@ytsaurus/ui

YTsaurus platform interface

User interface for a YTsaurus platform cluster.

How to work with the repo

First of all you have to provide clusters-config.json file with description of your cluster, the file should be placed in the root of the repository (see clusters-config.json.example).

Additionally you have to provide secrets/yt-interface-secret.json file with a token of a special user (UIRobot) for some service requests like handling settings and other, example:

{
  // common oauth token, the token is used if there is no cluster-specific token in the file
  "oauthToken": "special-user-secret-tocken",
  "cluster_name1": {
    // cluster_name1 specific oauth token
    "oauthToken": "cluster1-special-user-secret-token"
  },
  "cluster_name2": {
    // cluster_name2 specific oauth token
    "oauthToken": "cluster2-special-user-secret-token"
  }
}

Development

To run the development environment you need to prepare and run nginx:

1. Install nginx
2. Copy file deploy/nginx/yt.development.conf.example to /etc/nginx/sites-enabled/yt.development.conf
3. Modify /etc/nginx/sites-enabled/yt.development.conf
   • change server_name
   • replace all /path/to/the/repo to correct path
4. sudo systemctl restart nginx

Install required dependencies:

$ npm deps:install

A simple way to start development is using your local-yt cluster. The command bellow suggests to start local-yt cluster as docker-container:

$ npm run dev:localmode

Another way is to provide clusters-config.json and run the command like:

$ npm run dev:app

Development without nginx

You also could run application without nginx, right on localhost port. All that you need to do is specify port:

LOCAL_DEV_PORT=8080 YT_AUTH_ALLOW_INSECURE=true ALLOW_PASSWORD_AUTH=true npm run dev:oss

Docker

There is ability to build docker-image:

$ docker build . -t ytsaurus-ui:my-tag

All application files in a resulting docker-image will be placed in /opt/app, so you have to mount /opt/app/cluster-config.json and /opt/app/secrets/yt-interface-secret.json.

Environment variables

• YT_AUTH_ALLOW_INSECURE - if defined allows insecure (over http) authentication, do not use it for production
• ALLOW_PASSWORD_AUTH - If defined, the app requires a password for cluster access
• PROMETHEUS_BASE_URL - If defined enables monitoring dashboards with prometheus data, see Prometheus dashboards below (the value is forwarded to YTCoreConfig.prometheusBaseUrl). To use authorized requests to your prometheus instance provide prometheusAuthHeaders field in secrets/yt-interface-secret.json, like:

{
  "prometheusAuthHeaders": {"Authorization": "Basic xxxxxxx"}
}

• GRAFANA_BASE_URL - If defined enables links to specific dashbords when YTCoreConfig.prometheusBaseUrl is provided (the value is forwarded to YTCoreConfig.uiSettings.grafanaBaseUrl). By default the link is visible for all users, to control the visibility create //sys/interface_monitoring/allow_grafana_url node and add use permission for specific users/groups.

Prometheus dashboards

To activate monitoring ddashboards you have to provide PROMETHEUS_BASE_URL. Additionally you have to provide documents in //sys/interface_monitoring/ with description of dashboards:

• Account: master-accounts
• Tablet cell bundle: bundle-ui-cpu, bundle-ui-disk, bundle-ui-efficiency, bundle-ui-lsm, bundle-ui-memory, bundle-ui-network, bundle-ui-resource, bundle-ui-rpc-proxy, bundle-ui-rpc-proxy-overview, bundle-ui-user-load,
• System: cluster-resources, master-global, master-local, scheduler-internal,
• CHYT: chyt-monitoring,
• Navigation: queue-metrics, queue-consumer-metrics, flow-general,
• Operations: scheduler-operation, job,
• Scheduling pool: scheduler-pool

Cluster-specific features

There is yt-api command get_supported_feature and it is a good place to describe API features. But some cases require the ability to turn on/off a feature manually on a specific cluster. Such cluster specific optinos are placed in:

• //sys/@ui_config (values affect all users)
• //sys/@ui_config_dev_overrides (values affect only admins)

(see more detail in YTFRONT-2804)

UI determines a user as an admin on the cluster if he has write access to admins group.

Available options (default values are highlighted in bold):

Option name	Allowed values	Description
enable_per_bundle_tablet_accounting	true, false	Allows editing of resources of tablets through BundleEditorDialog YTFRONT-2851

| enable_per_account_tablet_accounting | false, true | Allows editing of resources of tablets through AccountEditorDialog YTFRONT-2851 |

| per_bundle_accounting_help_link | null, url as string | Help link for resources of tablets to display from AccountEditorDialog about moving the resources to bundles YTFRONT-2851 |

| enable_maintenance_api_nodes | null, boolean | Allows to use add_maintenance/remove_maintenance commands from Components/Nodes page YTFRONT-3792 |

| enable_maintenance_api_proxies | null, boolean | Allows to use add_maintenance/remove_maintenance commands from Components/HTTP Proxies and Components/RPC Proxies pages YTFRONT-3792 |

| chyt_controller_base_url | null, url as string | Base url for chyt-controller |

| livy_controller_base_url | null, url as string | Base url for spyt-controller | | job_trace_url_template | null, {title: string; url_template: string; enforce_for_trees?: Array<string>} | If defined adds Job trace item to meta-table on Job/Details page for a job with archive_features/has_trace == true and for jobs from a tree in enforce_for_trees, example: {title: 'Open in MyProfiling', url_template: 'https://my.profiling.service/{cluster}/{operationId}/{jobId}', enforce_for_trees: ['tree-with-traces'] } |

| query_tracker_default_aco | null, {stage1: string; stage2: string; } | Sets the default ACO in Query Tracker requests for each stage |

| operation_performance_url_template | null, {title: string; url_template: string} | Configuration for operation performance analysis system integration. Template must contain {operation_id} placeholder which will be replaced with actual operationId |

| tablet_errors_base_url | null, url as string | Base URL for tablet errors service to override uiSettings.tabletErrorsBaseUrl |

| ui_settings | null, Partial<UISettings \| SnakeCase<UISettings>> | Overrides for uiSettings |

Configuration

By default the application uses base configuration from path_to_dist/server/configs/common.js file. The behavior might be adjusted through APP_ENV and APP_INSTALLATION environment variables, see README.config.md for more details.

Migration

v1.17.0

• YT_AUTH_CLUSTER_ID environment variable has been replaced by ALLOW_PASSWORD_AUTH.
• config.ytAuthCluster option has been replaced by config.allowPasswordAuth.

How to run e2e on local machine

Here is an example of run & update of screenshot tests:

# Prerequsites: install docker and YTsaurus CLI
# https://www.docker.com/products/docker-desktop/
# https://ytsaurus.tech/docs/en/api/cli/install

# Terminal 1: launch local dev mode on 8080 port
LOCAL_DEV_PORT=8080 npm run dev:localmode:e2e

# Terminal 2: init local cluster
npm run e2e:localmode:init

# Terminal 2: mount repo in docker image and prepare everyting for tests run
# Make sure that you specify correct BASE_URL for e2e tests
# For linux use http://localhost:8080
# For macos use http://host.docker.internal:8080

print "Enter base url of your development stand: "; \
read BASE_URL; \
docker run --rm --network host -it -w /work \
    -v $(pwd):/work \
    -e BASE_URL=${BASE_URL} \
    "ghcr.io/gravity-ui/node-nginx:ubuntu20-nodejs18" \
    /bin/bash -c '
            cd tests
            npm ci
            npx playwright install --with-deps chromium
            cd ..
            npm run e2e:localmode:screenshots:update
'