Example project showing how to create & deploy a Nuxt.js / Vue.js based frontend and how to interact with a Spring Boot microservice (https://github.com/jonashackt/microservice-api-spring-boot)
The purpose of this project is elaborate the differences of a Nuxt.js / SSR or Static Site Generation based project to https://github.com/jonashackt/spring-boot-vuejs
┌────────────────────────────────┐
│ │
│ │
│ microservice-ui-nuxt-js │
│ │
│ │
└───────────────┬────────────────┘
│
│
┌───────────────▼────────────────┐
│ │
│ │
│ microservice-api-spring-boot │
│ │
│ │
└────────────────────────────────┘
"Universal Webapps" --> faster loading of web sites!
How? server does pre-rendering of the "start" HTML & CSS on server-side - and then loads JavaScript & dynamic CSS in the background, but user can already scroll and read.
All 3 popular web frameworks support Universal Web Apps:
- Angular: https://angular.io/guide/universal
- React: https://nextjs.org/ and https://www.gatsbyjs.com/
- Vue.js: https://nuxtjs.org/
For all concepts see https://vueschool.io/articles/vuejs-tutorials/hybrid-rendering-in-nuxt-js-3/
There are 4 modes now with Nuxt 3: Universal Rendering (aka Server Side Rendering (SSR)), Client Side Rendering (aka Static Site Generation, Hybrid Rendering & Edge-Side Rendering
-
Universal Rendering (aka Server Side Rendering (SSR)): SSR sends fully rendered page from server to the client -> which then gets hydrated (https://vuejs.org/guide/scaling-up/ssr.html), which means that Vue.js turns the server rendered page into dynamic DOM that can react to client-side data changes
-
Client Side Rendering (aka Static Site Generation): the application gets rendered during build phase and can then be deployed to any static hosting service (Netlify, GitHub Pages, AWS S3 static site hosting, Static hosting on Azure Storage Accounts etc.). There's no server needed for deployment & the content is delivered via Content Delivery Networks (CDNs). Additionally in Static Side Generation mode there's also a SPA Fallback for sites that should be rendered on client side and won't be served through the CDN.
Client Side Rendering aka Static Site Generation can be enabled inside the nuxt.config.ts
:
export default defineNuxtConfig({
ssr: false
})
// If you do use ssr: false, you should also place an HTML file in ~/app/spa-loading-template.html with some HTML you would like to use to render a loading screen that will be rendered until your app is hydrated.
-
Hybrid Rendering: Route Rules to decide which rendering method to use for every route
-
Edge-Side Rendering: With ESR, the rendering process is pushed to the 'edge' of the network - the CDN's edge servers. ESR is more a deployment target than an actual rendering mode.
See https://nuxt.com/docs/getting-started/installation
If you have node
and npm
installed, the npx
command should also work (see https://linuxtldr.com/npx-package-runner/).
npx nuxi@latest init microservice-ui-nuxt-js
A wizard will take you through the setup process.
If you want, you can add some UI candy like Element Plus https://nuxt.com/modules/element-plus (which leverages https://github.com/element-plus/element-plus). In the project root simply run:
npm i element-plus @element-plus/nuxt -D
And add the following lines to your nuxt.config.ts
:
modules: [
'@element-plus/nuxt'
],
elementPlus: { /** Options */ }
After project generation has finished, let's finally run our project skelleton with:
npm run dev -- -o
https://nuxt.com/docs/getting-started/views
By default, Nuxt will treat this file as the entrypoint and render its content for every route of the application.
So our file app.vue
is the starting point for everything.
https://nuxt.com/docs/getting-started/views#components
You can also create reusable Components in the components
directory, which may be an AppAlert, a Logo or something the like.
https://nuxt.com/docs/getting-started/views#pages
Pages represent views for each specific route pattern. Every file in the pages/ directory represents a different route displaying its content.
https://nuxt.com/docs/getting-started/views#layouts
Layouts are wrappers around pages that contain a common User Interface for several pages, such as a header and footer display
https://nuxt.com/docs/guide/directory-structure/app#minimal-usage
The CLI generates a file app.vue
, which is only useful for minimal sites like landing pages without the need for a Vue router:
"With Nuxt 3, the pages/ directory is optional. If not present, Nuxt won't include vue-router dependency. This is useful when working on a landing page or an application that does not need routing."
If you use app.vue
together with /pages
directory, you should use the <NuxtPage/>
component:
<template>
<div>
<NuxtLayout>
<NuxtPage/>
</NuxtLayout>
</div>
</template>
"
app.vue
acts a the main component of your Nuxt application. Anything you add to it (JS and CSS) will be global and included in every page"
https://nuxt.com/docs/getting-started/routing
Most websites will have more than one page (i.e. a home page, about page, contact page etc.). In order to show these pages, we need a Router. That's where vue-router comes in. When working with the Vue application, you have to set up a configuration file (i.e. router.js) and add all your routes manually to it. Nuxt.js automatically generates the vue-router configuration for you, based on your provided Vue files inside the pages directory. That means you never have to write a router config again! Nuxt.js also gives you automatic code-splitting for all your routes.
In other words, all you have to do to have routing in your application is to create .vue files in the pages folder.
Pretty cool :) Simply put your .vue
files into /pages
dir.
https://nuxt.com/docs/guide/directory-structure/pages#usage
"The pages/index.vue file will be mapped to the / route of your application."
