As modern software development shifts towards cloud-native architectures, service pipelines have become key to deploying, managing, and scaling microservices independently. In this post, we’ll explore the fundamentals of Tekton, a Kubernetes-native framework, is built to automate CI/CD processes for services and the power to build complex workflows.
We'll also discuss why you might want to use Tekton over traditional CI/CD tools like GitLab CI or GitHub Actions.
What is Tekton?
Tekton is a Kubernetes-based framework for defining, running, and managing CI/CD pipelines. Unlike traditional CI/CD systems that typically handle the deployment of entire applications, Tekton excels at managing service pipelines, which allow individual microservices to be built, tested, and deployed independently. This gives you the ability to release services faster and with more flexibility.
Why Use Tekton for Service Pipelines?
1. Kubernetes-Native and Cloud-Native Integration
Tekton is designed specifically for Kubernetes environments, making it a perfect fit for organizations running microservices or cloud-native applications. Unlike GitLab CI or GitHub Actions, which require some integration overhead, Tekton runs directly within Kubernetes clusters, taking full advantage of Kubernetes’ features.
2. Fine-Grained Control and Flexibility
Tekton provides more control over task execution, pipeline structure, and parameterization compared to traditional CI/CD tools. This flexibility allows you to build complex, customized workflows for each microservice from your code.
3. Vendor-Neutral, Open-Source Solution
Tekton is completely open-source and vendor-neutral, meaning you’re not locked into a specific CI/CD ecosystem like GitLab or GitHub. This gives you the freedom to use Tekton in any Kubernetes environment.
Key Concepts
- Tasks: Individual units of work, such as running a build or test.
- Pipelines: Collections of tasks that run in sequence or parallel.
- TaskRuns and PipelineRuns: Runtime instances of tasks and pipelines, allowing you to pass parameters dynamically.
Tekton Service Pipeline Example
Let’s walk through building a service pipeline that retrieves content from a URL, displays it, and then uses that content to generate a personalized message. I will add the in-line comments for clarification.
Tekton Task
A Task is the basic building block of a Tekton service pipeline. Below is a task that greets a user based on an input parameter.
cat <<EOF | kubectl apply --validate=false -f -
apiVersion: tekton.dev/v1
kind: Task
metadata:
name: greet-task
spec:
params:
- name: person
type: string
description: Who should we greet?
default: world
steps:
- name: greet
image: alpine
command:
- echo
args:
- "Hello \$(params.person)!"
EOF
This task uses the Alpine image to echo a personalized greeting, which can be customized by passing a value for the person
parameter.
A Task is like docker image, meanwhile the TaskRun is a container
TaskRun Definition
The TaskRun below executes the task, allowing us to pass a specific value to the person
parameter.
cat <<EOF | kubectl apply --validate=false -f -
apiVersion: tekton.dev/v1
kind: TaskRun
metadata:
name: greet-task-run-1
spec:
taskRef:
name: greet-task
params:
- name: person
value: "future engineers"
EOF
Applying this task, we can see the output in pod log:
$ kubectl get pod | grep greet-task-run-1
greet-task-run-1-pod 0/1 Completed 0 34s
$ kubectl logs greet-task-run-1-pod
Defaulted container "step-greet" out of: step-greet, prepare (init)
Hello future engineers!
There are a Tekton hub containing Tasks which has been written by community.
We will examine the community wget
Task in next section
Tekton Pipeline and PipelineRun
Now, let’s define a service pipeline that uses multiple tasks. We’ll use the wget task to fetch content from the internet and then pass that content into our greeting task.
Install the Wget Task
First, install the wget
community task from Tekton Hub:
kubectl apply -f https://api.hub.tekton.dev/v1/resource/tekton/task/wget/0.1/raw
Verify with kubectl get task
we'll see the new wget Task.
Pipeline Definition
The following pipeline defines a sequence of tasks: fetching content from a URL wget
, displaying that content cat
, and greeting the user with the content as a message greet-task
. Flow will be like this:
In this pipeline:
- fetch-content: Uses
wget
to download content from a URL and store it in a workspace.
- display-content: Reads the content and outputs it as a pipeline result.
- greet-with-content: Passes the fetched content as the greeting message in the final task.
cat <<EOF | kubectl apply --validate=false -f -
apiVersion: tekton.dev/v1
kind: Pipeline
metadata:
name: greet-service-pipeline
annotations:
description: |
Fetch content using wget, display it, and greet the user.
spec:
params:
- name: sourceUrl
description: The URL to fetch the content from
type: string
default: ""
workspaces:
- name: storage
tasks:
- name: fetch-content
taskRef:
name: wget
params:
- name: url
value: "\$(params.sourceUrl)"
- name: diroptions
value:
- "-P"
workspaces:
- name: wget-workspace
workspace: storage
- name: display-content
runAfter: [fetch-content]
workspaces:
- name: output
workspace: storage
taskSpec:
workspaces:
- name: output
steps:
- name: cat-content
image: busybox
script: |
#!/bin/sh
cat \$(workspaces.output.path)/name | tee /tekton/results/fetchedContent
results:
- name: fetchedContent
description: The content fetched from the URL
- name: greet-with-content
runAfter: [display-content]
taskRef:
name: greet-task
params:
- name: person
value: "\$(tasks.display-content.results.fetchedContent)"
EOF
PipelineRun Definition
Here’s how you run the pipeline using a PipelineRun, specifying a URL to fetch content from:
cat <<EOF | kubectl apply --validate=false -f -
apiVersion: tekton.dev/v1
kind: PipelineRun
metadata:
name: greet-service-pipeline-run-1
spec:
workspaces:
- name: storage
volumeClaimTemplate:
spec:
accessModes:
- ReadWriteOnce
resources:
requests:
storage: 2Mi
params:
- name: sourceUrl
value: "https://raw.githubusercontent.com/nhatsangvn/nhatsangvn/refs/heads/main/name"
pipelineRef:
name: greet-service-pipeline
EOF
Applying this pipeline will get the content from the specified URL, display it, and greet the information from the file.
$ kubectl get pod | grep greet-service-pipeline
greet-service-pipeline-run-1-display-content-pod 0/1 Completed 0 74s
greet-service-pipeline-run-1-fetch-content-pod 0/1 Completed 0 95s
greet-service-pipeline-run-1-greet-with-content-pod 0/1 Completed 0 67s
$ kubectl logs greet-service-pipeline-run-1-greet-with-content-pod
Defaulted container "step-greet" out of: step-greet, prepare (init)
Hello SangNN
!
Here we go: "Hello SangNN"
Conclusion
In this blog post, we’ve covered the basics of Tekton and how it can be used to build service pipelines tailored for microservices and cloud-native architectures. By leveraging Tekton, you gain the flexibility, control, and scalability needed to manage independent service deployments, allowing for faster and more frequent updates. Whether you're building for Kubernetes, looking to decouple CI/CD processes, or need a vendor-agnostic solution, Tekton's service pipeline approach is a strong candidate for modern DevOps workflows.
Stay tuned for more advanced topics on Tekton in future posts!