Should the _replicator database have 0 documents?

Innteresting... I am loading the audio prompt attachment from fauxton and couchdb is returning a 401 even though I specified the user. I only mess with couchdb on a very limited basis. Shouldn't I be able to download the wav attachment by calling it directly via curl while specifying the database admin credentials?

Ysean: yes, it stores it as an attachment inside couchdb

If you're getting 200 OK'd on one DB node, and 404 on another, there's something (possibly) wrong with your replication?

To get the attachment via couch directly, you need to GET /db/doc/attachmentName. eg; /06/bb/{accountid}/{documentid}/vm-enter_new_pin.wav

for system media, you need to make sure prompts are set up: sup kazoo_media_maintenance import_prompts /opt/kazoo/sounds/en/us/

doc relating to system prompts: docs.2600hz.com/kazoo_dev/doc/internationalization/prompts

Moosable, yup.. I did all that.

That's why I suspect a replication issue as well. Can someone give me an example of what the _replication should look like?

er _replicator

well, that's interesting. It seems I never actually configured the bigcouch nodes to cluster on this instance.

GAH

s/bigcouch/couchdb

Question... it slooks like the local.ini configs weren't updated for the cluster config... can I leave q=1 while setting r=2, w=2 and n=3 without negatively impacting the databases?

I have a feeling I'm going to be feeling some pain to make this right no matter what I do.

q=1 should be fine

Unless you suddenly need 12 nodes and you want to fragment the db across nodes for performance

But you can always set up a new cluster and copy/replicate data into the new cluster