Health Probes

🏥 Health Probes: Your Container’s Doctor

The Story of the Sleepy Restaurant

Imagine you own a restaurant. Every morning, you need to know three things:

1. Is the chef alive? (Can they breathe and move?)
2. Is the chef ready to cook? (Have they warmed up the kitchen?)
3. Did the chef just wake up? (Are they still getting ready?)

Kubernetes asks these same questions about your containers. These questions are called Health Probes.

🔍 What Are Health Probes?

Health Probes are like doctors that check on your containers.

They ask: “Are you okay? Can you work?”

If the container says “No,” Kubernetes helps fix the problem.

graph TD
    A["🐳 Container Running"] --> B{Health Check}
    B -->|Healthy| C["✅ Keep Running"]
    B -->|Unhealthy| D["🔄 Take Action"]

💓 Liveness Probes: “Are You Alive?”

The Simple Idea

A Liveness Probe checks if your container is still alive.

Think of it like this:

• A doctor checks if a patient is breathing
• If not breathing → Give CPR (restart)
• If breathing → Everything is fine

What Happens When It Fails?

Kubernetes restarts the container. Like rebooting a frozen computer.

Real Example

livenessProbe:
  httpGet:
    path: /healthz
    port: 8080
  initialDelaySeconds: 3
  periodSeconds: 5

What this says:

• Check the /healthz page every 5 seconds
• Wait 3 seconds before the first check
• If the page doesn’t respond, restart me

When Do You Need This?

Your app is stuck in a loop? It crashed but didn’t stop? Liveness Probe catches this.

🚦 Readiness Probes: “Are You Ready to Work?”

The Simple Idea

A Readiness Probe checks if your container can handle requests.

Think of it like this:

• A restaurant opens at 9 AM
• But the chef needs time to prep
• Don’t send customers until food is ready!

What Happens When It Fails?

Kubernetes stops sending traffic to this container. It doesn’t restart—just pauses.

Real Example

readinessProbe:
  httpGet:
    path: /ready
    port: 8080
  initialDelaySeconds: 5
  periodSeconds: 10

What this says:

• Check /ready every 10 seconds
• Wait 5 seconds before first check
• If not ready, don’t send me any work

When Do You Need This?

Your app needs to load data first? Connect to a database? Readiness Probe handles this.

🐣 Startup Probes: “Are You Done Starting?”

The Simple Idea

A Startup Probe gives slow-starting containers extra time.

Think of it like this:

• Some people wake up fast
• Others need 30 minutes and coffee
• Startup Probe is the snooze button

Why Is This Special?

Old apps take forever to start. Without Startup Probes, Liveness checks might kill them too early!

Real Example

startupProbe:
  httpGet:
    path: /healthz
    port: 8080
  failureThreshold: 30
  periodSeconds: 10

What this says:

• Try 30 times, every 10 seconds
• That’s 5 minutes to start up!
• After startup succeeds, hand off to Liveness Probe

When Do You Need This?

Legacy apps? Big data loads at startup? Startup Probe protects them.

⚙️ Probe Configuration: The Settings

Every probe has settings you can adjust. Like tuning a radio.

The Key Settings

Complete Example

livenessProbe:
  httpGet:
    path: /healthz
    port: 8080
  initialDelaySeconds: 10
  periodSeconds: 5
  timeoutSeconds: 3
  successThreshold: 1
  failureThreshold: 3

Translation:

• Wait 10 seconds, then start checking
• Check every 5 seconds
• Wait max 3 seconds for a reply
• 1 success = healthy
• 3 failures in a row = unhealthy

🔧 Probe Handlers: Three Ways to Check

Kubernetes can check health in three ways. Pick what works for your app.

1. HTTP Handler (Most Common)

Ask a web page: “Are you there?”

httpGet:
  path: /healthz
  port: 8080
  httpHeaders:
    - name: Custom-Header
      value: Awesome

✅ Best for: Web apps, APIs, anything with HTTP

2. TCP Handler

Just knock on the door: “Is anyone home?”

tcpSocket:
  port: 3306

✅ Best for: Databases, message queues, non-HTTP services

3. Exec Handler

Run a command inside the container.

exec:
  command:
    - cat
    - /tmp/healthy

✅ Best for: Custom checks, file-based health, scripts

graph TD
    A["Pick a Handler"] --> B["HTTP"]
    A --> C["TCP"]
    A --> D["Exec"]
    B --> E["Web Apps & APIs"]
    C --> F["Databases"]
    D --> G["Custom Scripts"]

⚔️ Liveness vs Readiness: The Big Difference

This is where people get confused. Let’s make it crystal clear.

Side-by-Side Comparison

The Restaurant Analogy

Liveness: Is the chef conscious?

• ❌ No → Call an ambulance (restart)
• ✅ Yes → Great!

Readiness: Can the chef take orders?

• ❌ No → Don’t seat customers
• ✅ Yes → Open for business!

When to Use Each

graph TD
    A["Container Started"] --> B{Startup Probe}
    B -->|Pass| C["Liveness + Readiness Active"]
    C --> D{Liveness Check}
    D -->|Fail| E["🔄 Restart Container"]
    D -->|Pass| F{Readiness Check}
    F -->|Fail| G["📵 No Traffic"]
    F -->|Pass| H["✅ Receive Traffic"]

Golden Rule

Liveness = “Should I restart?” Readiness = “Should I send traffic?”

🎯 Putting It All Together

Here’s a complete example with all three probes:

apiVersion: v1
kind: Pod
metadata:
  name: my-app
spec:
  containers:
  - name: app
    image: my-app:1.0
    ports:
    - containerPort: 8080

    startupProbe:
      httpGet:
        path: /healthz
        port: 8080
      failureThreshold: 30
      periodSeconds: 10

    livenessProbe:
      httpGet:
        path: /healthz
        port: 8080
      initialDelaySeconds: 0
      periodSeconds: 10
      failureThreshold: 3

    readinessProbe:
      httpGet:
        path: /ready
        port: 8080
      initialDelaySeconds: 0
      periodSeconds: 5
      failureThreshold: 3

What happens:

1. Startup Probe gives 5 minutes to start
2. Once started, Liveness checks every 10 seconds
3. Readiness checks every 5 seconds
4. Container receives traffic only when ready

🌟 Key Takeaways

1. Liveness = Is it alive? (Restart if dead)
2. Readiness = Is it ready? (No traffic if busy)
3. Startup = Is it done booting? (Extra time for slow apps)
4. HTTP/TCP/Exec = Three ways to check health
5. Configure wisely = Tune delays and thresholds for your app

You now understand Health Probes! Your containers will thank you for keeping them healthy. 🏥✨