Closed bookshelfdave closed 4 years ago
This is mostly done for the staging instance. Just prod to go as well as DNS for domains.
we should be fine pushing everything to prod before the DNS switch, right?
definitely. just haven't done it. I can push the prod branch and the k8s resources should be created.
Okay. Done. Seems to have worked. https://gitlab.com/mozmeao/www.firefox.com/-/jobs/560855599
what else do we need before we request the DNS cutover?
We should get some real internal domains for these. e.g. fxc-prod.moz.works
or fxc-stage.oregon-b.moz.works
. After that we're going to hook the prod one up to CloudFlare I think (issue #1309). Then we'll be ready for DNS. I'd like to do some load testing before we cut over as well though.
I'm going to aim for this pattern for the alb domain name:
<svc>-<env>.<cluster>moz.works. fxc-stage.oregon-b.moz.works/ fxc-prod.oregon-b.moz.works.
Then, point cloudflare at a fxc-prod.moz.works, which will temporarily just be a cname to fxc-prod.oregon-b.moz.works, but when we add a second region would point to a traffic policy, that can do healthchecks and geo-ip stuff like bedrock.
@pmac anything wrong with the above? I think this matches bedrock's pattern today.
looks good to me 👍
Should be all done! 🎉
fxc-prod.moz.works
orfxc-stage.oregon-b.moz.works
(following existing pattern in route53)