[rabbitmq-discuss] Monitoring Message Throughput

[Matthew Sackman]

On Sat, Feb 06, 2010 at 04:40:02PM -0500, Gavin M. Roy wrote:
> On Sat, Feb 6, 2010 at 4:17 PM, Matthias Radestock <matthias at lshift.net>wrote:
> 
> > By "throughput" do you mean:
> > - rate at which messages are enqueued?
> 
> For me, yes
> 
> > - rate at which messages are dequeued?
> 
> For me, yes

Ironically enough, in the new persister, we actually use these rates
(ingress and egress rates) internally to make various decisions.
Currently, you can get hold of them (if you're using the new persister -
branch bug21673) by the magic name "raw_vq_status". Note this is
undocumented and will not last too much longer - it exists for debugging
purposes only. However:

rabbitmqctl list_queues name messages messages_unacknowledged durable memory raw_vq_status
testA   3503    0       true    6657424  [{q1,0},{q2,0},{delta,{delta,undefined,0,undefined}},{q3,0},{q4,3503},{len,3503},{outstanding_txns,0},{target_ram_msg_count,undefined},{ram_msg_count,3503},{ram_index_count,0},{avg_egress_rate,0},{avg_ingress_rate,0},{next_seq_id,3503}]

the two fields there you're interested in are the avg_egress_rate and
avg_ingress_rate. It's a rolling average, smoothed over the last two
sample periods, where a period length is nominally 5 seconds.

Matthew