OriginChainDB / Guides

Current console and API guides

Find the workflow, connection details and availability limits for the latest documented OriginChainDB features.

These guides describe the released behavior reviewed against source and release evidence on 22 October 2026. Check each guide’s availability note: features enabled on dedicated deployments may not yet be available on the Free configuration.

The 19 October release adds migration live sync, embedding schedules, relationship backfills and expanded access controls. Check your instance’s /v1/capabilities before relying on a feature: a published change or newer source commit does not prove that every instance has been upgraded. See release notes and the CLI preview guide.

Choose your next task

• Use the current console

  Find your instance, connect a client, explore records and understand the controls available to your role.

• Copy and sync data with Migrate

  Copy a source database, validate mappings and rows, and use supported PostgreSQL or MySQL live sync on dedicated deployments.

• Saved embedders and embedding runs

  Generate vectors from selected table columns with a saved embedding endpoint, and control which rows a run processes.

• PostgreSQL compatibility

  Connect through the PostgreSQL endpoint, understand transaction behavior and handle the differences that matter to an application.

• Elasticsearch API: current support

  Connect a 7.x client with the supported authentication formats, and check the operations and limits available on dedicated deployments.

• Vector record API

  Create vector indexes, write dense or sparse records with metadata, fetch by ID and search within a namespace.

Search suggestions, timeouts and cancellation

Search suggestions apply the caller’s row policies. If a suggestion request cannot complete safely, show the returned error instead of presenting an empty suggestion list as a successful result.

Native SQL, graph, full-text and vector-table reads support the request headers X-OC-Statement-Timeout-Ms and X-OC-Query-Id. Use a valid timeout and a query ID unique among running requests. Malformed header values are refused, a duplicate active query ID conflicts, and a timed-out request returns a timeout response. Handle cancellation, throttling and failures explicitly.

The vector record search route has its own request contract and currently refuses a timeout_ms body field. Do not transfer options between APIs without checking their reference.

Learn a workflow and resolve questions

Browse the video tutorial library or the console FAQ. If a recording predates a new option, use the current guide for its availability and limits. The demo workspace offers browser-only sample data; sign in to the real console to use your own instances.