Kubernetes agents in Jenkins - Step-by-Step Execution

Process Flow - Kubernetes agents

Jenkins Master

↓

Request Agent Pod

↓

Kubernetes API

↓

Create Agent Pod

↓

Agent Pod Running

↓

Agent Executes Job

↓

Job Complete

↓

Agent Pod Deleted

Jenkins master requests a Kubernetes agent pod, which is created and runs the job, then is deleted after completion.

Execution Sample

Jenkins

podTemplate(label: 'k8s-agent') {
  node('k8s-agent') {
    stage('Build') {
      echo 'Running build on Kubernetes agent'
    }
  }
}

Defines a Kubernetes agent pod template and runs a build stage on that agent.

Process Table

Step	Action	Evaluation	Result
1	Jenkins master requests agent pod with label 'k8s-agent'	Request sent to Kubernetes API	Agent pod creation started
2	Kubernetes API creates pod based on template	Pod spec valid	Agent pod is running
3	Jenkins master schedules job on agent pod	Agent pod ready	Job starts executing
4	Job runs 'echo' command	Command executes	Output: 'Running build on Kubernetes agent'
5	Job completes	Success	Agent pod marked for deletion
6	Kubernetes deletes agent pod	Pod terminated	Agent pod removed

💡 Agent pod deleted after job completion, freeing resources

Status Tracker

Variable	Start	After Step 2	After Step 4	Final
agentPodStatus	Not created	Running	Running	Deleted
jobStatus	Not started	Not started	Running	Completed
jobOutput			Running build on Kubernetes agent	Running build on Kubernetes agent

Key Moments - 2 Insights

Why does Jenkins create a new pod for each agent instead of reusing one?

Each agent pod is created fresh to ensure a clean environment for the job, as shown in execution_table steps 1 and 2 where a new pod is created and started.

What happens if the agent pod fails to start?

If the pod fails to start, Jenkins cannot run the job on it, so the job will fail or retry. This is implied in step 3 where the job starts only if the pod is ready.

Visual Quiz - 3 Questions

Look at the execution table, what is the agentPodStatus after step 2?

A Not created

B Deleted

C Running

D Pending

💡 Hint

Check the variable_tracker table under agentPodStatus after Step 2

At which step does the job output 'Running build on Kubernetes agent' appear?

A Step 3

B Step 4

C Step 5

D Step 6

💡 Hint

Look at the execution_table row describing the job running the echo command

If the agent pod was not deleted after job completion, what would change in the variable_tracker?

A agentPodStatus would remain 'Running' or 'Terminated' instead of 'Deleted'

B jobStatus would be 'Failed'

C jobOutput would be empty

D No change in any variable

💡 Hint

Refer to the final value of agentPodStatus in variable_tracker

Concept Snapshot

Kubernetes agents in Jenkins:
- Jenkins master requests a pod from Kubernetes
- Pod is created fresh per job using podTemplate
- Job runs inside the pod as an agent
- Pod is deleted after job completes
- Ensures clean, isolated build environments

Full Transcript

This visual execution shows how Jenkins uses Kubernetes agents. First, Jenkins master requests an agent pod labeled 'k8s-agent'. Kubernetes API creates the pod based on the podTemplate. Once the pod is running, Jenkins schedules the job on it. The job runs the build stage, printing 'Running build on Kubernetes agent'. After the job finishes successfully, the agent pod is deleted to free resources. Variables like agentPodStatus and jobStatus change accordingly during these steps. This process ensures each job runs in a clean, isolated environment.