[rabbitmq-discuss] badarg in dict, fetch problem on basic.consume_ok (or not)

[Олег Росс]

This is how it is look in cacti (data from collectd - curl_json plugin):

<https://lh3.googleusercontent.com/-1GVErJK51R4/UtZR9W0XdBI/AAAAAAAAADs/kSTC_re4MH0/s1600/rabbitmq1.png>

<https://lh5.googleusercontent.com/-Crq2XN9Op5o/UtZSINHW3uI/AAAAAAAAAD0/UXqX1BIq9w0/s1600/rabbitmq2.png>

<https://lh4.googleusercontent.com/-_uLqxiSFY0A/UtZSqaQUIBI/AAAAAAAAAD8/UB0TXZBY9oY/s1600/rabbitmq3.png>

Last try of run (right most, pair of hours) is 
with collect_statistics_interval: 10000 and force_find_statistics: false: 
memory not grown, but management still become inaccessible. :(

On early memory grow peaks our server was stop responding to clients 
through big timeouts. Still no error in logs other than from my first post.

Oleg.

вторник, 14 января 2014 г., 23:25:18 UTC+4 пользователь Олег Росс написал:
>
> Hello.
>
> Some strange error we have sometimes (1-2 days or 1-2 weeks - 
> differently), see attachement.
>
> After this error memory consumption begin grow until watermark reached - 
> then rabbitmq start to pressure messages and our clients gave timeout.
>
> At the same time web management begin hanging on (HTTP requests never ends 
> without transmit any data).
>
> rabbitmq 3.2.2 (and 3.1.5)
> erlang R16B
>
> Oleg.
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.rabbitmq.com/pipermail/rabbitmq-discuss/attachments/20140115/11af148a/attachment.html>