When deploys go wrong

Prerequisites

• You've deployed at least one Render service (web service, worker, or cron job)
• You're comfortable reading stack traces and tailing logs
• Render CLI v2.7.0+ helps for the log-inspection sections, but is optional

Steps

1. 01 When deploys go wrong Why deploys break, who this tutorial is for, and the shape of the method you'll use to diagnose every failure in the same way. 5 min →
2. 02 The diagnostic method A six-step loop that turns any deploy failure into a named problem with a known fix - reproduce, locate, read, hypothesise, test smallest, escalate. 8 min →
3. 03 Reading Render logs Where build, deploy, and runtime logs live, how to find the first error in 10 seconds, and the Render Dashboard + CLI workflows that turn a wall of text into a specific failure. 8 min →
4. 04 Build-time failures The most common build errors on Render - module not found, language version mismatches, lockfile drift, missing build commands - with annotated logs and the one-line fix for each. 10 min →
5. 05 Boot and port binding The 'build passed, deploy failed' family - port binding, host address, missing `CMD`, hanging start commands, and the 'No open ports detected' message everyone hits eventually. 9 min →
6. 06 Health checks and crashes When the port binds but the deploy still fails - health check timeouts, crash loops, OOM kills, signal handling, and the difference between a flapping service and a dead one. 9 min →
7. 07 Runtime errors When the deploy went green but the app is misbehaving in production - the 400/404/500/502 family, database SSL mismatches, ALLOWED_HOSTS, Node keep-alive, and the rules for telling 'platform' from 'application' errors apart. 9 min →
8. 08 The playbook Stitch the method, logs cheat sheet, and failure library into a printable flowchart and a copy-pasteable runbook entry for your team. 7 min →