[rabbitmq-discuss] rabbitmq gets high load
tim at rabbitmq.com
Thu Jan 17 09:53:08 GMT 2013
On 17 Jan 2013, at 04:29, Doug wrote:
> PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND
> 17994 rabbitmq 20 0 11.6g 10g 1764 S 367 66.2 416834:07 beam.smp
> As you see, rabbitmq process has lots of system load (367% CPU, 66.2% memory).
Certainly looks like it. That level of CPU utilisation is a bit surprising.
> It even can't list the queue now:
> root at ubuntu:~# rabbitmqctl -p /pay list_queues
> Listing queues ...
> BREAK: (a)bort (c)ontinue (p)roc info (i)nfo (l)oaded
> (v)ersion (k)ill (D)b-tables (d)istribution
> So how to handle this case? Thanks.
You've not given us much to go on here. Normally I'd ask you to run the output of `rabbitmqctl report` but that's likely to lock up just as list_queues is doing. Can you post the regular as sasl rabbit logs please? If they're too bit to post, a /paste/ url will do fine and of course you may wish to scrub the logs to remove any private information if that's relevant.
Also what kind of /usage/ is this rabbit subject to at the time? Is it serving a large number of producers, consumers, both? Have you looked at any stats around the time this starts happening in order to see what the broker is up to?
More information about the rabbitmq-discuss