ContextQMD
Back to Kibana

Documentation for ML UI developers

x-pack/platform/plugins/shared/ml/readme.md

9.4.08.3 KB
Original Source

Documentation for ML UI developers

This plugin provides access to the machine learning features provided by Elastic.

Requirements

To use machine learning features, you must have a Platinum or Enterprise license or a free 14-day Trial. File Data Visualizer requires a Basic license. For more info, refer to Set up machine learning features.

Setup local environment

Kibana

  1. Fork and clone the Kibana repo.

  2. Install nvm, node, yarn (for example, by using Homebrew). See Install dependencies.

  3. Make sure that Elasticsearch is deployed and running on localhost:9200.

  4. Navigate to the directory of the kibana repository on your machine.

  5. Fetch the latest changes from the repository.

  6. Checkout the branch of the version you want to use. For example, if you want to use a 7.9 version, run git checkout 7.9.

  7. Run nvm use. The response shows the Node version that the environment uses. If you need to update your Node version, the response message contains the command you need to run to do it.

  8. Run yarn kbn bootstrap. It takes all the dependencies in the code and installs/checks them. It is recommended to use it every time when you switch between branches.

  9. Make a copy of kibana.yml and save as kibana.dev.yml. (Git will not track the changes in kibana.dev.yml but yarn will use it.)

  10. Provide the appropriate password and user name in kibana.dev.yml.

  11. Run yarn start to start Kibana.

  12. Go to http://localhost:560x/xxx (check the terminal message for the exact path).

For more details, refer to this getting started page.

Adding sample data to Kibana

Kibana has sample data sets that you can add to your setup so that you can test different configurations on sample data.

  1. Click the Elastic logo in the upper left hand corner of your browser to navigate to the Kibana home page.

  2. Click Load a data set and a Kibana dashboard.

  3. Pick a data set or feel free to click Add on all of the available sample data sets.

These data sets are now ready be analyzed in ML jobs in Kibana.

Running tests

Jest tests

Documentation: https://www.elastic.co/guide/en/kibana/current/development-tests.html#_unit_testing

Run the test following jest tests from kibana/x-pack/platform/plugins/shared/ml.

New snapshots, all plugins:

yarn test:jest

Update snapshots for the ML plugin:

yarn test:jest -u

Update snapshots for a specific directory only:

yarn test:jest public/application/settings/filter_lists

Run tests with verbose output:

yarn test:jest --verbose

Functional tests

Before running the test server, make sure to quit all other instances of Elasticsearch.

Run the following commands from the x-pack directory and use separate terminals for test server and test runner. The test server command starts an Elasticsearch and Kibana instance that the tests will be run against.

Functional tests are broken up into independent groups with their own configuration. Test server and runner need to be pointed to the configuration to run. The basic commands are

node scripts/functional_tests_server.js --config PATH_TO_CONFIG
node scripts/functional_test_runner.js --config PATH_TO_CONFIG

With PATH_TO_CONFIG and other options as follows.

  1. Functional UI tests with Trial license:

    GroupPATH_TO_CONFIG
    anomaly detection jobs group1src/platform/test/functional/apps/ml/anomaly_detection_jobs/group1/config.ts
    anomaly detection jobs group2src/platform/test/functional/apps/ml/anomaly_detection_jobs/group2/config.ts
    anomaly detection jobs group3src/platform/test/functional/apps/ml/anomaly_detection_jobs/group3/config.ts
    anomaly detection jobs group4src/platform/test/functional/apps/ml/anomaly_detection_jobs/group4/config.ts
    anomaly detection result viewssrc/platform/test/functional/apps/ml/anomaly_detection_result_views/config.ts
    anomaly detection integrationssrc/platform/test/functional/apps/ml/anomaly_detection_integrations/config.ts
    data frame analyticssrc/platform/test/functional/apps/ml/data_frame_analytics/config.ts
    data visualizersrc/platform/test/functional/apps/ml/data_visualizer/config.ts
    permissionssrc/platform/test/functional/apps/ml/permissions/config.ts
    stack management jobssrc/platform/test/functional/apps/ml/stack_management_jobs/config.ts
    short testssrc/platform/test/functional/apps/ml/short_tests/config.ts

    The short tests group contains tests for page navigation, model management, feature controls, settings and notifications. Test files for each group are located in the directory of their configuration file.

  2. Functional UI tests with Basic license:

    GroupPATH_TO_CONFIG
    permissionsplatform/test/functional_basic/apps/ml/permissions/config.ts
    data visualizer group1 (file data viz)platform/test/functional_basic/apps/ml/data_visualizer/group3/config.ts
    data visualizer group2 (index data viz)platform/test/functional_basic/apps/ml/data_visualizer/group2/config.ts
    data visualizer group3 (actions panel, discover grid)platform/test/functional_basic/apps/ml/data_visualizer/group3/config.ts

  3. API integration tests with Trial license:

    • PATH_TO_CONFIG: x-pack/platform/test/api_integration/apis/ml/config.ts

  4. Accessibility tests:

    We maintain a suite of accessibility tests (you may see them referred to elsewhere as a11y tests). These tests render each of our pages and ensure that the inputs and other elements contain the attributes necessary to ensure all users are able to make use of ML (for example, users relying on screen readers).

    • PATH_TO_CONFIG: src/platform/test/accessibility/config.ts
    • Add --grep=ml to the test runner command
    • Tests are located in x-pack/platform/test/accessibility/apps/group2

Generating docs screenshots

The screenshot generation uses the functional test runner described in the Functional tests section above.

Run the following commands from the x-pack directory and use separate terminals for test server and test runner. The test server command starts an Elasticsearch and Kibana instance that the tests will be run against.

node scripts/functional_tests_server.js --config platform/test/screenshot_creation/config.ts
node scripts/functional_test_runner.js --config platform/test/screenshot_creation/config.ts --include-tag ml

The generated screenshots are stored in x-pack/platform/test/functional/screenshots/session/ml_docs. ML screenshot generation tests are located in x-pack/platform/test/screenshot_creation/apps/ml_docs.

Shared functions

Note: We are in the process of moving shared code to packages, for example @kbn/ml-anomaly-utils. We'll update the docs accordingly once this work is done.

You can find the ML shared functions in the following files in GitHub:

https://github.com/elastic/kibana/blob/main/x-pack/platform/plugins/shared/ml/public/index.ts

https://github.com/elastic/kibana/blob/main/x-pack/platform/plugins/shared/ml/server/index.ts

These functions are shared from the root of the ML plugin, you can import them with an import statement. For example:

import { MlPluginSetup } from '@kbn/ml-plugin/server';

or

import { ML_ANOMALY_SEVERITY } from '@kbn/ml-plugin/common';

Functions are shared from the following directories:

ml/common
ml/public
ml/server