Test DNS cutover on development environment

[neilmb]

Feature/what we're after

We want to know how fast we can switch from CKAN to Catalog-beta and back. Try this on the development environment and take notes on the right way to make the switch and how long it takes

Anticipated/hypothesized benefits

• Confidence!!

Measurements/metrics

• number of times we have successfully switched and switched back

References/background

• For production, we need to consider the interaction with Cloudfront distributions and Cloud.gov-provided Cloudfront.

[SueValente]

Dry run that allows us to do #5839

[FuhuXia]

Dryrun with catalog-dev.data.gov on development space. The actual cutover did not involve DNS change, only route mapping change on cloud.gov environment. With the following cf commands, I can make catalog-dev alternate content between old catalog app and new catalog app within minutes.

# reads datagov-catalog-app
cf map-route datagov-catalog-proxy catalog-dev.data.gov --app-protocol http1
cf unmap-route catalog-proxy catalog-dev.data.gov
invalidate cache on aws console

# reads catalog-app
cf map-route catalog-proxy catalog-dev.data.gov --app-protocol http1
cf unmap-route datagov-catalog-proxy catalog-dev.data.gov
invalidate cache on aws console

changing cloudfront origin between catalog-dev-datagov.app.cloud.gov and datagov-catalog-dev.app.cloud.gov seems has no effect and not need. I assume that is because it relies on hostname catalog-dev.data.gov to route traffic to the right app.