Some questions:

Does the test bugzilla5.redhat.com instance have a queue we can test against in our staging?

Do we know what the new queue name is yet for prod?

Where do we request these acls?

I think @jfearn will have to answer the first two. I don't know.

The ACLs can be requested via a ticket to Matt Stratton's team in RHIT. Search on mojo for the UMB Getting Started page which describes how to get acls for a new cert. You'll need a similar ticket to upgrade acls for an existing cert. Failing that - email Matthew directly.

Some questions:
Does the test bugzilla5.redhat.com instance have a queue we can test against in our staging?

Only the internal dev and qe servers are talking to the message bus ATM. I'll poke our Ops and see if they can set it up.

Do we know what the new queue name is yet for prod?

It's not a queue, it's a virtual topic. /topic/VirtualTopic.eng.bugzilla

Where do we request these acls?

AIUI anyone who has access to the bus can read virtual topics.

It's not a queue, it's a virtual topic. /topic/VirtualTopic.eng.bugzilla

Yeah, if bugzilla will be publishing to the virtual topic: /topic/VirtualTopic.eng.bugzilla then consumers have to read from a queue of their own that gets dynamically created when they try to read from it. Something like /queue/Consumer.client-fedora.PUT-A-UNIQUE-UUID-HERE.VirtualTopic.eng.bugzilla

AIUI anyone who has access to the bus can read virtual topics.

That's usually true, but since the fedora client account was created before the days when we had virtual topics, I bet they're going to need to request access to even read. Maybe, maybe.

FYI the date for the Bugzilla 5 upgrade is Sep 7th 2018.

I've requested acls for our cert now. Once thats done we should be able to switch queues and be done.

I set the stomp queue to be used in staging only at: https://infrastructure.fedoraproject.org/cgit/ansible.git/commit/?id=644d5d5

It won't work until the acls are granted in @kevin's request above.

Note to self - bz5 went live. The topic probably changed, and is breaking us.

I'll try to follow up on this later today.

Yeah, I changed the queue to the new one, but it's giving a permission denied (even though we had been granted permissions months ago).

So not sure why its not just working...

Hm, I'll ping the UMB team asking.

Looks like the problem was on our side. Two commits, fixing:

• https://infrastructure.fedoraproject.org/cgit/ansible.git/commit/?id=864a7f66836805b0e7813cf71502daa0f705a545
• https://infrastructure.fedoraproject.org/cgit/ansible.git/commit/?id=62701210b43fc226fb37c4cdd9793330332f17ea

Hm, messages seem to not be coming through. Raised this with @yulwang on the bz5 side.

Looking further into this. Messages are making it to our system but getting lost in software. May need a small bugzilla2fedmsg change. I'm on it. :)

Turning into a bigger change than I expected. Shutting down the process for now.

https://apps.fedoraproject.org/datagrepper/raw?category=bugzilla&rows_per_page=1&delta=127800

PR posted at https://github.com/fedora-infra/bugzilla2fedmsg/pull/10

That PR has been applied on the bugzilla2fedmsg VM. Leaving this open to track merge and release of that change.

This is all done.

:goat: