Stabilizing Container Deployments in a Self-Hosted Stack
While deploying my Rails app using Kamal, I started seeing intermittent failures pulling containers from GitHub Container Registry (GHCR). Nothing catastrophic but just enough flakiness to erode trust in automation: occasional timeouts during docker pull inconsistent deploy reliability retries sometimes fixing it, sometimes not These are the kinds of issues worth fixing early to maintain both reliability and velocity. And once you stop trusting deploy automation, velocity drops fast. This is the “Hollywood” version for fun. The real work is quieter: logs, retries, and incremental fixes. ...