So index.vue
is the HOME
of the site.
https://nuxt.com/docs/guide/directory-structure/pages#navigation
To navigate between pages of your app, you should use the NuxtLink component. This component is included with Nuxt.js and therefore you don't have to import it as you do with other components
So if you come from Vue.js development like me, you may like the following mapping:
<router-link
in Nuxt is<NuxtLink>
(see https://nuxtjs.org/docs/2.x/get-started/routing#navigation)<router-view/>
is<Nuxt>
, which is the component you use to display your page components (see https://nuxtjs.org/docs/2.x/features/nuxt-components#the-nuxt-component)
https://nuxt.com/docs/getting-started/testing
https://dev.to/tao/adding-vitest-to-nuxt-3-2023-lpa
After I used jest some time ago, there's a new testing framework in town: https://vitest.dev
First we need to add Testing capabilities to our Nuxt project:
npm i --save-dev @nuxt/test-utils vitest @vue/test-utils happy-dom playwright-core
Now we can add @nuxt/test-utils/module
to our nuxt.config.ts
(optional). This adds a Vitest integration to our Nuxt DevTools which supports running our unit tests in development:
export default defineNuxtConfig({
devtools: { enabled: true },
modules: [
'@element-plus/nuxt',
'@nuxt/test-utils/module'
],
...
})
We also need to create a vitest.config.ts
:
import { defineVitestConfig } from '@nuxt/test-utils/config'
export default defineVitestConfig({
// any custom Vitest config you require
})
Here the Nuxt docs are a bit vary and if you leave out the configuration, you get the following error (see https://stackoverflow.com/a/70771291/4964553):
FAIL test/Logo.spec.ts > Logo > is a Vue instance
ReferenceError: document is not defined
To fix this, configure happy-dom
(or jsdom
, if you installed that alternatively):
import { defineVitestConfig } from '@nuxt/test-utils/config'
export default defineVitestConfig({
// any custom Vitest config you require
test: {
globals: true,
environment: 'happy-dom',
},
})
We also need to have a npm run test
goal, that will execute our tests e.g. in CI. Therefore add the following line to your package.json
:
"scripts": {
...
"test": "vitest",
...
As already used from jest we create a new test case in the /test
directory (which we need to create beforehand). As an example here I use Logo.spec.ts
:
import { describe, it, expect } from 'vitest'
import { mount } from '@vue/test-utils'
import Logo from '../components/Logo.vue'
describe('Logo', () => {
it('is a Vue instance', () => {
const wrapper = mount(Logo)
expect(wrapper.vm).toBeTruthy()
})
})
Now we can finally run our test using npm run test
:
$ npm run test
> test
> vitest
DEV v1.2.2 /home/jonashackt/dev/microservice-ui-nuxt-js
✓ test/Logo.spec.ts (1)
✓ Logo (1)
✓ is a Vue instance
Test Files 1 passed (1)
Tests 1 passed (1)
Start at 10:20:20
Duration 523ms (transform 127ms, setup 10ms, collect 103ms, tests 11ms, environment 117ms, prepare 89ms)
PASS Waiting for file changes...
press h to show help, press q to quit
https://nuxt.com/docs/guide/directory-structure/env
if you have a .env file in your project root directory, it will be automatically loaded at dev, build and generate time. Any environment variables set there will be accessible within your nuxt.config file and modules.
Thus we need to create a .env
file:
BASE_URL = http://localhost:8098
.env
files aren't checked into source control per default (.gitignore
). So you need to recreate them in every development environment.
https://nuxt.com/docs/guide/going-further/runtime-config#exposing
Nuxt provides a runtime config API to expose configuration and secrets within your application.
According to https://stackoverflow.com/a/73636631 add the following to your `nuxt.config.ts:
export default defineNuxtConfig({
...
runtimeConfig: {
public: {
baseUrl: process.env.BASE_URL || 'http://localhost:8098',
},
},
...
})
https://serversideup.net/using-environment-variables-in-nuxt-3/
https://stackoverflow.com/a/75870302
Now we need a composable to add our baseUrl
every time we want to use $fetch
. So let's create the composable
directory, if not already created and than create a file useApiFetch.ts
:
import { useFetch } from "#app"
type useFetchType = typeof useFetch
// wrap useFetch with configuration needed to talk to our API
export const useAPIFetch: useFetchType = (path, options = {}) => {
const config = useRuntimeConfig()
// baseURL is configured in nuxt.config.ts
options.baseURL = config.public.baseUrl + '/api'
return useFetch(path, options)
}
We now also use useFetch
instead of directly using $fetch
. That's why:
https://nuxt.com/docs/api/composables/use-fetch
useFetch
is a composable meant to be called directly in a setup function, plugin, or route middleware. It returns reactive composables and handles adding responses to the Nuxt payload so they can be passed from server to client without re-fetching the data on client side when the page hydrates.
Here's a great introduction: https://masteringnuxt.com/blog/data-fetching-basics
"How can I use the return values from useFetch, called in an event handler, in my template?": nuxt/nuxt#19447
As the docs about axios state, there's something new in town for fetching data from an API: https://nuxt.com/docs/getting-started/data-fetching
Nuxt provides composables to handle data fetching within your application.
useFetch
is the most straightforward way to handle data fetching in a component setup function.$fetch
is great to make network requests based on user interaction.useAsyncData
, combined with $fetch, offers more fine-grained control.
The new fetching API is based on https://unjs.io's https://github.com/unjs/ofetch
Here are the docs for the most used useFetch
https://nuxt.com/docs/api/composables/use-fetch
Here's a basic example:
<template>
<div class="service">
<h1>{{ msg }}</h1>
<h2>REST service call results</h2>
<button @click="fetchHelloApi">CALL Spring Boot REST backend service</button>
<h4 v-if="!apiCallPending && !apiCallError">Backend response: {{ helloMsg }}</h4>
<p v-if="apiCallError">Error: {{ apiCallError.message }}</p>
</div>
</template>
<script setup lang="ts">
const msg = 'HowTo call APIs using Nuxt3 useFetch():';
const helloMsg = ref('');
let apiCallPending = ref(false);
let apiCallError: any;
// See https://github.com/nuxt/nuxt/discussions/19447
async function fetchHelloApi() {
const { data, pending, error } = await useAPIFetch<string>('/hello', );
console.log(data);
helloMsg.value = data.value || '';
apiCallPending = pending;
apiCallError = error;
}
</script>
The combination of calling an async
function and then await
for it's promise to be filled is described also here https://stackoverflow.com/a/45018619/4964553
Common errors while using the new Nuxt 3 fetch api (ofetch):
How to access values from await createNewUser():
interface User {
id: number;
firstName: string;
lastName: string;
};
let user: User = { id: 0, firstName: '', lastName: ''};
...
var userId = await createNewUser();
user.id = userId; // --> ts(2322): Type 'Ref<number | null>' is not assignable to type 'number'.
