
error" message. But as far as I can see, create method is getting called
alright. Correct me, If i am wrong.



> 
-- 
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

-----Original Message-----
From: hemant kumar [mailto:gethemant at gmail.com] 
Sent: Monday, December 17, 2007 12:14 PM
To: Rob Nielsen
Cc: backgroundrb-devel at rubyforge.org
Subject: Re: [Backgroundrb-devel] Get "some read error" on calls to
worker

Hi Rob,


[SNIP]

Ok so, create method is getting invoked when you invoke:

MiddleMan.new_worker(...)

right?

Now, next i see an "exit!" statement there, which basically asks the
worker to quit. Now when a worker quits, master process gets an error
when reading from that socket, so thats why you are getting "some read
error" message. But as far as I can see, create method is getting called
alright. Correct me, If i am wrong.

----

Yes it gets called once. Subsequent calls fail for some reason. If I
take out the job_key, then I can call it again.

On Mon, 2007-12-17 at 12:34 -0500, Rob Nielsen wrote:
> 
> -----Original Message-----
> From: hemant kumar [mailto:gethemant at gmail.com] 
> Sent: Monday, December 17, 2007 12:14 PM
> To: Rob Nielsen
> Cc: backgroundrb-devel at rubyforge.org
> Subject: Re: [Backgroundrb-devel] Get "some read error" on calls
to
> worker
> 
> Hi Rob,
> 
> 
> [SNIP]
> 
> Ok so, create method is getting invoked when you invoke:
> 
> MiddleMan.new_worker(...)
> 
> right?
> 
> Now, next i see an "exit!" statement there, which basically asks
the
> worker to quit. Now when a worker quits, master process gets an error
> when reading from that socket, so thats why you are getting "some read
> error" message. But as far as I can see, create method is getting
called
> alright. Correct me, If i am wrong.
> 
> ----
> 
> Yes it gets called once. Subsequent calls fail for some reason. If I
> take out the job_key, then I can call it again.
When you can exit! on a worker..its dead, its gone. You can''t access it
any more. You shouldn''t call exit! on it, if you want to make further
calls to it.
> 
-- 
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

-----Original Message-----
From: hemant kumar [mailto:gethemant at gmail.com] 
Sent: Monday, December 17, 2007 12:36 PM
To: Rob Nielsen
Cc: backgroundrb-devel at rubyforge.org
Subject: RE: [Backgroundrb-devel] Get "some read error" on calls to
worker


On Mon, 2007-12-17 at 12:34 -0500, Rob Nielsen wrote:
> 
> -----Original Message-----
> From: hemant kumar [mailto:gethemant at gmail.com] 
> Sent: Monday, December 17, 2007 12:14 PM
> To: Rob Nielsen
> Cc: backgroundrb-devel at rubyforge.org
> Subject: Re: [Backgroundrb-devel] Get "some read error" on calls
to
> worker
> 
> Hi Rob,
> 
> 
> [SNIP]
> 
> Ok so, create method is getting invoked when you invoke:
> 
> MiddleMan.new_worker(...)
> 
> right?
> 
> Now, next i see an "exit!" statement there, which basically asks
the
> worker to quit. Now when a worker quits, master process gets an error
> when reading from that socket, so thats why you are getting "some read
> error" message. But as far as I can see, create method is getting
called
> alright. Correct me, If i am wrong.
> 
> ----
> 
> Yes it gets called once. Subsequent calls fail for some reason. If I
> take out the job_key, then I can call it again.
When you can exit! on a worker..its dead, its gone. You can''t access it
any more. You shouldn''t call exit! on it, if you want to make further
calls to it.

----
Right. But I''m calling new_worker. So that should start a new instance.
Right?
> 
-- 
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

On Mon, 2007-12-17 at 12:39 -0500, Rob Nielsen wrote:
> 
> -----Original Message-----
> From: hemant kumar [mailto:gethemant at gmail.com] 
> Sent: Monday, December 17, 2007 12:36 PM
> To: Rob Nielsen
> Cc: backgroundrb-devel at rubyforge.org
> Subject: RE: [Backgroundrb-devel] Get "some read error" on calls
to
> worker
> 
> 
> On Mon, 2007-12-17 at 12:34 -0500, Rob Nielsen wrote:
> > 
> > -----Original Message-----
> > From: hemant kumar [mailto:gethemant at gmail.com] 
> > Sent: Monday, December 17, 2007 12:14 PM
> > To: Rob Nielsen
> > Cc: backgroundrb-devel at rubyforge.org
> > Subject: Re: [Backgroundrb-devel] Get "some read error" on
calls to
> > worker
> > 
> > Hi Rob,
> > 
> > 
> > [SNIP]
> > 
> > Ok so, create method is getting invoked when you invoke:
> > 
> > MiddleMan.new_worker(...)
> > 
> > right?
> > 
> > Now, next i see an "exit!" statement there, which basically
asks the
> > worker to quit. Now when a worker quits, master process gets an error
> > when reading from that socket, so thats why you are getting "some
read
> > error" message. But as far as I can see, create method is getting
> called
> > alright. Correct me, If i am wrong.
> > 
> > ----
> > 
> > Yes it gets called once. Subsequent calls fail for some reason. If I
> > take out the job_key, then I can call it again.
> 
> When you can exit! on a worker..its dead, its gone. You can''t
access it
> any more. You shouldn''t call exit! on it, if you want to make
further
> calls to it.
> 
> ----
> Right. But I''m calling new_worker. So that should start a new
instance.
> Right?
> 
> > 
Ok, sync with latest svn code and see if its fixed. If not then, come to
#backgroundrb on freenode.



