Kubernetes Cluster Autoscaler
Вставка
- Опубліковано 27 вер 2024
- ===================================================================
1. SUBSCRIBE FOR MORE LEARNING :
/ @cloudquicklabs
===================================================================
2. CLOUD QUICK LABS - CHANNEL MEMBERSHIP FOR MORE BENEFITS :
/ @cloudquicklabs
===================================================================
3. BUY ME A COFFEE AS A TOKEN OF APPRECIATION :
www.buymeacoff...
===================================================================
This video shows (step-by-step demo) scratch to end how to set up Amazon EKS cluster Autoscaler . Here it also shows in depth pictorial overview of Kubernetes cluster Autoscaler and step-by-step required configuration set up using commands and manual.
At the end it shows live show of cluster auto scaling where number of nodes gets auto-increased when there is traffic or resource crunch at EKS cluster.
Command and support files used in demo can be found at Git hub repo link: github.com/Rek...
#kubernetes #clusterautoscaler #autoscale #cluster #aws #eks
Thank you! Helped a lot!
Thank you for watching my videos.
Hi, thanks for the video. I have a small doubt: can't we configure cluster autoscaler for the default node group?
Thank you for watching my videos.
Yes,We should be able to configure autoscaler for default node as well.
im getting crashloopbackoff can anyone help?
Thank you for watching my videos.
This means that Kubernetes is trying to start the pods but its not getting started. Please check two things one check if resources (memory / cpu) on your nodes, check if image that trying to pull exist or is happening as expected. Finally check if replica set configured as expected
Hi Thank you for this awesome video. I would like to ask two three queries
1: The moment when you increased the replica the cluster auto scaler increased the nodes but in the real time how its going to work because we can not manually increase the replica on the basis of traffic so my question is do we need HPA for that because without HPA how the nodes going to behave that it needs more pods with manual interruption.
2: Secondly I have four node groups in the EKS cluster and out of which one is on private subnet and I want to make that node group as a auto scaling one. So how can it auto scale by choosing the private subnet automatically because while creating new node group on private subnet I have to manually select the private subnet.
3: Thirdly in the node group, even though the tag is showing empty because I have not added any auto scale tag in it but in the ec2 auto scaling group the tag has value for the same node and whenever I am using that command "aws autoscaling describe-auto-scaling-groups" the value showing empty but in the ec2 auto scaling groups for the same node the lists of tags are showing.
Thank you for watching my videos.
And thank you for asking good questions.
1. Cluster autoscale and HPA both has to co-exist then the auto-scaling of nodes makes more sense. When HPA create more pods leads to consumption of more resource and leads to auto-scaling of nodes.
2. We can define which node groups to autoscale in configurations. Hence we can control it.
3. When tags are available it should show the tags even via command or console. Check if you have required access and running correct command here.
Hi Thanks for the tutorial. I was trying to understand worker group node can also be auto-scale correct? and what are the main benefits of using node groups? Can you please help me with this?
Thank you for watching my videos.
Yes EKS underlying EC2 node can also be autoscaled on demand and we call it cluster autoscaler.
There many benefits for using managed node groups at AWS EKS like below.
1. It helps us to segregate application nodes (worker nodes) from Kubernetes master nodes.
2. It provides capability to add/remove worker nodes to existing or newly creating EKS cluster where you can directly run Kubernetes applications.
3.It provides great scalability features on demand (where it uses EC2 Autoscaling group)
4. Its from AWS no additional cost for setting up managed node groups but only EC2 services within it.
Great Sir
Thank you for watching my videos.
Great tutorial. When I scaled down the pods, it not scale down the nodes. Can you explain how can i scale down the nodes.
Thank you for watching my videos.
Scale up and Scale down are automatic. Hencebit should scale down automatic when you stop resource consuming process. But you need to wait few minutes as it takes time to process.
@cloudquicklabs: Great videos to learn autoscaler . I have few question on same line 1) how my auto scaler of worker nodes increase based on the of CPU and memory (80%) threshold. 2) how my worker nodes should span in multiple az's with autoscaling. Can you suggest pls and thanks in advance.
Thank you for watching my videos.
1. Generally you need increase node count based on your application requirement , May be you need to increase by one.
2. Generally high available across AZs. But this also depends on your application requirements.
Great sir
Thank you for watching my videos.
Hi , it was a very useful tutorial. I was trying to scale down the node by reducing the no of pods but nodes are not scaling down .Could you plz help me with this?
Thank you for watching my videos.
When pods numbers increased does it scaled node number up?
Hi, It is a very useful video for me, I have done the workaround.
I scale down the pod from 80 to 20, In the meantime, How can I scale down the nodes automatically ...?
Please share if you have any documents or related videos.
Thanks in advance.
Thank you for watching my videos.
When you scale down the pods, which means you are reducing the load and it should eventually reduce the number of node at node group automatically.
Please check it again.
@@cloudquicklabs Thank you, I have checked, and works for me.
As I noticed, a Total of 4 Nodes, Each node 15 Pods are running Total PODS - 60
When I scale down the pods from 60 to 30, Out of 4 nodes one node has 5 pods are running and it is not automatically drained into another node.
Should I drain automatically to reduce the node ..?
Thank you for coming back on this. Based on node count definitions(at rest node count and final size count when there is load) at node group, cluster autoscaler (CA) does automatic scale nodes count up and down based on traffic (resource crunch), so in your case validate what you have given at node group definition.
You explained it very clearly. Thank you so much.
Thank you for watching my videos.
Happy learning.