Redis client had a 35s worst-case stall (10s read/write timeout x3
retries) on a single call, which under concurrent load exhausted the
connection pool and cascaded into a full outage. Timeouts and retries
are now tight enough that a degraded Redis fails fast instead of
tying up pooled connections.
Also fixes two userid/appuserid column mix-ups in delivery queries
(deliveryController.go, domain/delivery.go) and points the Dockerfile
build at main.go explicitly.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
36eb8cd194