K8s - Affinitiy and Anti-Affinity

-

 


Affinity refers to "Establishing a relationship based on similar characteristics".

Anti-Affinity refers to "Breaking a relationship based on a similar characteristics".

This affinity can be set at node level and pod level.

Both node and pod affinity has two parameters:

requiredDuringSchedulingIgnoredDuringExecution: Means the scheduler should look for the node matching the nodeAffinity condition mentioned in the pod configuration file.

preferredDuringSchedulingIgnoredDuringExecution: Means the scheduler tries look for the node matching the nodeAffinity condition mentioned in the pod configuration file. If no node is available then the scheduler creates pod on any available nodes.

What is a node affinity?

Node affinity says scheduler to place a pod on a node when one or more condition matches.

Node affinity is very similar to "Node Selector", but under node affinity we can mention one or more conditions.


I have a node labelled as datacenter=Chennai,operating_system=Linux,gpu=True.

Now, I want to schedule a pod onto a node matching all 3 conditions.



Now, the pod is created. Let's make change to the node affinity. Renaming datacenter reference from Chennai to chennai in the pod configuration.

Now the pod stays in the "Pending" state as the scheduler keeps looking for a node matching all the conditions.


Since we mentioned in the pod configuration as "requiredDuringSchedulingIgnoredDuringExecution" it has satisfy all the requirements.

Let's change to "preferredDuringSchedulingIgnoredDuringExecution"


When using preferredDuringSchedulingIgnoredDuringExecution, we need to mention "weight". The scheduler considers the weight of each node and adds the weight to the other scores for that node, and schedules the Pod onto the node with the highest final score.



Comments

Post a Comment

Popular posts from this blog

K8s - ETCD

-
Image
  etcd is a "strongly consistent , distributed key-value store". Why etcd? 1. Consistency : Since the API server is the central coordination point of the entire cluster; strong consistency is essential. It would be a disaster if, say, two nodes tried to attach the same persistent volume over iSCSI because the API server told them both that it was available. 2. Availability: API downtime means that the entire Kubernetes control plane comes to a halt, which is undesirable for production clusters. The  CAP theorem  says that 100% availability is impossible with strong consistency, but minimizing downtime is still a critical goal. 3. Consistent Performance: The API server for a busy Kubernetes cluster receives a fair amount of read and write traffic. The secret behind etcd's balance of strong consistency and high availability is the  Raft algorithm . Raft solves a particular problem: how can multiple independent processes decide on a single value for somethin...
Read more

SRE/DevOps Syllabus

-
Image
  DEVOPS/SRE Principles. Git What is Git? Architecture of Git. Working principle of Git. Create and cloning a repo. Version control branching. Version control commit. Version control managing workflows. Git hooks. Git Reflog. Git Stash. Git Cherry Picking. Undoing changes in different states of Git. Git based terraform template management. AWS CI/CD SDLC Automation. Code Commit. Code Build. Code Deployment. Code Pipeline. Elastic Beankstalk. Code Artifact. CodeGuru. Terraform Terraform Basics. Terraform State. Working with Terraform. Terraform with AWS. Remote State. Terraform Provisioners. Terraform Import, Tainting and Debugging. Terraform Modules. Terraform Functions and Conditional Expressions. Automation using Python - AWS AWS lambda. Automating EC2 with Lambda. Automating S3 with Lambda. Automating VPC with Lambda. Cost optimization with Lambda. SNS,SQS and SES with Python. Managing and Automating AWS Security with Python. Kubernetes - CKAD Application Developer. Kubernetes A...
Read more

K8s - Deployment and HPA replicas

-
Image
  We all know that replica is a critical part of a K8s deployment. Default replica is 1. Setting replicas as 3 will ensure 3 pods of that deployment is running at any given time.  Replica are internally managed by replica set which is in turn controlled by the K8s  ReplicationController   . When a pod is bad or deleted as part of replicaset, replication controller will ensure a new pod is created. Now, coming to HPA (Horizontal Pod Autoscaler) which is a dynamic scaler which works at the pod level based on the metrics configured.  When we configure we must mention min and max values. They should be greater than 1. Let's image I create a HPA with min as 5 and max as 10 under the condition of CPU utilization more than 50%. Initial deployment is set with the replica as 1. On top the deployment, I am attaching a HPA with the mentioned configuration. Under HPA, we are forcing the min replicas to 5. Hence, it autoscales the pod to 5 even though the CPU utilization is ...
Read more