evalaas is a serverless prototyping platform on top of Google Cloud Run which encourages the Save and Refresh development cycle, giving rapid feedback when developing serverless endpoints.
This file is an executable specification. The steps in this tutorial is continuously tested on GitHub Actions using Gauge. You can view the latest test result here.
This guide assumes that evalaas is already deployed.
Deploy a JavaScript endpoint
00:00:00After setting up, you can deploy endpoints using the PUT /admin/deploy/:endpointId endpoint with a bearer token.
Execution Time : 00:00:00
-
Compress and PUT "[example-files/hello.js](example-files/hello.js)" to "`https://test.evalaas.dev/admin/endpoints/hello`"
Execution Time : 00:00:00
-
Make a GET request to "`https://test.evalaas.dev/run/hello`"
Execution Time : 00:00:00
-
You should get the following JSON response:
json response `{ "ok": 1 }`
Updating a JavaScript endpoint
00:00:00One of the advantage of using evalaas is almost instant deploys. When you upload a new file to Google Cloud Storage, the change takes effect immediately. This is in contrast with some other serverless providers where you have to wait about 1 minute to get your code updated.
First, upload the 1st version:
Execution Time : 00:00:00
-
Compress and PUT "[example-files/hello.js](example-files/hello.js)" to "`https://test.evalaas.dev/admin/endpoints/hello`"
Execution Time : 00:00:00
-
Make a GET request to "`https://test.evalaas.dev/run/hello`"
Execution Time : 00:00:00
-
You should get the following JSON response:
json response `{ "ok": 1 }`
Then, upload the 2nd version:
Execution Time : 00:00:00
-
Compress and PUT "[example-files/hello-v2.js](example-files/hello-v2.js)" to "`https://test.evalaas.dev/admin/endpoints/hello`"
Execution Time : 00:00:00
-
Make a GET request to "`https://test.evalaas.dev/run/hello`"
Execution Time : 00:00:00
-
You should get the following JSON response:
json response `{ "ok": "it is working!" }`
Environment variables
00:00:00Since multiple endpoints may be run on the same Node.js process (unless you deploy multiple instances of evalaas), they all share the same set of system environment variables (process.env), so process.env cannot be customized per-endpoint.
However, we still may want to give each endpoint a different configuration, but we might not want to hard-code them into the source code or check them into source control. Alternatively, you can upload a .env file next to the .js.gz file, and it will be made available in the endpoint as req.env.
Execution Time : 00:00:00
-
Compress and PUT "[example-files/env-example.js](example-files/env-example.js)" to "`https://test.evalaas.dev/admin/endpoints/env-example`"
Execution Time : 00:00:00
-
PUT to "`https://test.evalaas.dev/admin/env/env-example`" with the following contents
`env-example.env` # this is a comment A=hello B="world"
Execution Time : 00:00:00
-
Make a GET request to "`https://test.evalaas.dev/run/env-example`"
Execution Time : 00:00:00
-
You should get the following JSON response:
json response `{ "env": { "A": "hello", "B": "world" } }`
Source map support
00:00:00To make compiled endpoints easy to debug, evalaas supports inline source maps.
Execution Time : 00:00:00
-
Compress and PUT "[example-files/source-map-example.js](example-files/source-map-example.js)" to "`https://test.evalaas.dev/admin/endpoints/source-map-example`"
Execution Time : 00:00:00
-
Make a GET request to "`https://test.evalaas.dev/run/source-map-example`"
Execution Time : 00:00:00
-
You should find these strings in the response:
text to find `src/f.js` `src/index.js`
You can use webpack to generate example-files/source-map-example.js with these files:
src/index.js:
import f from './f'
export default (req, res) => {
res.json({ stack: f() })
}
src/f.js:
export default function foo() {
return bar()
}
function bar() {
return new Error('test source map').stack
}
webpack.config.js:
module.exports = {
entry: './src/index.js',
devtool: 'inline-source-map',
target: 'node',
mode: 'production',
output: {
path: `${__dirname/dist}`,
filename: 'source-map-example.js',
library: 'endpoint',
libraryTarget: 'umd',
},
}
Then run:
yarn add --dev webpack webpack-cli && yarn webpack
Error handling
00:00:00If your endpoint function throws or returns a rejected promise, evalaas will send a response of status 500.
Execution Time : 00:00:00
-
Compress and PUT "[example-files/error-example.js](example-files/error-example.js)" to "`https://test.evalaas.dev/admin/endpoints/error-example`"
Execution Time : 00:00:00
-
Make a GET request to "`https://test.evalaas.dev/run/error-example`"
Execution Time : 00:00:00
-
You should get a response with status code "500"
Make HTTP requests with `fetch()`
00:00:00You can use fetch() to call other HTTP endpoints without installing any extra server.
Execution Time : 00:00:00
-
Compress and PUT "[example-files/fetch-example.js](example-files/fetch-example.js)" to "`https://test.evalaas.dev/admin/endpoints/fetch-example`"
Execution Time : 00:00:00
-
Make a GET request to "`https://test.evalaas.dev/run/fetch-example`"
Execution Time : 00:00:00
-
You should get the following JSON response:
json response `{ "dependenciesCount": 3 }`