[rabbitmq-discuss] Web Management Plugin Sluggishness
simon at rabbitmq.com
Fri Oct 19 17:41:03 BST 2012
In 2.8.x (and prior), /api/nodes will make an RPC call to each node in
turn to get the node-specific stats. If one node is heavily loaded or
slow in responding for whatever reason this will mean /api/nodes will
take a long time to return.
This has changed on default already, for basically the reason you're
running into, and the nodes emit stats events for the management DB to
If you invoke:
rabbitmqctl eval '[timer:tc(fun () ->
rabbit_mgmt_external_stats:info(N), N end) || N <-
(all one line)
then rabbitmqctl will make all these RPC calls and time them (the times
are in microseconds). That might let you know if you have a problem node.
On 19/10/12 17:26, Dave Seltzer wrote:
> Digging a little bit deeper, the symptom I'm seeing is that it's taking
> a while to load key json data from the http API.
> Specifically: _http://queue01:55672/api/nodes_ is taking more than 3
> seconds to load.
> Is anyone aware of what would cause this slowness?
> On Fri, Oct 19, 2012 at 11:47 AM, Dave Seltzer <dseltzer at tveyes.com
> <mailto:dseltzer at tveyes.com>> wrote:
> We have a number of RabbitMQ clusters in our environment. Recently
> I’ve started rolling out some new HA Clusters running RabbitMQ 2.8.7
> on Erlang R15B02 in Centos.
> In the most recent version of RabbitMQ we’re seeing some significant
> sluggishness in the Web Management Plugin. It almost seems like it’s
> Is anyone else seeing this?
> We have introduced one other change which is that I’ve switched from
> HAProxy to Zen Load Balancer. I don’t think this has anything to do
> with it though, because when I test against an individual server I
> still get sluggishness.
> Dave Seltzer <mailto:dseltzer at tveyes.com>
> Chief Systems Architect
> (203) 254-3600 x222
> rabbitmq-discuss mailing list
> rabbitmq-discuss at lists.rabbitmq.com
More information about the rabbitmq-discuss