I have been hearing from my clients more and more that I want to save on my cloud cost. Taking that ask, I found this really cool project called kube-downscaler which is a pod that runs in your Kubernetes cluster that will scale down a deployment/statefulset or a whole namespace.

You can configure it in many ways and I am outlining one way that I have been using it.

In our pre-production clusters we don’t get much usage out of them during the weekends. The idea is that we will scale down the pods in these namespaces and in turn our cluster-autoscaler will scale down the nodes since they are not in use. The reverse will happen on a scale up.

In this example, I will scale down the gar namespace for a sort period of time for 1 hour.

apiVersion: v1
kind: Namespace
metadata:
  annotations:
    downscaler/downtime: Fri-Fri 01:45-02:45 UTC
  labels:
    name: gar
  name: gar

Pods are running in that namespace:

kubectl -n gar get pods -o wide                                      
NAME                            READY   STATUS             RESTARTS   AGE    IP             NODE                           NOMINATED NODE
tornado-fb459c7f9-9mtm7         1/1     Running            0          2m9s   100.96.0.223   ip-172-17-51-20.ec2.internal   <none>
webapp-nginx-85fcf96f7f-5bxgs   1/2     CrashLoopBackOff   4          2m9s   100.96.6.19    ip-172-17-50-79.ec2.internal   <none>
webapp-nginx-85fcf96f7f-s2k4n   1/2     Error              4          2m9s   100.96.0.224   ip-172-17-51-20.ec2.internal   <none>

Date on the kube-downscaler

kubectl -n kube-downscaler exec kube-downscaler-6969d86595-7g8dg date
Fri Nov  8 02:21:11 UTC 2019

Logs from the kube-downscaler

2019-11-08 01:45:24,425 INFO: Scaling down Deployment gar/webapp-nginx from 2 to 0 replicas (uptime: always, downtime: Fri-Fri 01:45-02:45 UTC)
2019-11-08 01:45:24,442 DEBUG: https://100.64.0.1:443 "PATCH /apis/apps/v1/namespaces/gar/deployments/webapp-nginx HTTP/1.1" 200 None

Pods in the gar namespace:

kubectl -n gar get pods -o wide                                        
No resources found in gar namespace.

Looks like it downscaled the pods in the gar namespace. Cool!

Now im waiting for the upscale event…

Logs from the kube-downscaler

2019-11-08 02:46:03,626 INFO: Scaling up Deployment gar/webapp-nginx from 0 to 2 replicas (uptime: always, downtime: Fri-Fri 01:45-02:45 UTC)
2019-11-08 02:46:03,644 DEBUG: https://100.64.0.1:443 "PATCH /apis/apps/v1/namespaces/gar/deployments/webapp-nginx HTTP/1.1" 200 None

Pods in the gar namespace:

kubectl -n gar get pods -o wide                                      
NAME                            READY   STATUS             RESTARTS   AGE     IP             NODE                           NOMINATED NODE
tornado-fb459c7f9-n5hzj         1/1     Running            0          2m52s   100.96.0.225   ip-172-17-51-20.ec2.internal   <none>
webapp-nginx-85fcf96f7f-4flvx   1/2     CrashLoopBackOff   4          2m52s   100.96.5.222   ip-172-17-50-67.ec2.internal   <none>
webapp-nginx-85fcf96f7f-vl9b5   1/2     CrashLoopBackOff   4          2m52s   100.96.0.226   ip-172-17-51-20.ec2.internal   <none>

Date on the kube-downscaler

kubectl -n kube-downscaler exec kube-downscaler-6969d86595-7g8dg date  
Fri Nov  8 02:49:28 UTC 2019

Looks like it scaled down and up as I configured it.

Contact me if you have any questions about this or want to chat, happy to start a dialog or help out: blogs@managedkube.com {::nomarkdown}

Learn more about integrating Kubernetes apps

ManagedKube Website

{:/nomarkdown}

Learn more about integrating Kubernetes apps

ManagedKube Website

Kube-Downscale