Shutdown Error in RabbitMq sasl Log

Question

We are running RabbitMq 3.6.2 with Erlang 17.4 on Windows Server 2008. We see the following record over and over in the sasl log. It is occurring sparratically (every few hours) with no particular pattern.

=SUPERVISOR REPORT==== 05-Jul-2016::09:52:42 === 
    Supervisor: {<0.835.211>,amqp_channel_sup_sup} 
    Context: shutdown_error 
    Reason: shutdown 
    Offender: [ {nb_children,1},
                {name,channel_sup}, 
                {mfargs, 
                    {amqp_channel_sup,start_link, 
                        [direct,<0.864.211>, <<"">>]}},
                {restart_type,temporary}, 
                {shutdown,brutal_kill}, 
                {child_type,supervisor}]

I read here that there may be a correlation to the aliveness test through the HTTP API from the admin plugin. We are also using the aliveness test from our load balancer to define which nodes should receive traffic.

Any help on understanding what the log actually means and how we can prevent it (if need be) would be greatly appreciated. I published this question on Server Fault as well here.

Mccrary

Answer 1

I don't know how much you know about Erlang/OTP, but:

Context: shutdown_error suggests there's something wrong (like throwing an exception) during the process's shutdown procedure

Reason: shutdown suggests that this process is finished normally, and is trying to gracefully shutdown before it crashes

The Offender is the child (start) specification of which is having shutdown_error and it lives under the supervisor specified by Supervisor.

Combined with your mention of the aliveness test. My first guess would be somehow the aliveness test code created some amqp channel to the RabbitMQ cluster but failed to gracefully shut down. Did a little bit of digging in rabbitmq-management code but haven't found anything yet.

Also this looks like a pretty old question, not sure whether it still affects you or someone else.

Empress

It affects me as well. I get this error daily almost in every hour on my rabbitmq clusters. – Croak

reading the following link, "Yup, that's the culprit. It's just straight closing the TCP socket which causes these warnings to be emitted into the logs. I know they say error, but you can safely ignore them." it sounds like an ignorable error. rabbitmq.1065348.n5.nabble.com/… – Hypostatize