On Thu, Jun 17, 2010 at 7:41 AM, Christian-Rolf Grün <[email protected]> wrote:
> I am currently evaluating beanstalkd as a message queue solution for
> an application.

Awesome! As others have noted many times before, you should stay open
to the possibility of auxiliary storage for mutable job state, such as
completion status. A SQL database or key-value store can do well here.

For beanstalkd-only solutions...

> I'd like to know if there are any built-in facilities
> for the worker to send back data to the job producer after a job is
> done.

Not yet, but see http://github.com/kr/beanstalkd/issues/issue/3

> Failing that, can the worker modify the data associated with a job?
> Can the producer alter the data of a buried job before kicking it?

No. Jobs are immutable.

> The last resort for sending data back would be the creation of reply
> tubes for each job. How expensive is it to create a tube?

That is a great strategy if you change it just a bit: create one reply
tube for each producer, rather than for each job.

It is cheap to create a tube, but performance with large numbers of
tubes (thousands or more) has not yet been optimized.

The "mailbox" tubes discussed in
http://github.com/kr/beanstalkd/issues/issue/3 might make it easier to
manage reply jobs in this strategy, depending on how you want to
organize things, but they aren't strictly necessary.

kr

-- 
You received this message because you are subscribed to the Google Groups 
"beanstalk-talk" group.
To post to this group, send email to [email protected].
To unsubscribe from this group, send email to 
[email protected].
For more options, visit this group at 
http://groups.google.com/group/beanstalk-talk?hl=en.

Reply via email to