[rabbitmq-discuss] slave_nodes and synchronised_slave_nodes from HTTP API - Something's odd:

[Simon MacMullen]

On 25/01/12 17:37, Matt Pietrek wrote:
> Following on to an earlier thread
> (http://lists.rabbitmq.com/pipermail/rabbitmq-discuss/2012-January/017146.html)
> (Thanks Simon!), I've found something interesting.

Hmm...

> In one case (the "VMMunger" queue below), the returned 'slave_nodes'
> field is empty, while the 'synchronised_slave_nodes' field is
> 'rabbit at play2'. In other words, exactly the *opposite* of what you'd
> expect if the queue was not in sync. The strangeness is further
> corroborated by the copy/paste results of the Queues tab in the Web
> management UI, which reports that the VMMunger queue *is* in sync:

Well, the UI is just basing that on what it got from the HTTP API. In 
general synchronised_slave_nodes should be a subset of slave_nodes.

However the mgmt plugin gets the data from different sources, so you may 
have found a bug. What does rabbitmqctl show in this situation? (It goes 
via a different route.) Can you reproduce it?

Cheers, Simon

-- 
Simon MacMullen
RabbitMQ, VMware