[rabbitmq-discuss] [New project] amqp-utils, CLI utilities for AMQP queues

Fri Feb 6 15:46:54 GMT 2009

Valentino,

Thanks, that is a very useful contribution.  I'll ask Tony how hard it
would be to expose this using his JSON stuff.

alexis

On Thu, Feb 5, 2009 at 9:55 AM, Valentino Volonghi <dialtone at gmail.com> wrote:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
>
>
> On Feb 5, 2009, at 1:35 AM, Alexis Richardson wrote:
>
>> Thanks for the feedback on this.
>>
>> As Valentino and Al point out, keeping things compact is important,
>> and not all users are running or want to run a web framework with
>> RabbitMQ.  Though of course some will.
>>
>> So one option here is to use the built-in HTTP capability rather than
>> a web framework.  For example, to provide an interface to rabbitmqctl
>> that lets people use a browser to keep an eye on the broker.
>>
>> On the other hand people may prefer programmatic management.
>
>
> I'm pretty sure that with an HTTP server built-in you can do anything
> to rabbitmq, even writing consumers in javascript :).
>
> More seriously:
> An HTTP server that exposes some kind of simple REST API that
> can be queries also from javascript would mean that even a static
> page (hardly a match for a built-in http server) could be served
> efficiently and act as a monitoring screen.
>
> As for the REST API I'm thinking along the lines of:
>
> /stats/ <- general stats and list of exchanges
> /stats/<EXCHANGE>/ <- list of queues under exchange
> /stats/<EXCHANGE>/<QUEUE>/ <- queue stats
> /stats/users/
> /stats/consumers/
> /stats/acls/
> ...
>
> simple http-auth is more than enough if you really want to make
> this secure and could use the same users credentials that rabbitmq
> stores for the rest of the system. And the response format for each
> request would simply be JSON, then a single html page can be
> used to show all the stats that one might need. This has no memory
> overhead because everything is done on the client side and can be
> freely extended by simply changing javascript without the need to
> deploy a new web server somewhere (etc etc).
>
> Then by quering (polling at the beginning, comet when there's more
> time to develop something like this) a url multiple times one can build
> a chart of the stats (and if polling the built-in http server doesn't even
> need to know that it is happening).
>
> - --
> Valentino Volonghi aka Dialtone
> Now running MacOS X 10.5
> Home Page: http://www.twisted.it
> http://www.adroll.com
>
> -----BEGIN PGP SIGNATURE-----
> Version: GnuPG v1.4.9 (Darwin)
>
> iEYEARECAAYFAkmKt4sACgkQ9Llz28widGW65ACgtJTQCgE7ZYMLTFkuALRxSM+O
> pzwAnRlrN8uI6BS7vTEHKgOLFOzj1Lqd
> =7ifr
> -----END PGP SIGNATURE-----
>