Solution: useFetch
returns refs as Ref<number | null>
.
So in order to access the value returned by useFetch
, we need to append .value
:
var userId = await createNewUser();
user.id = userId.value;
interface User {
id: number;
firstName: string;
lastName: string;
};
let user: User = { id: 0, firstName: '', lastName: ''};
async function createNewUser () {
const {data: id, pending, error, refresh} = await useAPIFetch(`/user/` + user.firstName + '/' + user.lastName);
return id;
};
var userId = await createNewUser();
user.id = userId.value; // --> ts(2322): Type 'unknown' is not assignable to type 'number'.
Solution: https://stackoverflow.com/a/72529632/4964553
Use useFetch
with explicit type definition <number>
:
await useAPIFetch<number>
interface User {
id: number;
firstName: string;
lastName: string;
};
let user: User = { id: 0, firstName: '', lastName: ''};
async function createNewUser () {
const {data: id, pending, error, refresh} = await useAPIFetch<number>(`/user/` + user.firstName + '/' + user.lastName);
return id;
};
var userId = await createNewUser();
user.id = userId.value; // --> error ts(2322): Type 'number | null' is not assignable to type 'number'
When Typescript is used with strict null checks (see https://www.typescriptlang.org/tsconfig#strictNullChecks), the compiler checks if there could be situations, where your variables are undefined. In this case it emits the error. And it's definitely unclear, if our API responds finally. So we could circumvent this by defining a default value (see https://stackoverflow.com/a/74796026/4964553). In our case the userId.value
is maybe of type number
, but maybe not defined. So we use an appended || 0
here:
thisUser.id = userId.value || 0
The UI () doesn't get updated
We have the following in our template:
<button @click="createNewUser">Create User</button>
<div v-if="showResponse"><h6>User created with Id: {{ user.id }}</h6></div>
with the <script setup lang="ts">
section:
var showResponse: boolean = false;
async function createNewUser () {
const {data: id, pending, error} = await useAPIFetch<number>(`/user/` + user.firstName + '/' + user.lastName, {
method: 'post'
});
...
showResponse = true;
...
};
Now clicking on the Create User
button, doesn't update the UI :(
Solution: We need to use ref()
here! And Why refs?
https://vuejs.org/guide/essentials/reactivity-fundamentals.html#why-refs:
You might be wondering why we need refs with the .value instead of plain variables. To explain that, we will need to briefly discuss how Vue's reactivity system works.
When you use a ref in a template, and change the ref's value later, Vue automatically detects the change and updates the DOM accordingly. This is made possible with a dependency-tracking based reactivity system. When a component is rendered for the first time, Vue tracks every ref that was used during the render. Later on, when a ref is mutated, it will trigger a re-render for components that are tracking it.
In standard JavaScript, there is no way to detect the access or mutation of plain variables. However, we can intercept the get and set operations of an object's properties using getter and setter methods. The .value property gives Vue the opportunity to detect when a ref has been accessed or mutated.
Luckily our template can stay the same:
Notice that we did not need to append .value when using the ref in the template. For convenience, refs are automatically unwrapped when used inside templates
But our script
section now changes slightly:
import { ref } from 'vue'
const showResponse = ref(false);
async function createNewUser () {
const {data: id, pending, error} = await useAPIFetch<number>(`/user/` + user.firstName + '/' + user.lastName, {
method: 'post'
});
...
showResponse.value = true;
...
};
Now our UI gets updated as expected :)
Also we need to use ref()
for our User objects:
// from this
let user: User = { id: 0, firstName: '', lastName: ''};
// to this
let user = ref({ id: 0, firstName: '', lastName: ''});
The type is automatically inferred. We could alternatively write the equivalent:
let user: Ref<User> = ref({ id: 0, firstName: '', lastName: ''});
Configure Axios & Spring Boot to handle CORS/SOP
See https://github.com/jonashackt/spring-boot-vuejs#the-problem-with-sop
As our Spring Boot App is deployed separately, we don't really need the CORS setup right? Nah, we need at least something!
See https://stackoverflow.com/a/56168756/4964553 and https://stackoverflow.com/questions/52230470/how-to-use-webpack-dev-proxy-with-nuxt
So let's configure the correct header in Axios. Therefore head over to our plugins/axios.ts and add the 'Access-Control-Allow-Origin': '*',
header:
// TODO: We need to make the baseURL configurable through environment variables for sure in the next step!
const axiosApi = axios.create({
baseURL: `http://localhost:8098/api`,
timeout: 1000,
headers: {
'Access-Control-Allow-Origin': '*',
'Content-Type': 'application/json'
}
});
Don't forget the backend to get CORS enabled
Our Spring Boot based backend https://github.com/jonashackt/microservice-api-spring-boot needs to support CORS too.
See https://spring.io/guides/gs/rest-service-cors/
So inside backend Controller BackendController.java we need to use the org.springframework.web.bind.annotation.CrossOrigin
and should mind our Spring Security configuration also (see https://stackoverflow.com/a/67583232/4964553).
We should change the
allowedOrigins
from the wildcard*
to a concrete URL as we go to production.
Make baseUrl configurable via environment variables
A common problem is how to make the baseUrl
variable configurable inside our axios.ts:
import axios, {AxiosResponse} from 'axios'
// TODO: We need to make the baseURL configurable through environment variables for sure in the next step!
const axiosApi = axios.create({
//baseURL: `http://fargatealb-81c02c2-1301929463.eu-central-1.elb.amazonaws.com:8098/api`,
baseURL: `http://localhost:8098/api`,
timeout: 1000,
headers: {
'Access-Control-Allow-Origin': '*',
'Content-Type': 'application/json'
}
});
Using comments isn't elegant in any way - and the production baseUrl
will differ every time we do a new Pulumi/IaC deployment...
