test(e2e): Add e2e test for AWS lambda in ESM mode #12833
Conversation
| @@ -23,7 +23,8 @@ | |||
| "esbuild": "0.20.0", | |||
| "glob": "8.0.3", | |||
| "ts-node": "10.9.1", | |||
| "yaml": "2.2.2" | |||
| "yaml": "2.2.2", | |||
| "rimraf": "^5.0.0" | |||
There was a problem hiding this comment.
not sure why but without adding rimraf explicitly here, I can't use the yarn test:run command to run e2e tests locally as rimraf isn't found otherwise 🤔
Lms24
force-pushed
the
lms/test-aws-serverless-esm
branch
from
5b962b7 to
8d1fa8f
Compare
|
|
||
| const handler = Sentry.wrapHandler(async () => { | ||
| await new Promise(resolve => { | ||
| const req = http.request( |
There was a problem hiding this comment.
took me ages to find out why the previously used http.get call didn't produce a span. Turns out it's an Otel ESM bug. Reported it to Otel here: open-telemetry/opentelemetry-js#4857
Anyway, this http.request call now emits a span which we assert on in the test. So this shows that otel instrumentation generally works
| { | ||
| /* Run your local dev server before starting the tests */ | ||
| webServer: [ | ||
| { | ||
| command: `node start-event-proxy.mjs && pnpm wait-port ${eventProxyPort}`, | ||
| port: eventProxyPort, | ||
| stdout: 'pipe', | ||
| }, | ||
| ], | ||
| }, | ||
| ); |
There was a problem hiding this comment.
l: Is this needed? The base config is already doing this ootb.
There was a problem hiding this comment.
Right now, it somewhat is necessary because the shared playwright config always expects an additional startCommand to start an application/development server. For our AWS Lambda tests, we don't need this because we invoke the function from within the test file. I'll open a separate PR to make startCommand optional since this concerns all e2e tests relying on the shared config.
There was a problem hiding this comment.
Ah, understood. Thanks for the explanation!
This PR adds an E2E test app that simulates running an AWS lambda function in ESM mode. At the same time, this test app represents the state of how to manually use the
@sentry/aws-serverlessSDK for ESM-based lambda functions. For the moment, the plan is to document this as follows:@sentry/aws-serverlesswith the package manager of your choiceNODE_OPTIONS: '--import @sentry/aws-serverless/awslambda-autoSENTRY_DSN: '<your-dsn>'SENTRY_TRACES_SAMPLE_RATE: '1.0'Sentry.wrapHandlernode_modulesdirectory is deployed alongside your lambda codeUnfortunately, we can't get around highest friction points:
Sentry.wrapHandlerwrappernode_modulesare deployed alongside lambda codeThis, we'll only get rid off with an ESM lambda layer, and possibly not at all for the
Sentry.wrapHandlerpart.ref #12409