[rabbitmq-discuss] RabbitMQ crashed

Kapil Goyal goyalk at vmware.com
Thu Nov 7 21:10:20 GMT 2013


Hi Michael,

Is this issue fixed in 3.2.1? I noted this:

25838 prevent crashes due to timeouts when calling into the limiter
      (since 3.1.0)

If not, is there a way to track the progress on the issue or a timeline
you could provide?

Thanks
Kapil

-----Original Message-----
From: Michael Klishin [mailto:mklishin at gopivotal.com] 
Sent: Thursday, October 24, 2013 2:39 PM
To: Kapil Goyal
Cc: Simon MacMullen; Discussions about RabbitMQ
Subject: Re: [rabbitmq-discuss] RabbitMQ crashed

On 25 Oct 2013, at 01:29, Kapil Goyal <goyalk at vmware.com> wrote:

>> Is this the only error message in the logs?
> There are many errors like this before the crash:
> =ERROR REPORT==== 21-Oct-2013::20:01:09 === webmachine error: 
> path="/api/connections"
> "Unauthorized"

This is a warning from HTTP API. Something tries to access
/api/connections without having permissions to do so. So, it's a red
herring.

> 
>> I don't think it took down the entire node, I expect the queue crashed.

>> Still unacceptable though.
> Error message says "Generic server <0.764.0> terminating". Does this 
> not indicate that node crashed?

Server here means gen_server, an Erlang process behavior (think of it as a
class named *Server, if it helps). It should not bring down the node.

> 
>> Kapil, this will be fixed in the next release.
> This is good news. But, our product is already shipped with RabbitMQ 
> 3.1.5 and upgrade won't be easy. Nonetheless, is this totally a server 
> issue, being hit because of high load, or can we do anything at the 
> client end to avoid it?

This is a server issue. From time to time one process does not respond to
another process in < 5 seconds and the caller crashes without a timeout.
Presumably this is because of short hardware resources.

There is no client-side work around I'm afraid.

MK

Software Engineer, Pivotal/RabbitMQ


More information about the rabbitmq-discuss mailing list