[rabbitmq-discuss] Debugging rabbitmq crash
mvcalder at gmail.com
Fri Jun 25 14:19:27 BST 2010
So, I still get the same error with latest erlang and latest rabbitmq.
I sent a message yesterday that I think did not go through ( I got an
error message from the mail delivery system). Anyway that had the
stack trace from the two python apps doing the messaging. The server
process died thusly:
line 125, in write_frame
File "<string>", line 1, in sendall
socket.error: [Errno 32] Broken pipe
and the client:
line 215, in read_method
IOError: Socket closed
Rabbit is still running and I can just restart the pair of processes
and they work (thanks Matthias for the suggestion to check that).
I don't think rabbit is the culprit in this. I should have mentioned,
I am running the messaging through stunnel (it is all local to one
machine at the moment, but will be deployed to more machines once it
has been tested). So, I am thinking that may be part of the problem
(nest step, turn off stunnelling).
If anyone has further suggestions at tests I can run, or logs I can
inspect, let me know. Thanks for the help,
On Fri, Jun 25, 2010 at 8:56 AM, Matthew Sackman <matthew at rabbitmq.com> wrote:
> On Thu, Jun 24, 2010 at 07:19:29AM -0400, Matt Calder wrote:
>> The second run stopped in the same manner. I was running rabbitmq
>> 1.7.2 and erlang 13B03. I had started before I got Matthias's
>> suggestion to upgrade rabbit. This time I captured the stack traces
>> of the two processes when they stopped.
> Hmm, if Rabbit itself is in trouble then there should be entries in
> Rabbit's logs. It's possible it's a memory exhaustion issue, or it could
> be a client issue - there are many possibilities at this point. What
> information is returned by commands such as "rabbitmqctl list_queues"
> etc? The rabbit logs should be the most revealing though.
> How are you getting on with 1.8.0?
> rabbitmq-discuss mailing list
> rabbitmq-discuss at lists.rabbitmq.com
More information about the rabbitmq-discuss