Skip to main content

Documentation Index

Fetch the complete documentation index at: https://datost.com/docs/llms.txt

Use this file to discover all available pages before exploring further.

Datost connects to ClickHouse so your team can ask event, product, and OLAP questions in Slack and get answers backed by real columnar data. It’s a good fit for high-volume event streams, product analytics tables, and ClickHouse Cloud clusters.

​
Prerequisites

Before connecting, gather the following from your ClickHouse cluster:
  • Host — e.g. your-cluster.clickhouse.cloud or an IP address
  • HTTP port — 8123 for HTTP, 8443 for HTTPS (ClickHouse Cloud defaults to HTTPS on 8443)
  • Protocol — HTTP or HTTPS
  • Username — defaults to default
  • Password
  • Database (optional) — leave empty to expose all non-system databases
Datost uses the official @clickhouse/client package over the HTTP interface. The native TCP protocol is not supported.

​
Connect in the admin panel

1

Open Data Sources

In the Datost web app, go to Settings → Data Sources and click Add data source.
2

Choose ClickHouse

Select ClickHouse from the provider grid.
3

Paste a connection URL or fill fields manually

Paste a ClickHouse Cloud curl command or connection URL to auto-fill host, port, protocol, username, and password — or enter them manually. Supported shapes include clickhouse://user:pass@host:port/db, bare https://host.clickhouse.cloud:8443, and the curl --user 'user:pass' https://host:port snippet from the ClickHouse Cloud connect dialog.
4

Whitelist the Datost egress IP

If your cluster restricts traffic, add the egress IP shown in the modal to your ClickHouse IP allowlist.
5

Test and save

Click Test connection. Datost runs SELECT version() to confirm reachability and returns the server version. Save to finish.

​
Permissions required

The connecting user needs:
  • SELECT on the databases and tables you want Datost to query.
  • SELECT on system.tables and system.columns for schema introspection (used for table listing and column metadata).
Read-only users work well. Datost does not issue writes during normal operation.
If you provide a specific database, Datost scopes schema discovery to that database. Leaving it blank exposes every database except system, INFORMATION_SCHEMA, and information_schema.

​
Limitations

Large result sets are capped. Queries run with max_result_rows (default 1000) and result_overflow_mode = 'break', so ClickHouse stops streaming once the cap is hit and the response is marked as truncated.
  • HTTP only. The native TCP protocol isn’t supported — use the HTTP/HTTPS interface (ports 8123/8443).
  • Client caching. Datost caches up to 50 ClickHouse clients per process with a 30-minute idle TTL; connections refresh automatically when credentials change.
  • Statement termination. Trailing semicolons are stripped before the query is sent.
  • Views vs. tables. Engines containing View are surfaced as views; everything else is treated as a table.

​
Troubleshooting

  • Connection test fails with a network error — verify the port matches your protocol (8123 for HTTP, 8443 for HTTPS) and that the Datost egress IP is whitelisted.
  • Authentication error — confirm the username (ClickHouse Cloud defaults to default) and that the password has no leading/trailing whitespace from the paste.
  • Missing tables — check that the user has SELECT on system.tables and system.columns, and that you didn’t scope the connection to a database that doesn’t contain them.