[rabbitmq-discuss] badarg in dict, fetch problem on basic.consume_ok (or not)
blake-r at mail.ru
Wed Jan 15 11:15:40 GMT 2014
No, all queues was created on same channel which closed with
connection, with exclusive=true. And...
$ rabbitmqctl list_queues -p poker name messages_ready
messages_unacknowledged messages consumers
shows that there are no queues with messages or without consumer.
At this night whiout sleep I'm downgrade erlang (from
erlang-solutions.com R16B to native Ubuntu R14B) and now management
plugin hangs more quickly, but there are no errors in log at all, and
about badarg too. Memory grow, but for a while can't qualify reason -
may be R14B have other allocations (because number of clients now also
grow). At least it grow not so quickly as prior to.
2014/1/15 Matthias Radestock <matthias at rabbitmq.com>:
> On 15/01/14 10:53, Blake-R wrote:
>> OK. But as I'm wrote with graphics server still serve other requests -
>> just with memory grow. I'm looked at list connections/queues and
>> others with rabbitmqctl - it all in good values, not overflow (I'm
>> already has bug with non-closing channels and fix them).
> The illegal command sequence will result in the Erlang process representing
> the AMQP channel dying - which is what the error report you posted
> indicates. That in turn will cause the associated connection to be closed,
> which too is shown in the error report.
> Other than that rabbit will continue to function normally. The memory growth
> you are seeing is presumably due to messages accumulating in queues since
> their consuming channels have crashed due to the above problem.
More information about the rabbitmq-discuss