Open it from Monitoring → Health Dashboard in the left navigation. The dashboard is turned on per environment; if you don’t see it, contact StarTree support. See Permissions for what each action needs.
Reading the dashboard
The top of the page shows how many tables are failing and which checks they fail, with a View failing shortcut. Last run shows when the health checks last ran. They run every 20 minutes by default, and Refresh runs them now (it takes a minute or two). The dashboard has three tabs:- Overview lists each check with its passing and failing table counts. Fix N tables opens the failing tables for that check.
- Tables lists every table you can read, with its workspace and health. Search, or filter by status or by check. Hover a failing table and click Fix to see and fix its failing checks in one place, or click the row to open the table’s details.
- Notifications is where you set up alerts. See Alerts.
Checks on the dashboard
The dashboard shows four checks by default. Each one has a fix you can run from the Data Portal.
Tables that StarTree runs for you, such as
system_query_log, don’t appear on the dashboard and never trigger alerts.
The other table-level checks still run and show on Cluster Health. The health check reference explains every check and its diagnostic fields. To surface more checks on this dashboard, contact StarTree support.
Table details
A table’s details page lists each check with a Passed or Failed badge. Failing checks open to show:- Findings: a one-line verdict and the numbers behind it, such as stale segments per server, topics with too-short retention, or the flagged columns. More info opens the full detail where there is one.
- Remediation: the fixes for that check, described below.
Remediating failing checks
Fix buttons appear on failing checks and are enabled when you have permission to change the table. After a fix, the check updates on the next health run; click Refresh to run it sooner. Segment Assignment- Rebalance segments moves segments to where the assignment expects them. Rebalance status shows progress, and Cancel rebalance stops one that’s running.
- Reload segments first shows what a reload would change, then applies the current schema and index settings to every segment. Queries keep running during the reload. Reload status shows progress.
- Lower flush threshold seals segments sooner, so Kafka still holds the messages if a server has to catch up. Each affected topic gets a pre-filled value that keeps twice the threshold under its retention, and you choose which topics to change. New segments seal at the new interval; nothing is reloaded.
- Raise topic retention keeps the table as it is and gives you
kafka-configs.shcommands to run on your Kafka cluster, with a suggested retention of three times the flush threshold. StarTree doesn’t change your Kafka cluster.
- Remove dictionary encoding lists the flagged columns, all selected. Columns that still need their dictionary are shown but locked, with the reason: an inverted, FST, timestamp or (on a non-numeric column) range index, the sorted column, an on-heap dictionary, a star-tree dimension, or a rollup table. For some index locks, Edit indexes opens the table’s indexes so you can change them first.
- Applying reloads every segment of the table. On a large table that takes a while and adds load on the servers, and queries may be slower until it finishes.
Alerts
Alerts go to channels. A channel is one email address or one Slack incoming webhook, and each has its own coverage and schedule. Manage them on the dashboard’s Notifications tab.Get alerts on your own email
On the Notifications tab, click Set up my alerts. You can also click Subscribe on a check from a table’s details page to get alerts for that check on that table.Add a channel
Adding channels for other people, and any Slack channel, needs the notification admin permission (see Permissions).1
Open Notifications
On the Health Dashboard, open the Notifications tab and click Add channel.
2
Choose where alerts go
Under Send to, pick Email and enter one or more addresses (each becomes its own channel with the same settings), or pick Slack and paste an incoming webhook URL that starts with
https://hooks.slack.com/services/ (see How to create a Slack incoming webhook).3
Choose what it covers
Under Covers, pick Every table and check, or Choose tables and checks and select them in the grid.
4
Choose when it sends
Under Frequency, turn on one or more of the options described below.
5
Test and save
Click Test to send a test message, then Save.
The Slack option appears only when Slack delivery is turned on for your environment. If you don’t see it, contact StarTree support.
Create a Slack incoming webhook
A Slack channel needs an incoming webhook for the Slack channel you want alerts in. Follow How to create a Slack incoming webhook, then paste the URL under Send to when you add the channel. Add one StarTree channel per webhook.What a channel covers
- Every table and check includes tables created later.
- Choose tables and checks covers only what you select. Tables created later aren’t included until you add them.
When a channel sends
At least one option must be on. Without reminders, a failing check is announced once and not again while it stays failing. A new channel also won’t hear about checks that were already failing before it was added until something changes, so turn on reminders if you want a standing list.
A check that keeps flipping between failing and passing doesn’t flood the channel. After an alert, further changes to that check are held for 30 minutes, then announced once if its state is still different.
Mute, pause, and remove
- Mute stops alerts for a failing check until it next passes, then alerts resume on their own. Mute from a table’s details page or from the channel’s grid.
- The On switch pauses or resumes a whole channel.
- Test on a channel’s row sends a test message. The trash icon removes the channel.
Permissions
Related
- Health checks: the Cluster Health view and every check, with what it verifies, its threshold, and what to do about a failure.
- RBAC Actions: the permission actions referenced above.
- Troubleshooting: what to do once a check is failing, for query failures, ingestion problems, and resource pressure.
- How to create a Slack incoming webhook: the Slack-side steps for a Slack channel.
- Table health API reference: read health results and create, change, and test alert channels from a script, for example across several clusters.

