Get started with Auto Scaling

Build a scale group of two nodes from your own saved image, and see its nodes listed with their addresses. The group adds a node when CPU use is high and removes one when it is low.

Auto Scaling is offered in Delhi and Chennai. In Delhi, an account created on or after May 15, 2026 needs an exemption from E2E before it can create a scale group. If that is your account, see Why is creating a scale group refused in Delhi? first.

You also need:

• A saved image in that region, in the Ready state, that was saved from a C3, M3, SDC3, E1, or E1WC node, or from the older C, C2, or M series. See Save a node as an image.
• On a prepaid account, enough infra credits to run two nodes on the plan you choose for 24 hours.

Every node in the scale group starts from this image, so save it from a node that already starts your application at boot. If your application runs as a systemd service, run sudo systemctl enable <service-name> on that node before you save the image, where <service-name> is the service’s name.

The nodes also keep the image’s SSH keys and root password, and you sign in to them with those.

Choose the image and the plan

1. In MyAccount, under Compute, click Auto Scaling. Manage Auto Scaling opens. Click Create New Scale Group, or Get Started if you have no scale group yet.

2. Under Select Saved Image, select your image. Only images in the Ready state are listed.

3. Under Select Plan, choose a plan and click Select. The list shows the plans your image can run on, and every node in the scale group gets the vCPU and RAM of the plan you pick. The plan is set once, when you create the scale group.

   From this step on, prices are shown per month and marked Hourly Billed, which means the nodes are billed by the minute at the hourly price. For current prices, see E2E’s pricing page.

Set up the scale group

1. Scale Group Name is filled in from the image name and a three-digit number. Keep it, or enter your own name, two to 40 characters long, using letters, digits, underscores, or hyphens. You choose the name once, when you create the scale group.
2. Under Scale group parameters, keep Minimum Node at 2 and Maximum Node at 5. The scale group starts with as many nodes as Minimum Node.
3. Under Select Policy, choose Elastic Policy, which changes the number of nodes as CPU use goes up and down.
4. Keep Policy Parameter Type on Default, with CPU as the parameter. Keep the starting values, which add a node when CPU use is more than 60% and remove one when it is less than 30%.
5. Under Security Group, your default security group is selected. Keep it, or choose one that allows the port your application listens on.
6. Check the Summary panel, and click Create Scale.

Back on the list, with Scale group creation in progress., your scale group reads Creating. The list refreshes itself every five seconds while the scale group is being created. Wait until it reads Running, which takes about three minutes in Delhi, or under a minute for a one-node scale group in Chennai. If it still reads Creating after that, see Troubleshooting Auto Scaling.

Check that the scale group runs its nodes

Click the scale group in the list to open its tabs, and open Active Node Details. It lists the two nodes running now, each with its Scale Node Name, Public IP, Private IP, and Status.

Each time CPU use stays above 60% for the watch period, the scale group adds a node, up to five. Each time CPU use stays below 30% for the watch period, it removes one, down to two. With the starting values, once a steady load goes above 60%, it starts adding a node after about 4.5 minutes, and the node takes about 2.5 minutes more to be added. The Audit Logs tab records each step. See Read the audit logs.

(Optional) Open your application on a node

When Active Node Details shows an address under Public IP, open this address in a browser, where <node-ip> is that address and <port> the port your application listens on:

http://<node-ip>:<port>/

When it shows N/A, open the same address with the node’s Private IP instead, from a machine in the scale group’s private network.

You will see your application, served by a node the scale group started from your image. If the page fails to load, check that the security group allows inbound traffic on <port>; see Add or edit rules.

(Optional) Put a load balancer in front

To give your users one address for all the nodes, attach the scale group to an Application Load Balancer as an Auto Scale Group backend. See Put a load balancer in front of a scale group.

Clean up

Each node is billed by the minute from when it starts until it is removed. When you have finished trying the scale group:

• To keep its settings for later, stop it. Stopping removes every node and anything stored on the nodes. While the scale group is stopped, you pay nothing for nodes. The saved image keeps its own charges. See Resize, stop, or start a scale group.
• To remove it completely, delete it. Deleting is permanent: it removes the scale group and all its nodes. See Delete a scale group.
• If you created a load balancer for the scale group, delete it too when you no longer need it. It is a separate product with its own charges.

What to do next

• Create a scale group: every choice on the create page, including a VPC and encryption.
• Set an elastic policy: tune the CPU values, the watch period, and the cooldown.
• Set a scheduled policy: scale up and down at the times you know are busy.
• Put a load balancer in front of a scale group: give your users one address for all its nodes.
• Monitor a scale group: follow the scale group’s status, its nodes, and each scaling step.
• Troubleshooting Auto Scaling and Auto Scaling FAQ: fixes for common problems, and short answers to common questions.