[rabbitmq-discuss] Recovering a busted instance

Wed May 22 10:44:54 BST 2013

Hi François.

The stack trace you're seeing is consistent with ones we've seen before 
after something causes a bunch of extra zeroes to be written to the end 
of the queue index.

Currently we're not absolutely sure what causes this but suspect the OS 
not completing a write during shutdown.

The good news is that we have a fix for reading queue indices in this 
state. The bad news is that it can't go into the 3.1.x series since it 
requires a change to the on disc format for the queue index.

However, this fix now is in the nightly releases (and will be in 3.2.0 
eventually). So I think you should be able to get a nightly build to 
boot with that data store. I'd take a backup first though :-) Once you 
do, you won't be able to go back to 3.1.x, but you should be able to get 
messages out to a 3.1.x broker with the shovel.

Hope this helps.

Cheers, Simon

On 21/05/13 18:25, François Beausoleil wrote:
> I have an instance which I tried upgrading from 3.0.2 to 3.1.0.
> Something broke during the upgrade and I booted a fresh new instance
> and connected my clients to it.
>
> Now, I have valuable data in the 3.0.2 instance which I'd like to
> retrieve. If I could boot the instance, I'd shovel the data to the
> 3.1.0 instance.
>
> I have attached the log files from the crashing instance.
>
> Is there anything I can do to recuperate those messages?
>
> Thanks, François Beausoleil
>
>
>
>
>
>
>
> _______________________________________________ rabbitmq-discuss
> mailing list rabbitmq-discuss at lists.rabbitmq.com
> https://lists.rabbitmq.com/cgi-bin/mailman/listinfo/rabbitmq-discuss
>

-- 
Simon MacMullen
RabbitMQ, Pivotal