Node pools and auto scaling in GCP - Cheat Sheet & Quick Revision

Recall & Review

beginner

What is a node pool in Google Kubernetes Engine (GKE)?

A node pool is a group of nodes within a GKE cluster that share the same configuration, such as machine type and labels. It allows you to manage and scale sets of nodes independently.

beginner

How does auto scaling work in GKE node pools?

Auto scaling automatically adjusts the number of nodes in a node pool based on the workload demand. It adds nodes when more resources are needed and removes nodes when demand decreases.

intermediate

Why use multiple node pools in a GKE cluster?

Using multiple node pools lets you run different types of workloads with different machine types or configurations. For example, one pool for CPU-intensive tasks and another for memory-intensive tasks.

intermediate

What is the difference between cluster autoscaler and horizontal pod autoscaler?

Cluster autoscaler adjusts the number of nodes in the cluster (node pools), while horizontal pod autoscaler adjusts the number of pods (containers) running inside the cluster based on CPU or custom metrics.

beginner

What happens if auto scaling is disabled on a node pool?

If auto scaling is disabled, the node pool size stays fixed. You must manually add or remove nodes, which can lead to wasted resources or insufficient capacity during workload changes.

What does a node pool in GKE represent?

A A group of nodes with the same configuration

B A single pod running in the cluster

C A storage bucket for logs

D A network firewall rule

A node pool is a set of nodes with the same machine type and settings in a GKE cluster.

What triggers the cluster autoscaler to add nodes?

A When a pod is deleted

B When pods need more resources than current nodes provide

C When a new node pool is created

D When the cluster is idle

The cluster autoscaler adds nodes when pods cannot be scheduled due to lack of resources.

Which of these is NOT a benefit of using multiple node pools?

A Automatically scaling pods inside nodes

B Isolating workloads for security or compliance

C Running different workloads with optimized resources

D Using different machine types in the same cluster

Scaling pods is done by horizontal pod autoscaler, not by node pools.

If auto scaling is disabled, what must you do to change node count?

A Delete the node pool

B Wait for automatic scaling

C Restart the cluster

D Manually add or remove nodes

Without auto scaling, node count changes require manual intervention.

What is the main difference between cluster autoscaler and horizontal pod autoscaler?

A Both scale nodes only

B Cluster autoscaler changes pods; pod autoscaler changes nodes

C Cluster autoscaler changes nodes; pod autoscaler changes pods

D Both scale pods only

Cluster autoscaler adjusts node count; horizontal pod autoscaler adjusts pod count.

Explain what a node pool is and why it is useful in GKE.

Think about how you organize workers with similar tools for specific tasks.

Describe how auto scaling works for node pools and why it helps manage resources.

Imagine a restaurant adding or removing tables depending on how many customers arrive.

Practice

1. What is the main purpose of a node pool in a GKE cluster?

easy

A. To group similar machines for easier management and scaling

B. To store data permanently in the cluster

C. To monitor network traffic between nodes

D. To create backups of the cluster configuration

Solution

1. Step 1: Understand node pool concept

   A node pool is a set of nodes with the same configuration in a cluster.

2. Step 2: Identify node pool purpose

   Node pools help manage and scale groups of similar machines easily.

3. Final Answer:

   To group similar machines for easier management and scaling -> Option A

4. Quick Check:

   Node pool = group similar machines [OK]

Hint: Node pools group similar nodes for scaling and management [OK]

Common Mistakes:

• Confusing node pools with storage or backup features
• Thinking node pools monitor network traffic
• Assuming node pools store data permanently

2. Which of the following is the correct way to enable auto scaling on a node pool using gcloud CLI?

easy

A. gcloud container clusters create CLUSTER_NAME --autoscaling

B. gcloud container node-pools update POOL_NAME --enable-autoscaling --min-nodes=1 --max-nodes=5

C. gcloud container node-pools create POOL_NAME --autoscaling-enabled

D. gcloud container node-pools set-autoscale POOL_NAME --min=1 --max=5

Solution

1. Step 1: Recall gcloud syntax for node pool auto scaling

   The correct command uses 'node-pools update' with '--enable-autoscaling' and min/max nodes.

