Back to updates

Add startup probe guidance to Kubernetes deployment docs

Pending review
by Sarah KimJun 30, 2026

2

Total

0

Changes

New documentDeployment Guide20 lines
New documentTroubleshooting20 lines
No text changes to review.
Deployment Guide

Startup Probes

If pods enter CrashLoopBackOff after deployment, the liveness probe may be failing before the application finishes starting.

Symptoms

Pods cycle between Running and CrashLoopBackOff
kubectl describe pod shows: "Liveness probe failed: HTTP probe failed with statuscode: 503"
The app takes longer to start than the liveness probe's initialDelaySeconds

Solution

Add a startupProbe to your deployment spec:

startupProbe:

httpGet:

path: /healthz

port: 8080

failureThreshold: 30

periodSeconds: 10

This allows up to 5 minutes for startup (30 failures × 10s interval). Once the startup probe succeeds, Kubernetes switches to the regular liveness probe.

When to use startupProbe vs initialDelaySeconds

startupProbe: Preferred for apps with variable startup times. Kubernetes will not run liveness/readiness checks until the startup probe succeeds.
initialDelaySeconds: Use when startup time is predictable and consistent. Simpler but less flexible — if startup takes longer than expected, the pod will be killed.