[rabbitmq-discuss] Shovel restart command hanging

[Tim Watson]

On 28/08/2012 10:57, Tim Watson wrote:
> On 27/08/2012 23:47, Elizabeth Liao wrote:
>> Hi,
>>
>> Currently we are running Rabbitmq 2.8.2 and are running into some
>> problems with a shovel configuration.  We were running into the shovel
>> wedge issues that was mentioned in another discussion on this mailing
>> list. It appeared to be sometimes resolved by the command: rabbitmqctl
>> eval 'application:stop(rabbitmq_shovel),
>> application:start(rabbitmq_shovel).'  However, other sometimes,
>> executing this command would hang.
>>
>
> Are you talking about the issue discussed in
> http://rabbitmq.1065348.n5.nabble.com/Shovel-stops-receiving-acks-from-cluster-td21384.html?
>

Er, I was a bit trigger happy with the send button there. I *meant* to 
say that as per 
http://rabbitmq.1065348.n5.nabble.com/Excessive-memory-consumption-of-one-server-in-cluster-setup-td21617.html, 
that bug appears to have nothing to do with using the shovel plugin.

If this is the issue you're facing, I'm actually working on a fix at the 
moment, and Matthias mentioned in 
http://lists.rabbitmq.com/pipermail/rabbitmq-discuss/2012-August/022133.html 
that it might be possible to work around this by using `rabbitmqctl 
set_vm_memory_high_watermark` to temporarily disable producers (by 
setting a very low value and make the broker apply back pressure) before 
restarting a node.