We need a couple of things for this to work:

• fedmsg certs for the resultsdb machines (created already).
• a local firewall rule on our inbound fedmsg relay to let the resultsdb machines in (created already).
• a specialized config for fedmsg on the resultsdb boxes that tells it where to connect to.
• an RHIT firewall rule to let those 124 boxes talk to the relays on the 126 net.

nirik volunteered to file a ticket with RHIT asking for that.

I've filed the ticket, will let you know when it's done.

Which machines were part of that request? It seems silly not to add the production resultsdb machine to that request so we don't have to go through this again later

I did add both stg and prod to it.

This change has been completed. Please test. ;)

Got a test message through: https://apps.stg.fedoraproject.org/datagrepper/id?id=2015-2153588e-7684-4b08-a913-73dddb88fa3f&is_raw=true&size=extra-large

OK, and I added a config file for resultsdb in staging that should tell it to use the relay:

https://infrastructure.fedoraproject.org/cgit/ansible.git/commit/?id=453d9a007b6c7a3be28e82111422e87b0ee69357

Should be ready for testing. Lemme know where you hit problems and we can work through it in #fedora-apps.

Looks like this is working now. ;)