Crack 31 LPA DevOps Interviews 🔥 | ArgoCD + Kubernetes + Helm

Summary

The video is a live DevOps training session focused on Argo CD and GitOps. It includes a hands-on deployment exercise using Kubernetes, Helm, AWS EKS, and Terraform. The instructor explains how Argo CD keeps a Kubernetes cluster aligned with configuration stored in Git, then demonstrates setting up infrastructure, deploying applications, and troubleshooting common problems.

Main Concepts

How the Tools Fit Together

The instructor describes the progression from virtual machines to lightweight containers, and then to Kubernetes for managing many containers.

• Helm packages Kubernetes application configuration into charts and provides commands for installing, upgrading, and rolling back releases.
• Argo CD automates Kubernetes delivery by monitoring a Git repository and reconciling the cluster with the configuration stored there.

GitOps and the Source of Truth

Git is described as the source of truth for desired application and cluster configuration. When a developer changes and commits Kubernetes YAML or Helm configuration, Argo CD detects the change and syncs it to the target cluster.

Argo CD can also detect when the live cluster drifts from the Git configuration. For example, if a deployment is manually deleted, Argo CD can restore the declared state.

The instructor distinguishes between:

• CI, which builds and tests application code.
• CD, which deploys already-built artifacts and Kubernetes configuration.

Argo CD is presented as a CD/GitOps tool, not as a build system.

Applications and Multiple Clusters

An Argo CD application definition specifies the Git repository, branch or revision, path within the repository, and destination cluster and namespace. Applications can use plain Kubernetes YAML or Helm charts. Different applications can be organized in separate folders in one repository or in separate repositories.

One Argo CD instance can manage applications across multiple Kubernetes clusters. The instructor also introduces ApplicationSets, which can generate application deployments across multiple registered clusters.

Practical Workflow Demonstrated

1. Prepare the AWS Environment

• Create a jump server and an EKS/Kubernetes cluster using Terraform.
• Install the required command-line tools, configure AWS access, and clone the project repository.
• Run Terraform initialization and planning before applying infrastructure changes.
• Check networking and security-group rules if the jump server cannot reach the cluster API.

2. Install and Access Argo CD

• Install Argo CD into a Kubernetes cluster.
• For the lab, expose its interface through a NodePort rather than creating a load balancer, to avoid additional cost.
• Use the correct node’s public IP and exposed port, and ensure the relevant security-group rule allows access.
• Retrieve or reset the Argo CD admin credentials when needed. The session demonstrates decoding a stored secret and generating a bcrypt password hash; the instructor advises consulting documentation or an AI assistant for exact commands.

3. Connect Git and Deploy Applications

• Add the Git repository to Argo CD and define applications for services such as Grafana and Nginx.
• Set each application’s repository, revision, path, destination cluster, and namespace.
• Review sync policy options, including manual and automatic sync.
• With automatic sync enabled, changes committed to Git can be applied without manually running Helm or kubectl for every deployment.
• Monitor application sync and health status in the Argo CD interface. The instructor also demonstrates creating an application through the UI.

4. Troubleshoot Using Kubernetes and Argo CD Status

• Inspect pod status, events, deployments, services, and logs when an application is pending, unhealthy, or in a crash loop.
• Check that the namespace exists and that the application’s manifests and service type match the intended access method.
• Use the correct node IP and port for NodePort access. Allowing a port in a security group is not enough if the wrong IP or service endpoint is being used.
• If an application’s live state is changed manually, check whether Argo CD will reconcile it with Git. For a lasting change, update the Git configuration rather than changing only the live cluster.
• For the Grafana issue shown, the instructor investigates database and persistent-volume configuration, replica count, and container logs. The discussion highlights that multiple replicas sharing unsuitable storage can cause problems.

Scaling Concepts Discussed

• Cluster Autoscaler adds or removes worker nodes when pods cannot be scheduled or when capacity is no longer needed. Its behavior depends on the configured node-group limits; in the lab, the maximum node count must be increased before the group can grow further.
• Metrics Server supplies resource-usage metrics used by Kubernetes components such as the Horizontal Pod Autoscaler.
• HPA changes the number of pod replicas based on metrics such as CPU or memory.
• VPA adjusts resource requests or limits for individual pods.
• DaemonSets create a pod on each eligible node, so their pod count rises when nodes are added.
• Karpenter is also mentioned as an event-driven way to provision worker nodes.

The session emphasizes checking dependencies and installation order—for example, ensuring required scaling components are present—and confirming what Terraform plans to create, change, or destroy.

Broader Lessons and Closing Message

• Understand a tool’s purpose and how it fits into a system rather than memorizing commands.
• Use Git as the authoritative record for desired deployments, and understand the difference between declared state and live cluster state.
• Practice diagnosing real failures, including scheduling problems, missing namespaces, configuration errors, storage issues, and network-access mistakes.
• The instructor promotes a free training program and says it requires substantial study time: around 10–12 hours daily for non-working students or 5–6 hours for working students. The instructor also cites placement and package figures and invites viewers to register for a future batch; these are claims made during the session.

Speakers and Sources Featured

• Instructor/trainer: Leads the lesson, explains the concepts, and performs the demonstrations. The subtitles do not clearly identify the instructor by name.
• Students and live-chat participants: Ask questions, respond to prompts, and discuss the demonstrations and errors.
• ChatGPT/AI assistant: Mentioned as a resource for finding installation commands and troubleshooting guidance.
• YouTube viewers/commenters: Addressed by the instructor during the live session and closing remarks.