[rabbitmq-discuss] Mirrored Queues Disappearing

Plock, Justin (GE Corporate) Justin.Plock at ge.com
Tue Jun 11 16:54:32 BST 2013


We unfortunately also run our logs at the warning level due to the amount of connection closing information our load balancer puts in there. 

That's what all of the connection closures you're probably seeing is from.

-Justin

On Jun 11, 2013, at 10:35, "Emile Joubert" <emile at rabbitmq.com> wrote:

> 
> Hi,
> 
> On 11/06/13 15:02, Plock, Justin (GE Corporate) wrote:
>> Could a network partition may have caused this?  I guess I have two main questions:
> 
> It is possible that a partition could cause less logging on some nodes
> and that could end up looking like a gap. Access to the logs may help,
> even if there is a gap.
> 
>> 1.) If my queues/exchange are set to HA, why do they continually get lost during network partitions?
> 
> That shouldn't happen. From the sasl logfiles provided it is clear that
> you are dealing with the effects of a bug. We are trying to replicate
> the bug, and access to the logfiles will probably help. Please feel free
> to contact me directly to arrange transfer, if that will help.
> 
>> 2.) Would using autoheal handle this scenario better?
> 
> If the root cause relates to partition healing then a different
> partition strategy may fail to trigger the bug. Please let us know if
> that is the case.
> 
> 
> We will be grateful for any information that you can provide about how
> to replicate it. How reliably can you replicate the bug? You mentioned
> network partitions - do queues get lost each time a network partition
> occurs? Does changing the partition handling strategy make any difference?
> 
> 
> 
> 
> 
> -Emile
> 
> 
> 
> 


More information about the rabbitmq-discuss mailing list