But there's help inside the Nuxt.js docs: https://nuxtjs.org/docs/2.x/configuration-glossary/configuration-env We can use the env
Property:
Nuxt.js lets you create environment variables client side, also to be shared from server side.
All we have to do is to add the following lines to our nuxt.config.js:
env: {
baseUrl: process.env.BASE_URL || 'http://localhost:8098'
}
We can even define a default using ||
which comes in really handy for local development where our Spring Boot backend runs on http://localhost:8098
!
Now inside our axios.ts we can use the env
property like this:
import axios, {AxiosResponse} from 'axios'
const axiosApi = axios.create({
baseURL: process.env.baseUrl,
timeout: 1000,
headers: {
'Access-Control-Allow-Origin': '*',
'Content-Type': 'application/json'
}
});
The baseUrl
property that will be equal to the BASE_URL
server side environment variable if available or defined.
Now on your local dev machine you can create a BASE_URL
env var like this:
export BASE_URL=http://fargatealb-81c02c2-1301929463.eu-central-1.elb.amazonaws.com:8098/api
Running npm run dev
should now use the Fargate URL as definded.
Generate Static Site from Nuxt.js application
Now we should generate our Nuxt.js static site. So inside our project's root directory let's run:
npm run generate
This will result in a normal NPM build, followed by the static site generation:
> generate
> nuxt generate
Nuxt 3.9.3 with Nitro 2.8.1 2:33:39 PM
ℹ Using Nitro server preset: static 2:33:39 PM
ℹ Building client... 2:33:41 PM
ℹ vite v5.0.11 building for production... 2:33:41 PM
ℹ ✓ 1553 modules transformed. 2:33:44 PM
Inspect report generated at /home/jonashackt/dev/microservice-ui-nuxt-js/.nuxt/analyze/.vite-inspect
ℹ .nuxt/dist/client/manifest.json 2.94 kB │ gzip: 0.58 kB 2:33:44 PM
...
ℹ .nuxt/dist/client/_nuxt/useAPIFetch.77dgfZ2z.js 12.37 kB │ gzip: 5.06 kB 2:33:44 PM
ℹ .nuxt/dist/client/_nuxt/Element.70MerkGn.js 29.38 kB │ gzip: 10.27 kB 2:33:44 PM
ℹ .nuxt/dist/client/_nuxt/entry.FHaf2s3N.js 158.96 kB │ gzip: 60.07 kB 2:33:44 PM
ℹ ✓ built in 3.46s 2:33:44 PM
✔ Client built in 3473ms 2:33:44 PM
ℹ Building server... 2:33:44 PM
ℹ vite v5.0.11 building SSR bundle for production... 2:33:44 PM
ℹ ✓ 825 modules transformed. 2:33:46 PM
Inspect report generated at /home/jonashackt/dev/microservice-ui-nuxt-js/.nuxt/analyze/.vite-inspect
ℹ .nuxt/dist/server/_nuxt/entry-styles.r_RyA2SE.mjs 0.08 kB 2:33:46 PM
ℹ .nuxt/dist/server/_nuxt/User-styles.J7VmNqcw.mjs 0.14 kB 2:33:46 PM
...
ℹ .nuxt/dist/server/_nuxt/Element-vW6V9Woz.js 40.91 kB │ map: 72.14 kB 2:33:46 PM
ℹ .nuxt/dist/server/server.mjs 49.94 kB │ map: 115.45 kB 2:33:46 PM
ℹ ✓ built in 2.19s 2:33:46 PM
✔ Server built in 2198ms 2:33:46 PM
ℹ Initializing prerenderer nitro 2:33:46 PM
ℹ Prerendering 6 initial routes with crawler nitro 2:33:47 PM
├─ /Service (72ms) nitro 2:33:47 PM
├─ /200.html (71ms) nitro 2:33:47 PM
├─ /User (73ms) nitro 2:33:47 PM
├─ /404.html (71ms) nitro 2:33:47 PM
├─ /Element (78ms) nitro 2:33:47 PM
├─ / (76ms) nitro 2:33:47 PM
├─ /service (17ms) nitro 2:33:47 PM
├─ /user (8ms) nitro 2:33:47 PM
├─ /element (14ms) nitro 2:33:47 PM
├─ /Service/_payload.json (18ms) nitro 2:33:47 PM
├─ /User/_payload.json (7ms) nitro 2:33:47 PM
├─ /Element/_payload.json (1ms) nitro 2:33:47 PM
├─ /_payload.json (0ms) nitro 2:33:47 PM
├─ /service/_payload.json (0ms) nitro 2:33:47 PM
├─ /user/_payload.json (0ms) nitro 2:33:47 PM
├─ /element/_payload.json (1ms) nitro 2:33:47 PM
✔ Generated public .output/public nitro 2:33:47 PM
✔ You can preview this build using npx serve .output/public nitro 2:33:47 PM
✔ You can now deploy .output/public to any static hosting!
Have a look into the .output/public
folder - it should contain all files necessary for your site to host in a static hosting service like AWS S3 (or GitHub Pages etc.).
