[M9] HTTP API server crash takes down the entire operator

[pradeepto]

Severity: Medium

Category

Reliability

Files

cmd/main.go:256-261

Description

The API server runs in a goroutine. If apiServer.Start() fails (port conflict, bind error, etc) it calls os.Exit(1) from that goroutine. This bypasses controller-runtime's shutdown sequence and can leave leader election leases held, blocking other replicas from taking over.

Fix

Signal the manager to shut down gracefully instead of calling os.Exit directly. Use the manager's context cancellation or a shared error channel.