feat: broadcast queue state

[bithckr]

Resolves: #13

Start decomposing the pipeline into a Prepare step and and an Execute step. During prepare we backup and broadcast term/index. During Execute we recover the data.

[dlawregiets]

I haven't quite caught up on reading all of the code yet to answer these questions but I have a few.

It looks like you get the vhost from the log message. Is it documented that the rabbit user that phorcys uses needs to have access to all vhosts so it can publish everywhere it needs to?

The cfg.AMQPURL I assume has the users credentials baked in.

Do the log messages when a queue gets corrupted all go to the address on the root vhost? Is there any way to configure that so those messages go to a non-root vhost so clients on the root vhost can't read these messages?

Does phorcys use it's own vhost for message isolation or the root one?

[bithckr]

I haven't quite caught up on reading all of the code yet to answer these questions but I have a few.

It looks like you get the vhost from the log message. Is it documented that the rabbit user that phorcys uses needs to have access to all vhosts so it can publish everywhere it needs to?

The cfg.AMQPURL I assume has the users credentials baked in.

Do the log messages when a queue gets corrupted all go to the address on the root vhost? Is there any way to configure that so those messages go to a non-root vhost so clients on the root vhost can't read these messages?

Does phorcys use it's own vhost for message isolation or the root one?

We get a list of queues from RabbitMQ along with the Vhost. We use that list to match log messages with an actual queue. The log messages are too inconsistent to parse. I will update the README in the next commit to indicate that
full access to all VHost is required. It would not be hard to add VHost support, but I think that becomes moot because of the log exchange.

Yes, cfg.AMQPURL does have the user/password baked in.

The log exchange is only available on the root vhost(/). This is not configurable.

Right now Phorcys uses the vhost configured by the AMQPURL, but as we continue development beyond the POC, we may need to rethink that.