Deploy Static Site Generated Nuxt.js app to AWS S3 with Pulumi
https://www.pulumi.com/docs/reference/pkg/aws/s3/bucket/
https://www.pulumi.com/docs/reference/pkg/aws/s3/bucketobject/
https://www.pulumi.com/docs/tutorials/aws/s3-website/
Let's create a separate deployment
directory for our Pulumi sources (since we can't override our package.json
etc. of our root project):
mkdir deployment && cd deployment
pulumi new aws-typescript
I named the Pulumi project after my root project microservice-ui-nuxt-js-deployment
Now inside our deployment/index.ts Pulumi TypeScript program let's create an S3 Bucket for static website hosting:
import * as pulumi from "@pulumi/pulumi";
import * as aws from "@pulumi/aws";
// Create an AWS resource (S3 Bucket)
const nuxtBucket = new aws.s3.Bucket("microservice-ui-nuxt-js-hosting-bucket", {
acl: "public-read",
website: {
indexDocument: "index.html",
}
});
// S3 Objects from Nuxt.js static site generation will be added through aws CLI instead of Pulumi like this
// (see https://www.pulumi.com/docs/tutorials/aws/aws-ts-static-website/#deployment-speed):
// aws s3 sync ../.output/public/ s3://$(pulumi stack output bucketName) --acl public-read
// Export the name of the bucket
export const bucketName = nuxtBucket.id;
export const bucketUrl = nuxtBucket.websiteEndpoint;
And for every file inside the Nuxt.js target dir .output/public
we create a new S3 object inside the S3 Bucket.
But this shouldn't be done using Pulumi's
BucketObject
for multiple files really - see this so Q&A for more details: https://stackoverflow.com/questions/67318524/pulumi-typescript-aws-how-to-upload-multiple-files-to-s3-incl-nested-files
Instead we should use AWS CLI directly to copy (and later incrementally sync, when new builds ran) our static website files to our S3 Bucket like this:
aws s3 sync ../.output/public/ s3://$(pulumi stack output bucketName) --acl public-read
Using $(pulumi stack output bucketName) we simply get the S3 Bucket name that was created by Pulumi. Mind the --acl public-read parameter at the end, since you have to enable public read access on each of your static web files in S3, although the Bucket itself already has public read access!
Before we finally run our Pulumi program, make sure to have an apropriate AWS
AWS_ACCESS_KEY_ID
andAWS_SECRET_ACCESS_KEY
configured. If you don't have them, you can generate them inside theIAM
service for your AWS user in the AWS management console. Make sure to runaws configure
to configure both to your local terminal.
Now it's time to run our Pulumi deployment. cd
into deployment
and run:
pulumi stack select dev
pulumi up
Use Pulumi with GitHub Actions CI
As already described here: https://github.com/jonashackt/azure-training-pulumi#pulumi-with-github-actions there are some steps to take in order to use Pulumi with GitHub Actions.
https://www.pulumi.com/docs/guides/continuous-delivery/github-actions/
It's really cool to see that there's a Pulumi GitHub action project https://github.com/pulumi/actions already ready for us.
Create needed GitHub Repository Secrets
First we need to create 5 new GitHub Repository Secrets (encrypted variables) in your repo under Settings/Secrets
.
We should start to create a new Pulumi Access Token PULUMI_ACCESS_TOKEN
at https://app.pulumi.com/jonashackt/settings/tokens
Now we need to create the AWS specific variables: AWS_ACCESS_KEY_ID
and AWS_SECRET_ACCESS_KEY
(they need to be exactly named like this, see https://www.pulumi.com/docs/intro/cloud-providers/aws/setup/#environment-variables). Create them all as GitHub Repository Secrets.
There should be all these vars defined:
Create GitHub Actions workflow
Let's create a GitHub Actions workflow static-site-pulumi-aws-s3-deploy.yml:
name: static-site-pulumi-aws-s3-deploy
on:
push:
pull_request:
env:
AWS_ACCESS_KEY_ID: ${{ secrets.AWS_ACCESS_KEY_ID }}
AWS_SECRET_ACCESS_KEY: ${{ secrets.AWS_SECRET_ACCESS_KEY }}
PULUMI_ACCESS_TOKEN: ${{ secrets.PULUMI_ACCESS_TOKEN }}
jobs:
static-site-pulumi-aws-s3-deploy:
runs-on: ubuntu-latest
steps:
- name: Checkout
uses: actions/checkout@master
- name: Setup node env
uses: actions/[email protected]
with:
node-version: '14'
- name: Cache node_modules
uses: actions/cache@v2
with:
path: ~/.npm
key: ${{ runner.os }}-node-${{ hashFiles('**/package-lock.json') }}
restore-keys: |
${{ runner.os }}-node-
- name: Install Nuxt.js dependencies
run: npm install
- name: Install Pulumi dependencies before npm run generate to prevent it from breaking the build
run: npm install
working-directory: ./deployment
- name: Run tests
run: npm run test
- name: Generate Static Site from Nuxt.js application
run: npm run generate
- name: Install Pulumi CLI
uses: pulumi/[email protected]
- name: Run pulumi preview & pulumi up
run: |
pulumi stack select dev
pulumi preview
pulumi up -y
working-directory: ./deployment
- name: Configure AWS credentials for GitHub pre-installed AWS CLI
uses: aws-actions/configure-aws-credentials@v1
with:
aws-access-key-id: ${{ secrets.AWS_ACCESS_KEY_ID }}
aws-secret-access-key: ${{ secrets.AWS_SECRET_ACCESS_KEY }}
aws-region: eu-central-1
- name: Deploy Nuxt.js generated static site to S3 Bucket via AWS CLI
run: |
aws s3 sync ../.output/public/ s3://$(pulumi stack output bucketName) --acl public-read
echo "Access the Nuxt.js app at the following URL:"
pulumi stack output bucketUrl
working-directory: ./deployment
We use the possibility to define the environment variables on the workflow's top level to reduce the 3 definition to one.
Then after using the great cache Action, we need to install our Nuxt project's dependencies. And we also need to install our Pulumi project's npm packages - otherwise npm run generate
(which generates the Nuxt.js Static Site) won't work (see this build for example):
$ Run npm run generate
> [email protected] generate /home/runner/work/microservice-ui-nuxt-js/microservice-ui-nuxt-js
> nuxt generate
[fatal] Nuxt build error
ERROR in deployment/index.ts:1:25
TS2307: Cannot find module '@pulumi/pulumi' or its corresponding type declarations.
