362d540aac
This PR mixes various initiatives to improve visibility on sentry **1. Catch errors on workflow jobs** commit [catch workflowTriggerExceptions in job handle](1dbba8c9e2) @thomtrp **2. Fix type in messagingImportExceptionHandler** commit [fix type issue on messagingImportExceptionHandler](919bb3844c) @guillim **3. Catch invalid uuid errors thrown by Postgres by rightfully typing expected id as uuid** commits [use UUIDFilter instead of IDFilter to get graphqlError in case of malformed id](57cc315efe), [use UUIDFilter (2)](304553d770), [fix ids typed as UUID instead of ID](f95d6319cf) @Weiko ⚠️⚠️⚠️ when we deploy this PR we need to flush the schema types from redis as this PR changes them ⚠️⚠️⚠️ **4. Do not group UNKNOWN errors together** commit [do not group unknown errors together](c299b39c8f) Some CustomException classes have introduced UNKNOWN error codes as a default fallback error code. We use CustomException codes to group issues together, but we don't want to do it with UNKNOWN error as they may not have anything in common. For exemple [this sentry for UNKNOWN code](https://twenty-v7.sentry.io/issues/6605750776/events/a72272d8941b4fa2add9b1f39c196d3f/?environment=prod&environment=prod-eu&project=4507072499810304&query=Unknown&referrer=next-event&stream_index=0) groups together "Unknown error importing calendar events for calendar channel...", "Insufficent permissions...", to name a few. **5. Improve postgres error grouping** commit [group together postgres errors](567c25495e) Postgres error are thrown by typeORM as QueryFailedError. we have a lot of them on sentry where they are badly grouped They are currently grouped on sentry according to the stack trace, which leads them to sometimes be grouped even if they don't have anything in common : for exemple [this sentry for QueryFailedError](https://twenty-v7.sentry.io/issues/6563624590/events/2d636821e27a448595b647b4b5a7d6a8/?environment=prod&environment=prod-eu&project=4507072499810304&query=is%3Aunresolved%20%21issue.type%3A%5Bperformance_consecutive_db_queries%2Cperformance_consecutive_http%2Cperformance_file_io_main_thread%2Cperformance_db_main_thread%2Cperformance_n_plus_one_db_queries%2Cperformance_n_plus_one_api_calls%2Cperformance_p95_endpoint_regression%2Cperformance_slow_db_query%2Cperformance_render_blocking_asset_span%2Cperformance_uncompressed_assets%2Cperformance_http_overhead%2Cperformance_large_http_payload%5D%20timesSeen%3A%3E10&referrer=previous-event&sort=date&stream_index=0) groups together "user mapping not found for "postgres" and "invalide type for uuid: 'fallback-id'" to name a few. I attempted to improve the grouping by grouping them with a new custom fingerPrint composed of the [code returned by Postgres](https://www.postgresql.org/docs/current/errcodes-appendix.html) + the truncated operation name (Find, Aggregate, Check...). This is still not ideal as postgres code are quite broad - we could have the same error code for two Find operations with different causes. let's give this a try !
Twenty end-to-end (E2E) Testing
Prerequisite
Installing the browsers:
npx nx setup twenty-e2e-testing
Run end-to-end tests
npx nx test twenty-e2e-testing
Start the interactive UI mode
npx nx test:ui twenty-e2e-testing
Run test in specific file
npx nx test twenty-e2e-testing <filename>
Example (location of the test must be specified from the root of twenty-e2e-testing package):
npx nx test twenty-e2e-testing tests/login.spec.ts
Runs the tests in debug mode.
npx nx test:debug twenty-e2e-testing
Show report after tests
npx nx test:report twenty-e2e-testing
Q&A
Why there's path.resolve() everywhere?
That's thanks to differences in root directory when running tests using commands and using IDE. When running tests with commands,
the root directory is twenty/packages/twenty-e2e-testing, for IDE it depends on how someone sets the configuration. This way, it
ensures that no matter which IDE or OS Shell is used, the result will be the same.