2. Step 2: Match options with correct syntax

   gcloud container node-pools update POOL_NAME --enable-autoscaling --min-nodes=1 --max-nodes=5 matches the correct syntax and flags for enabling auto scaling.

3. Final Answer:

   gcloud container node-pools update POOL_NAME --enable-autoscaling --min-nodes=1 --max-nodes=5 -> Option B

4. Quick Check:

   Enable autoscaling with update and min/max nodes [OK]

Hint: Use 'node-pools update' with --enable-autoscaling and min/max nodes [OK]

Common Mistakes:

• Using cluster create command to enable node pool autoscaling
• Incorrect flag names like --autoscaling-enabled
• Commands that do not specify min and max nodes

3. Given a node pool with auto scaling enabled (min nodes=2, max nodes=4), what happens when workload increases beyond current capacity?

medium

A. The node pool scales up by adding nodes up to the max limit

B. The node pool scales down to reduce costs

C. The cluster deletes existing nodes to balance load

D. The workload is rejected until manual scaling is done

Solution

1. Step 1: Understand auto scaling behavior on workload increase

   Auto scaling adds nodes when workload exceeds current capacity, up to max nodes.

2. Step 2: Analyze options

   The node pool scales up by adding nodes up to the max limit correctly describes scaling up; others describe incorrect or opposite behaviors.

3. Final Answer:

   The node pool scales up by adding nodes up to the max limit -> Option A

4. Quick Check:

   Workload increase triggers scale up [OK]

Hint: Auto scaling adds nodes when workload grows, up to max nodes [OK]

Common Mistakes:

• Thinking auto scaling scales down on workload increase
• Assuming nodes are deleted to balance load
• Believing workload is rejected without manual scaling

4. You configured auto scaling on a node pool but notice nodes are not scaling up despite high CPU usage. What is the most likely cause?

medium

A. Auto scaling is disabled on the node pool

B. The node pool has no nodes initially

C. The cluster is in maintenance mode

D. The max nodes limit is set too low

Solution

1. Step 1: Check auto scaling settings

   If auto scaling is enabled but nodes don't increase, max nodes limit may be reached.

2. Step 2: Identify cause of no scale up

   The max nodes limit is set too low explains that max nodes limit is too low to add more nodes despite workload.

3. Final Answer:

   The max nodes limit is set too low -> Option D

4. Quick Check:

   Max nodes limit blocks scaling up [OK]

Hint: Check max nodes limit if scaling doesn't happen [OK]

Common Mistakes:

• Assuming auto scaling is disabled without checking limits
• Ignoring max nodes limit as a cause
• Thinking cluster maintenance mode blocks scaling always

5. You want to optimize costs by using multiple node pools with different machine types and enable auto scaling only on the pool handling variable workloads. Which approach is best?

hard

A. Disable auto scaling on all node pools and manually adjust nodes daily

B. Use a single node pool with auto scaling enabled for all workloads

C. Create a small fixed node pool for stable workloads and a separate auto scaling node pool for variable workloads

D. Create multiple auto scaling node pools with the same machine type

Solution

1. Step 1: Understand workload types and node pools

   Stable workloads benefit from fixed nodes; variable workloads benefit from auto scaling.

2. Step 2: Choose best cost optimization strategy

   Create a small fixed node pool for stable workloads and a separate auto scaling node pool for variable workloads separates stable and variable workloads into different pools with appropriate scaling.

3. Final Answer:

   Create a small fixed node pool for stable workloads and a separate auto scaling node pool for variable workloads -> Option C

4. Quick Check:

   Separate pools for stable and variable workloads optimize cost [OK]

Hint: Use fixed pool for stable, auto scaling pool for variable workloads [OK]

Common Mistakes:

• Using single auto scaling pool for all workloads
• Disabling auto scaling and relying on manual scaling
• Creating multiple auto scaling pools with same machine type unnecessarily