> 1 | import * as pulumi from "@pulumi/pulumi";
| ^^^^^^^^^^^^^^^^
2 | import * as aws from "@pulumi/aws";
3 |
4 | // Create an AWS resource (S3 Bucket)
ERROR in deployment/index.ts:2:22
TS2307: Cannot find module '@pulumi/aws' or its corresponding type declarations.
1 | import * as pulumi from "@pulumi/pulumi";
> 2 | import * as aws from "@pulumi/aws";
| ^^^^^^^^^^^^^
3 |
4 | // Create an AWS resource (S3 Bucket)
5 | const nuxtBucket = new aws.s3.Bucket("microservice-ui-nuxt-js-hosting-bucket", {
╭─────────────────────────────╮
│ │
│ ✖ Nuxt Fatal Error │
│ │
│ Error: Nuxt build error │
│ │
╰─────────────────────────────╯
npm ERR! code ELIFECYCLE
npm ERR! errno 1
npm ERR! [email protected] generate: `nuxt generate`
npm ERR! Exit status 1
npm ERR!
npm ERR! Failed at the [email protected] generate script.
npm ERR! This is probably not a problem with npm. There is likely additional logging output above.
npm ERR! A complete log of this run can be found in:
npm ERR! /home/runner/.npm/_logs/2021-05-04T09_54_52_203Z-debug.log
Error: Process completed with exit code 1.
The nuxt generate seems to look for package.json
files in subdirectories also.
After the obligatory jest test run via npm run test
, we finally need to generate our Nuxt.js static site files with
npm run generate
This will generate all files into the .output/public
directory we'll later use with the AWS CLI to sync into our S3 Bucket.
After having installed the Pulumi CLI with the pulumi/action-install-pulumi-cli action, we can use Pulumi to create our AWS resources - which is our static website hosting enabled S3 Bucket mainly.
We also use GitHub Actions ability to define a working-directory: ./deployment
(as discussed here) so that we can issue our pulumi up
in the right directory.
To not run into problems using the pre-installed AWS CLI on GitHub Actions we should also configure our AWS credentials using the aws-actions/configure-aws-credentials action. Otherwise we'll run into errors like this:
Run aws s3 sync ../.output/public/ s3://$(pulumi stack output bucketName) --acl public-read
<botocore.awsrequest.AWSRequest object at 0x7f01644bd070>
Finally we can use the AWS CLI to sync our Nuxt.js generated static site files into the Pulumi created S3 Bucket!
Create GitHub environment with the S3 Bucket URL
Running a workflow that references an environment that does not exist will create an environment with the referenced name. The newly created environment will not have any protection rules or secrets configured. Anyone that can edit workflows in the repository can create environments via a workflow file, but only repository admins can configure the environment.
So first we need to define the environment inside our static-site-pulumi-aws-s3-deploy.yml:
environment:
name: microservice-ui-nuxt-js-deployment
url: ${{ steps.step_name.outputs.url_output }}
Now we need to give our step Deployment step a id
so that we can reference it inside the environment:url
. Also we need to
set a variable like s3_url
that will hold the S3 Buckets url with echo "::set-output name=s3_url::http://$(pulumi stack output bucketUrl)"
(see this so answer also):
- name: Deploy Nuxt.js generated static site to S3 Bucket via AWS CLI
id: aws-sync
run: |
aws s3 sync ../.output/public/ s3://$(pulumi stack output bucketName) --acl public-read
echo "Access the Nuxt.js app at the following URL:"
pulumi stack output bucketUrl
echo "::set-output name=s3_url::http://$(pulumi stack output bucketUrl)"
working-directory: ./deployment
With this we can use the output inside our environment:url
:
environment:
name: microservice-ui-nuxt-js-deployment
url: ${{ steps.aws-sync.outputs.s3_url }}
Now we should be able to see (and click on) the URL as an environment inside the GitHub Actions UI:
Configure BASE_URL of microservice-api-spring-boot in frontend deployment
Initally let's simply define the environment variable BASE_URL
inside our GitHub Actions workflow static-site-pulumi-aws-s3-deploy.yml:
name: static-site-pulumi-aws-s3-deploy
on:
push:
pull_request:
env:
AWS_ACCESS_KEY_ID: ${{ secrets.AWS_ACCESS_KEY_ID }}
AWS_SECRET_ACCESS_KEY: ${{ secrets.AWS_SECRET_ACCESS_KEY }}
PULUMI_ACCESS_TOKEN: ${{ secrets.PULUMI_ACCESS_TOKEN }}
BASE_URL: "http://fargatealb-81c02c2-1301929463.eu-central-1.elb.amazonaws.com:8098/api"
...
Server-Side Rendering (SSR) incl. Node.js Container Build with Paketo
Here's a great description of the differences between
static
andserver
mode in Nuxt.js: https://nuxtjs.org/blog/going-full-static#commands
Nuxt.js also provides us with a server
mode inside the nuxt.config.js:
export default {
// Target: https://go.nuxtjs.dev/config-target
target: 'server',
Now instead of using nuxt generate
(configured inside our package.json
) or npm run generate
to generate a static version of our app, we should use nuxt build
- or npm run build
to trigger the SSR mode build:
$ npm run build
ℹ Production build 11:12:27
ℹ Bundling for server and client side 11:12:27
ℹ Target: server 11:12:27
ℹ Using components loader to optimize imports 11:12:27
ℹ Discovered Components: .nuxt/components/readme.md 11:12:27
✔ Builder initialized 11:12:27
✔ Nuxt files generated
...
Use Paketo.io CNB to build a Node.js Container for our Nuxt app
Be sure to have pack CLI
installed (see https://buildpacks.io/docs/tools/pack/) and then run:
pack build microservice-ui-nuxt-js --path . --builder paketobuildpacks/builder-jammy-base
I had several errors starting with node js npm ERR! missing:
and read through https://paketo.io/docs/buildpacks/language-family-buildpacks/nodejs/#npm-installation-process
Paketo runs npm rebuild
when a local node_modules
directory & package-lock.json
is present, which ran into the errors.