-- 
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

On Mon, 2007-12-17 at 14:48 -0500, Rob Nielsen wrote:
> 
> -----Original Message-----
> From: hemant kumar [mailto:gethemant at gmail.com] 
> Sent: Monday, December 17, 2007 1:09 PM
> To: Rob Nielsen
> Cc: backgroundrb-devel at rubyforge.org
> Subject: RE: [Backgroundrb-devel] Get "some read error" on calls
to
> worker
> 
> 
> On Mon, 2007-12-17 at 12:39 -0500, Rob Nielsen wrote:
> > 
> > -----Original Message-----
> > From: hemant kumar [mailto:gethemant at gmail.com] 
> > Sent: Monday, December 17, 2007 12:36 PM
> > To: Rob Nielsen
> > Cc: backgroundrb-devel at rubyforge.org
> > Subject: RE: [Backgroundrb-devel] Get "some read error" on
calls to
> > worker
> > 
> > 
> > On Mon, 2007-12-17 at 12:34 -0500, Rob Nielsen wrote:
> > > 
> > > -----Original Message-----
> > > From: hemant kumar [mailto:gethemant at gmail.com] 
> > > Sent: Monday, December 17, 2007 12:14 PM
> > > To: Rob Nielsen
> > > Cc: backgroundrb-devel at rubyforge.org
> > > Subject: Re: [Backgroundrb-devel] Get "some read error"
on calls to
> > > worker
> > > 
> > > Hi Rob,
> > > 
> > > 
> > > [SNIP]
> > > 
> > > Ok so, create method is getting invoked when you invoke:
> > > 
> > > MiddleMan.new_worker(...)
> > > 
> > > right?
> > > 
> > > Now, next i see an "exit!" statement there, which
basically asks the
> > > worker to quit. Now when a worker quits, master process gets an
> error
> > > when reading from that socket, so thats why you are getting
"some
> read
> > > error" message. But as far as I can see, create method is
getting
> > called
> > > alright. Correct me, If i am wrong.
> > > 
> > > ----
> > > 
> > > Yes it gets called once. Subsequent calls fail for some reason.
If I
> > > take out the job_key, then I can call it again.
> > 
> > When you can exit! on a worker..its dead, its gone. You can''t
access
> it
> > any more. You shouldn''t call exit! on it, if you want to make
further
> > calls to it.
> > 
> > ----
> > Right. But I''m calling new_worker. So that should start a new
> instance.
> > Right?
> > 
> > > 
> 
> Ok, sync with latest svn code and see if its fixed. If not then, come to
> #backgroundrb on freenode.
> 
> ----
> OK that seems to have fixed the issue. Thanks.
> 
> A couple questions: 
> 1. What happens if you do a new_worker with a given job_key while a
> worker with the same key is already running? Does the second task get
> queued or does the request simply wait for the first process to finish?
Currently none of the above happens. Your request will be simply
ignored.
> 
> 2. Does ask_status throw an exception if the worker isn''t running?
I
> would like to be able to ask if the worker is running before starting
> another one.
Nope, It won''t throw any exception in user code. It does throw an
exception internally. However, you can query status of all workers that
are currently running, using 

  MiddleMan.query_all_workers()

See, if that output is helpful.

-- 
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

-----Original Message-----
From: hemant kumar [mailto:gethemant at gmail.com] 
Sent: Monday, December 17, 2007 3:09 PM
To: Rob Nielsen
Cc: backgroundrb-devel
Subject: RE: [Backgroundrb-devel] Get "some read error" on calls to
worker

[SNIP]
> 
> 2. Does ask_status throw an exception if the worker isn''t running?
I
> would like to be able to ask if the worker is running before starting
> another one.
Nope, It won''t throw any exception in user code. It does throw an
exception internally. However, you can query status of all workers that
are currently running, using 

  MiddleMan.query_all_workers()

See, if that output is helpful.

----
I do get an exception: BackgrounDRb::BdrbConnError when I do the
following and the worker is not running.

worker_status = MiddleMan.ask_status(:worker =>
:update_facebook_account_info_worker, :job_key => @id)

I don''t mind using a begin rescue block if that is the way to verify
that the worker is not running. Is there a better way?

P.S. I''m on IRC now but no one seems to be answering (I''m an
IRC noob -
sorry).

-- 
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