Where failover meets your application
Bloodraven's job ends at a label selector and a DNS record; your application's job starts there. This unit closes the gap Unit 3 opened — the promotion that worked perfectly while the counter app went on serving stale reads from the demoted site and failing every write against it, with nothing paging for either. You will fix a connection pool, move a primary on purpose at an RPO of zero, decide what your cache and sessions owe you, and give the deploy pipeline a lease instead of the cluster. Everything so far has assumed somebody can see the truth about playground; Unit 5 removes that assumption.
Topics
Unit test
Quick check: can you say why a pooled connection keeps serving stale reads after a correct promotion, name the one path that actually drains connections, and read an absent dragonfly.sessions without guessing, and say what a deployment hold defers and what it never touches?
Project
Make the writer survive
Instrument a writer against playground, run both a planned and an emergency failover underneath it, and produce a drill record with the measured write-gap for each — so the recovery time you claim for your application is one you observed rather than one you assumed.