Simply deleting the local node_modules
folder causes Paketo to run a npm ci
instead, which worked like a charm:
$ pack build microservice-ui-nuxt-js --path . --builder paketobuildpacks/builder-jammy-base
base: Pulling from paketobuildpacks/builder
Digest: sha256:3e2ee17348bd901e7e0748e0e1ddccdf8a602b624e418927145b5f84ca26f264
Status: Image is up to date for paketobuildpacks/builder-jammy-base
base-cnb: Pulling from paketobuildpacks/run
Digest: sha256:b6b1612ab2dfa294514fff2750e8d724287f81e89d5e91209dbdd562ed7f7daf
Status: Image is up to date for paketobuildpacks/run:base-cnb
===> DETECTING
4 of 7 buildpacks participating
paketo-buildpacks/ca-certificates 2.2.0
paketo-buildpacks/node-engine 0.4.0
paketo-buildpacks/npm-install 0.3.0
paketo-buildpacks/npm-start 0.2.0
===> ANALYZING
Previous image with name "microservice-ui-nuxt-js" not found
===> RESTORING
===> BUILDING
Paketo CA Certificates Buildpack 2.2.0
https://github.com/paketo-buildpacks/ca-certificates
Launch Helper: Contributing to layer
Creating /layers/paketo-buildpacks_ca-certificates/helper/exec.d/ca-certificates-helper
Paketo Node Engine Buildpack 0.4.0
Resolving Node Engine version
Candidate version sources (in priority order):
-> ""
<unknown> -> ""
Selected Node Engine version (using ): 14.17.0
Executing build process
Installing Node Engine 14.17.0
Completed in 5.795s
Configuring build environment
NODE_ENV -> "production"
NODE_HOME -> "/layers/paketo-buildpacks_node-engine/node"
NODE_VERBOSE -> "false"
Configuring launch environment
NODE_ENV -> "production"
NODE_HOME -> "/layers/paketo-buildpacks_node-engine/node"
NODE_VERBOSE -> "false"
Writing profile.d/0_memory_available.sh
Calculates available memory based on container limits at launch time.
Made available in the MEMORY_AVAILABLE environment variable.
Paketo NPM Install Buildpack 0.3.0
Resolving installation process
Process inputs:
node_modules -> "Not found"
npm-cache -> "Not found"
package-lock.json -> "Found"
Selected NPM build process: 'npm ci'
Executing build process
Running 'npm ci --unsafe-perm --cache /layers/paketo-buildpacks_npm-install/npm-cache'
Completed in 14.988s
Configuring launch environment
NPM_CONFIG_LOGLEVEL -> "error"
Configuring environment shared by build and launch
PATH -> "$PATH:/layers/paketo-buildpacks_npm-install/modules/node_modules/.bin"
Paketo NPM Start Buildpack 0.2.0
Assigning launch processes
web: nuxt start
===> EXPORTING
Adding layer 'paketo-buildpacks/ca-certificates:helper'
Adding layer 'paketo-buildpacks/node-engine:node'
Adding layer 'paketo-buildpacks/npm-install:modules'
Adding layer 'paketo-buildpacks/npm-install:npm-cache'
Adding 1/1 app layer(s)
Adding layer 'launcher'
Adding layer 'config'
Adding layer 'process-types'
Adding label 'io.buildpacks.lifecycle.metadata'
Adding label 'io.buildpacks.build.metadata'
Adding label 'io.buildpacks.project.metadata'
Setting default process type 'web'
Saving microservice-ui-nuxt-js...
*** Images (5eb36ba20094):
microservice-ui-nuxt-js
Adding cache layer 'paketo-buildpacks/node-engine:node'
Adding cache layer 'paketo-buildpacks/npm-install:modules'
Adding cache layer 'paketo-buildpacks/npm-install:npm-cache'
Successfully built image microservice-ui-nuxt-js
Running our Paketo build Nuxt.js container (the Nuxt HOST configuration)
Now everything should be straight forward, right?! Simply run our app with:
docker run --rm -i --tty -p 3000:3000 microservice-ui-nuxt-js
But this doesn't show up in our browser!
Does it work inside the container? Install curl...
But still not in the browser
I came upon it in https://nuxtjs.org/docs/2.x/deployment/deployment-cloud-run
and then read https://medium.com/i22digital/development-setup-with-nuxt-node-and-docker-b008a241c11d
Last thing is the environment variable HOST . Nuxt needs this variable when started from a container, otherwise you won’t be able to reach it. Host 0.0.0.0 is designated to tell Nuxt to resolve a host address, which is accessible to connections outside of the host machine.
See https://stackoverflow.com/a/67871934/4964553
docker run --rm -i --tty --env "HOST=0.0.0.0" -p 3000:3000 microservice-ui-nuxt-js
Another problem arose after upgrading to Nuxt 3.x / Vue.js 3.x:
Executing the docker run
gave a:
ERROR: failed to launch: determine start command: when there is no default process a command is required
Luckily I stumbled upon the depts of the Paketo node.js docs, where it is described how to specify a custom entrypoint right at the Paketo build time - for more details see https://stackoverflow.com/a/78000323/4964553
So a working docker run
could be achieved by using the following pack build
command:
pack build ghcr.io/jonashackt/microservice-ui-nuxt-js:latest \
--builder paketobuildpacks/builder-jammy-base \
--env BP_LAUNCHPOINT=".output/server/index.mjs" \
--path .
