Skip to main content
Everything on the Table Health Dashboard is backed by REST endpoints on the StarTree Cloud data-manager service. Use them to read health check results and to create, change, and test alert channels from a script, for example to apply the same alert setup to every cluster you run.
The dashboard and its alerts are turned on per environment. Slack channels need Slack delivery turned on as well. If an endpoint below returns 404, or Slack isn’t listed by GET /destinations/supported-types, contact StarTree support.

Endpoints

All paths are relative to https://<your-domain>/api/data/manager, where <your-domain> is your Data Portal host. Run the same calls against each cluster’s domain to configure several clusters. In the API, an alert channel is called a destination.

Authentication and Headers

All requests require a valid StarTree Cloud Bearer token.

Permissions

GET /destinations/access tells you which of these applies to the token you’re using.

Health check results

GET /api/clusterHealth

Returns the latest result of every table-level check shown on the dashboard, for the tables you can read. Checks run every 20 minutes by default (see GET /api/clusterHealth/schedule below).
Each check also carries its own diagnostic fields, such as details for KAFKA_CHECK or numServersNeedingReload and reasons for TABLE_SEGMENTS_RELOAD_CHECK. The health check reference describes them. Free-text error fields are redacted. The check keys shown by default are:

GET /api/clusterHealth/list

Returns every check the cluster runs, with its description, as {"<CHECK_KEY>": {"description": "..."}}. This includes cluster- and instance-level checks that the dashboard doesn’t show.

POST /api/clusterHealth/refresh

Runs the health checks now. The new results appear in GET /api/clusterHealth after a minute or two. Calls are limited to one every 30 seconds.

GET /api/clusterHealth/schedule


Alert channels (destinations)

A destination is one email address or one Slack incoming webhook, with its own coverage and schedule. Destinations are cluster-wide: one destination can cover tables in every workspace.

GET /supported-types

/api/v1/table-health-check-notifications/destinations/supported-types returns the destination types you can create on this cluster: ["EMAIL"], or ["EMAIL", "SLACK"] when Slack delivery is turned on.

GET /access

/api/v1/table-health-check-notifications/destinations/access returns {"canManageAllChannels": true} when the caller has ManageTableHealthNotifications on the cluster, and false otherwise.

GET /destinations

Lists destinations, ordered by id. Add ?type=EMAIL or ?type=SLACK to filter. Without ManageTableHealthNotifications, the list contains only your own email destination, if you have one.

Destination object

POST /destinations

Creates a destination and returns 201 with the destination object.

Request Body

Only one destination can exist per address or webhook; a second one returns 409.

Example: Slack channel for every table

Example: email for two tables

PUT /destinations/{id}

Changes a destination and returns the updated destination object. The body has the same fields as POST /destinations; any field you leave out keeps its current value.
  • To keep a Slack destination’s webhook, leave target out or send back the masked value from GET.
  • Sending subscriptions replaces all of the destination’s rows. Leave it out to keep them.
  • Changing target drops messages still queued for the old address or webhook.
To pause a channel without losing its settings:

DELETE /destinations/{id}

Removes the destination, its subscriptions, and any messages still queued for it. Returns 204 with no body.

GET /destinations/{id}/subscriptions

Returns the destination’s subscription rows. For a coversAll destination these are its overrides, often none.

POST /destinations/test

Sends a test message right away and returns once it’s delivered. Test a saved destination by ID, or test an address or webhook before saving it.
If Slack or the mail provider rejects the message (for example, a revoked webhook or an archived Slack channel), the call returns 400. A temporary delivery failure returns 503; retry it.

Subscriptions

When coversAll is false, a destination gets alerts only for what its subscription rows cover. When coversAll is true, rows act as exceptions to full coverage, such as opting one table out or muting one check. For each table and check, the most specific matching row wins: an exact table and check, then (table, *), then (*, check), then (*, *), then the destination’s coversAll. For example, this coversAll destination gets every table except staging_events_REALTIME:
A mute on an exact table and check sends the recovery message when the check passes, then clears itself. A mute on a wildcard row stays until you remove it. An email destination gets alerts only for tables its recipient can read. A Slack destination gets every table it covers, so anyone in that Slack channel can see those table names.

Reminder schedule

reminderCron uses six fields: second, minute, hour, day of month, month, day of week. It’s evaluated in reminderTimeZone and can fire at most once an hour.

Error Response

Errors return a JSON body with a human-readable message and the HTTP status. Some environments also return a dpeCode (for example DPE-5241) and remediation steps; see Error codes.

Example: set up the same alerts on several clusters

Each cluster keeps its own destinations, so run the script once per cluster you add. A rerun returns 409 for clusters that already have the webhook; use PUT /destinations/{id} to change those.

Table Health Dashboard & Alerting

The dashboard, each check’s fix, and how alerts behave.

Health checks

Every check, its diagnostic fields, and what to do about a failure.

Slack incoming webhook

Create the webhook URL a Slack destination needs.

RBAC Actions

The permission actions referenced on this page.