Percolator with rabbitmq river


(Steve Casey) #1

Hi, I'd like some advice/feedback on whether this scenario is something you
could do with ES & some customization of the rabbitmq river...

Is it possible to get the rabbitmq river to send messages to a percolator?
(i dont need the docs indexed)

Then when the percolator responds with a set of matching queries, I send
the matching queries with the initial document onto another rabbitmq queue?
(eg. if 1 doc matches 5 queries, then 5 msgs are pushed onto queue)

(let me know if an example is needed to better illustrate :))

Basically, using ES as a filtering engine for queue messages.

Is it a really bad idea? would it have severe limitations? (scale? too many
moving parts?) why?

Or is it awesome and I should get the rabbitmq river code forked now!?! :slight_smile:

Regards,
Steve


(Shay Banon) #2

Its possible to do with rabbitmq, but it does not feel like a "river". Just
write code that does it yourself in your own codebase, why introduce the
complexity of a river? If you still want to do a "river", I would create a
different type of river, not one that does both indexing and percolation.

On Sat, Jan 14, 2012 at 9:03 AM, Steve Casey sa.casey@gmail.com wrote:

Hi, I'd like some advice/feedback on whether this scenario is something
you could do with ES & some customization of the rabbitmq river...

Is it possible to get the rabbitmq river to send messages to a percolator?
(i dont need the docs indexed)

Then when the percolator responds with a set of matching queries, I send
the matching queries with the initial document onto another rabbitmq queue?
(eg. if 1 doc matches 5 queries, then 5 msgs are pushed onto queue)

(let me know if an example is needed to better illustrate :))

Basically, using ES as a filtering engine for queue messages.

Is it a really bad idea? would it have severe limitations? (scale? too
many moving parts?) why?

Or is it awesome and I should get the rabbitmq river code forked now!?! :slight_smile:

Regards,
Steve


(system) #3