Kubernetes Interview Question with answer by farshad nick
0 likes37 views
The document discusses Kubernetes components and architecture, highlighting key elements like the master node's API server, etcd, and the scheduler. It also addresses operational concepts such as maintaining cluster availability through rolling updates, Kubernetes cron jobs, and managing pod communications with network policies. Additionally, it explains the differences between daemonsets and deployments, as well as handling evicted pods and utilizing static pods.
Kubernetes Interview Question with answer by farshad nick
- 2. 1- What is kubernetes components and Architecture ? Master Node: ● API Server: The API server is the central management point for the Kubernetes cluster. It exposes the Kubernetes API and serves as the interface for communication with the control plane components. ● etcd: This is a distributed key-value store that stores the configuration data and the current state of the cluster. All persistent cluster data is stored in etcd. ● Controller Manager: The controller manager is responsible for monitoring the state of the cluster through the API server and making necessary changes to bring the current state closer to the desired state. ● Scheduler: The scheduler is responsible for assigning pods to nodes based on resource requirements, policies, and constraints. It decides on which node a pod should run. ● Cloud Controller Manager (CCM) : the CCM is responsible for managing cloud-specific aspects of the cluster, such as load balancers, persistent volumes, and virtual machines, based on the underlying cloud infrastructure.
- 3. Node (Minion/Worker): ● Kubelet: Kubelet is an agent that runs on each node in the cluster. It ensures that containers are running in a pod and reports the node's status to the master. Kubelet takes instructions from the API server and manages containers on the node. ● Kube Proxy: Kube Proxy maintains network rules on nodes. It handles communication between pods and services, forwarding traffic as necessary. ● Container runtime Interface : This is the software responsible for running containers. It can be any runtime that implements the CRI, such as Docker, containerd, or others.
- 5. 2- Why Kubernetes Cluster must be in odd number? As you probably know, Kubernetes relies on etcd, which uses the RAFT consensus algorithm (similar to Paxos) RAFT algorithm uses majority voting to elect a leader for cluster You can See Article about rafting and this topic
- 6. 3- What is the solution for Zero-Downtime in kubernetes Rolling Updates: Use Kubernetes' built-in rolling update feature for deployments. This allows you to update your application incrementally, ensuring that a portion of the pods remains available during the update process. Readiness Probes: Implement readiness probes to check if your application is ready to serve traffic before it receives requests. This prevents traffic from being sent to pods that are not fully initialized or ready. Pod Disruption Budgets (PDB): Define PDBs to specify the maximum number of pods that can be unavailable during voluntary disruptions (like updates or node maintenance). This helps maintain a minimum level of availability. Horizontal Pod Autoscaler (HPA): Automatically scale your application based on metrics such as CPU or memory usage. This ensures you have enough pods to handle increased load, minimizing the risk of downtime.
- 7. Service Mesh: Use a service mesh like Istio or Linkerd to manage traffic routing and implement features like canary deployments and circuit breaking, which help maintain availability during deployments. Load Balancing: Utilize Kubernetes Services and Ingress controllers for load balancing traffic across pods. This distributes requests evenly and helps prevent overload on individual pods. Multi-Zone or Multi-Region Deployments: Deploy your applications across multiple availability zones or regions to ensure redundancy. If one zone goes down, traffic can be routed to healthy instances in another zone. Health Checks: Implement both liveness and readiness probes to automatically restart or stop sending traffic to unhealthy pods, ensuring that only healthy instances serve requests.
- 8. 4- What is difference between daemonset and deployment ? DaemonSet ensures that a specific Pod runs on all (or a subset of) nodes in the cluster, typically used for cluster-level services. Deployment manages a set of identical Pods, typically used for stateless applications, with features like rolling updates and replica management.
- 9. 5-There are some evicted pod in my kubernetes what is it and when this situation happens ? In Kubernetes, evicted pods are pods that get terminated because the node they were running on couldn't provide sufficient resources, like memory or disk space. This typically happens due to: 1. Out of Memory (OOM) Pressure: When the node runs out of memory, Kubernetes evicts lower-priority pods to free up resources. 2. Disk Pressure: When the node’s disk usage is critically high, Kubernetes evicts pods to prevent system failures. 3. Pod Quality of Service (QoS) Levels: Kubernetes assigns pods different QoS classes (Guaranteed, Burstable, BestEffort) based on their resource requests and limits. Pods with lower QoS classes 4. Resource Overcommitment 5. Node Failure or Maintenance To avoid evictions, ensure your pods have proper resource requests and limits defined.
- 10. 6- What is difference between linux cronjob and kubernetes cronjob ? Linux Cron Job: Runs tasks on a single server at the OS level. It's limited to that machine's capacity, with no built-in fault tolerance or automatic rescheduling if the server fails and its always consume resources Kubernetes CronJob: Runs tasks as pods in a Kubernetes cluster. It scales across multiple nodes, provides automatic failover, retries, and centralized management. Ideal for containerized and distributed workloads and it just run on a specific time that you defined
- 11. 7- What is static pods ? Static Pods in Kubernetes are pods that are directly managed by the kubelet on a specific node, rather than by the Kubernetes API server. They are usually used for system-level pods or critical infrastructure components (like control plane components) that need to run on a specific node. Static pods are defined using YAML manifests placed in a specific directory (e.g., /etc/kubernetes/manifests/). The kubelet watches this directory, and any changes to the manifests trigger pod updates.
- 12. 8- How can we Control pods to pods communication ( Deny or allow incoming/outgoing traffics ) ? Network Policy allows you to Control on Kubernetes network communication
- 13. 9-What is finalizer in kubernetes ? Finalizers are namespaced keys that tell Kubernetes to wait until specific conditions are met before it fully deletes resources marked for deletion. Finalizers alert controllers to clean up resources the deleted object owned. https://kubernetes.io/docs/concepts/overview/working-with-objects/finalizers/
- 14. 10- How to Deploy App X on kubernetes nodes that has SSD ? ● Node Selector ● Node Affinity ● Taints and Tolerations ● Pod Topology Spread Constraints ● Storage Classes (for Persistent Volumes with SSD backing) ● Persistent Volume Claims (targeting SSD-based storage) ● Dynamic Provisioning with SSD-backed Storage Classes
- 15. Follow for more Devops Content Farshad_nickfetrat @farshadnick @packops.ir packops.dev