Scaling

Learn to scale a ReplicaSet with kubectl scale.

Scaling

A ReplicaSet is also scaled with kubectl scale. But if you directly scale a ReplicaSet that belongs to a Deployment, the Deployment immediately undoes it, since the desired state is in the Deployment. Scaling is always done through the top-level owner:

kubectl scale rs/web-7c5b9f6d8 --replicas=10    # the Deployment pulls it back to 3
kubectl scale deploy/web --replicas=10          # the right way

Conditional scaling

--current-replicas scales only if the current count is the expected value; it prevents accidental overwrites when two people make changes at the same time. Several deployments can also be scaled with one command:

kubectl scale deploy/web --current-replicas=3 --replicas=5
kubectl scale deploy web api --replicas=2
kubectl scale deploy -l katman=arka --replicas=0

Exercise Scale at the right level

Write two commands: scale the web deployment to 6 only if its current replica count is 3; scale all deployments labelled katman=arka down to 1 replica.

Show example solution

kubectl scale deploy/web --current-replicas=3 --replicas=6
kubectl scale deploy -l katman=arka --replicas=1

Test yourself

What happens if you scale a ReplicaSet owned by a Deployment to 10 replicas?
- The Deployment soon pulls it back to its own replicas value
- 10 replicas stay permanently
- The Deployment is deleted
- It errors
Show answer

The Deployment soon pulls it back to its own replicas value — Controllers continuously enforce the desired state.

Scale only if the current count is 3.

kubectl scale deploy/web ____=3 --replicas=5

Show answer

kubectl scale deploy/web --current-replicas=3 --replicas=5

Match.

scale rs/...

The owner undoes it
scale deploy/...

The right level
--current-replicas

Conditional scale
-l katman=arka

Bulk scale by label

Next lesson: Self-healing

We Are Enhancing Your Experience

Cookie Preferences

Strictly Necessary Cookies

Analytical Cookies