Running the Paketo build on GitHub Actions
Now simply add a new GitHub Actions workflow server-side-rendering-nodejs-container-paketo.yml:
server-side-rendering-nodejs-container-paketo:
runs-on: ubuntu-latest
steps:
- name: Checkout 🛎
uses: actions/checkout@master
- name: Login to GitHub Container Registry
uses: docker/login-action@v1
with:
registry: ghcr.io
username: ${{ github.actor }}
password: ${{ secrets.GITHUB_TOKEN }}
- name: Install pack CLI via the official buildpack Action https://github.com/buildpacks/github-actions#setup-pack-cli-action
uses: buildpacks/github-actions/[email protected]
# Caching Paketo Build see https://stackoverflow.com/a/66598693/4964553
# BP_OCI_SOURCE as --env creates the GitHub Container Registry <-> Repository link (https://paketo.io/docs/buildpacks/configuration/#applying-custom-labels)
- name: Build app with pack CLI & publish to bc Container Registry
run: |
pack build ghcr.io/jonashackt/microservice-ui-nuxt-js:latest \
--builder paketobuildpacks/builder-jammy-base \
--path . \
--env "BP_OCI_SOURCE=https://github.com/jonashackt/microservice-ui-nuxt-js" \
--cache-image ghcr.io/jonashackt/microservice-ui-nuxt-js-paketo-cache-image:latest \
--publish
Let's try to run our container which got published to the GitHub Container Registry with:
docker run --rm -i --tty --env "HOST=0.0.0.0" -p 3000:3000 ghcr.io/jonashackt/microservice-ui-nuxt-js:latest
This will result in the following error:
$ docker run --rm -i --tty --env "HOST=0.0.0.0" -p 3000:3000 ghcr.io/jonashackt/microservice-ui-nuxt-js:latest
Unable to find image 'ghcr.io/jonashackt/microservice-ui-nuxt-js:latest' locally
latest: Pulling from jonashackt/microservice-ui-nuxt-js
d0c52a206d00: Already exists
1879a9e18f37: Already exists
1320020c14ba: Already exists
0c300ce812d0: Pull complete
09e2d34e431f: Pull complete
49948b98bd8d: Pull complete
d18fee29c74f: Pull complete
49d2052d2301: Pull complete
feeffabb64e4: Pull complete
0fae4bb31aa5: Pull complete
87f3d5cd60d8: Pull complete
aa7a1571fd83: Pull complete
0b0022959657: Pull complete
5a44e4f7b58d: Pull complete
Digest: sha256:1d1e3125a8fdab6136b490891c7d9717bb9e108a3df870f55f77f64ecea4d597
Status: Downloaded newer image for ghcr.io/jonashackt/microservice-ui-nuxt-js:latest
FATAL No build files found in /workspace/.nuxt/.output/public/server. 13:09:38
Use either `nuxt build` or `builder.build()` or start nuxt in development mode.
Use either `nuxt build` or `builder.build()` or start nuxt in development mode.
at VueRenderer._ready (/layers/paketo-buildpacks_npm-install/modules/node_modules/@nuxt/vue-renderer/.output/public/vue-renderer.js:758:13)
at async Server.ready (/layers/paketo-buildpacks_npm-install/modules/node_modules/@nuxt/server/.output/public/server.js:637:5)
at async Nuxt._init (/layers/paketo-buildpacks_npm-install/modules/node_modules/@nuxt/core/.output/public/core.js:482:7)
╭─────────────────────────────────────────────────────────────────────────────────────╮
│ │
│ ✖ Nuxt Fatal Error │
│ │
│ Error: No build files found in /workspace/.nuxt/.output/public/server. │
│ Use either `nuxt build` or `builder.build()` or start nuxt in development mode. │
│ │
╰─────────────────────────────────────────────────────────────────────────────────────╯
It seems that our Paketo build only runs npm install
or npm ci
, but doesn't run npm run generate
(which in turn runs nuxt generate
) to generate our site.
So in order to get our Container working, we need to run the npm run generate
first (as we did locally) - and then the Paketo build should pick up our build artifacts.
This is not ideal, since there should be everything handled inside the container build - but reading the docs I don't see a hook where I could place the npm run generate
inside the Paketo build.
So here's the full GitHub Action job:
server-side-rendering-nodejs-container-paketo:
runs-on: ubuntu-latest
steps:
- name: Checkout 🛎
uses: actions/checkout@master
- name: Setup node env 🏗
uses: actions/[email protected]
with:
node-version: '14'
- name: Cache node_modules 📦
uses: actions/cache@v2
with:
path: ~/.npm
key: ${{ runner.os }}-node-${{ hashFiles('**/package-lock.json') }}
restore-keys: |
${{ runner.os }}-node-
- name: Install Nuxt.js dependencies
run: npm install
- name: Install Pulumi dependencies before npm run generate to prevent it from breaking the build
run: npm install
working-directory: ./deployment
- name: Run tests 🧪
run: npm run test
- name: Generate Static Site from Nuxt.js application (nuxt build, see package.json)
run: npm run build
- name: Login to GitHub Container Registry
uses: docker/login-action@v1
with:
registry: ghcr.io
username: ${{ github.actor }}
password: ${{ secrets.GITHUB_TOKEN }}
- name: Install pack CLI via the official buildpack Action https://github.com/buildpacks/github-actions#setup-pack-cli-action
uses: buildpacks/github-actions/[email protected]
# Caching Paketo Build see https://stackoverflow.com/a/66598693/4964553
# BP_OCI_SOURCE as --env creates the GitHub Container Registry <-> Repository link (https://paketo.io/docs/buildpacks/configuration/#applying-custom-labels)
- name: Build app with pack CLI & publish to bc Container Registry
run: |
pack build ghcr.io/jonashackt/microservice-ui-nuxt-js:latest \
--builder paketobuildpacks/builder-jammy-base \
--path . \
--env "BP_OCI_SOURCE=https://github.com/jonashackt/microservice-ui-nuxt-js" \
--cache-image ghcr.io/jonashackt/microservice-ui-nuxt-js-paketo-cache-image:latest \
--publish
Links
Nuxt.js TypeScript Components cookbook: https://typescript.nuxtjs.org/cookbook/components/
VSCode with Vetur Plugin: https://marketplace.visualstudio.com/items?itemName=octref.vetur