Client configuration

from gwenlake import Gwenlake

client = Gwenlake(
    api_key=None,          # see Authentication
    credentials=None,
    profile=None,
    base_url=None,         # default https://api.gwenlake.com/v1
    timeout=60,            # seconds; connect timeout is 5 s
    max_retries=2,
)

Base URL

Short name	URL	Serves
`api` (default)	`https://api.gwenlake.com/v1`	the Gwenlake APIs and the Atlas catalog
`atlas`	`https://atlas.gwenlake.com/api/v1`	the same API, as the Atlas application serves it
a tenant	`https://atlas.acme.example/v1` or `/api/v1`	a platform deployed for one client

The constants are gwenlake.constants.GWENLAKE_API_URL, GWENLAKE_ATLAS_URL and KNOWN_ENDPOINTS; base_url="atlas" is accepted as a short name. Every resource is served under one gateway, https://api.gwenlake.com/v1. A tenant deployed on its own host, a self-hosted platform or a staging environment is reached by naming its endpoint, in order of precedence: base_url= on the client, the GWENLAKE_BASE_URL environment variable, or the base_url key of the credentials profile in use (see Authentication). An API host serves the API under /v1, an Atlas application host under /api/v1 (https://atlas.gwenlake.com/api/v1); a bare host is completed to https://<host>/v1, a host with /api to /api/v1, any other path (http://localhost:8000/v1) is used as given. The command line’s config set-profile probes a host for the right prefix.

Timeouts and retries

The default timeout is one minute per request. Failed requests are retried twice with an exponential backoff, starting at half a second and capped at eight. Pass timeout= on the client for every call, or on a SQL statement for that call only:

client.statements.create(statement="SELECT ...", timeout=600)

Environment variables

Variable	Meaning
`GWENLAKE_API_KEY`	the API key, when no key, profile or credentials object is passed
`GWENLAKE_BASE_URL`	the gateway URL
`GWENLAKE_PROFILE`	the credentials profile the command line uses

A .env file in the working directory is loaded by the training module and by the command line.

Resources

Attribute	Covers
`client.projects`	projects
`client.filesystem`	folders, search, move / rename / restore of any resource
`client.datasets`	datasets: create, update, summary, branches, `delete_rows`, maintenance, vector search
`client.files`	files inside a dataset
`client.statements`	ad hoc SQL: `create`, `read_arrow`, `iter_arrow`, `read_all`, `to_dataframe`, `to_file`
`client.queries`	stored SQL queries and their versions
`client.transforms`	the producers of the lineage: `set_parameters`, `execute`, `functions`
`client.builds`	builds (`run`, `plan`, `wait`, `run_and_wait`, `logs`, `cancel`) and asynchronous tasks
`client.schedules`	schedules: `time_trigger`, `build_action`, `trigger`, `enable`
`client.lineage`	the lineage graph, data branches of a lineage cone, workflow positions
`client.repositories`	repositories, `scan_lineage`, the git routes (tree, file, commit, branches, import)
`client.connections`	connections, their files, `query`, `test`, credentials
`client.ml_models`	the catalog’s ML models and their artifacts
`client.mediasets`, `client.syncs`, `client.flows`	mediasets, syncs, flows
`client.agents`, `client.skills`, `client.tools`, `client.ontologies`, `client.applications`	agents and what they use
`client.organizations`, `client.users`, `client.api_keys`	the organization, you, the API keys
`client.models`	inference models
`client.chat`	chat completions
`client.embeddings`	embeddings

client.resolve("<project>.<dataset>") turns a handle into a dataset id. Every resource has an async twin on AsyncGwenlake.

Get started

Concepts

Workspace

Data

Code

Build

Models

Inference

Agents

Client

Changelog

Base URL

Timeouts and retries

Environment variables

Resources

​Base URL

​Timeouts and retries

​Environment variables

​Resources

Base URL

Timeouts and retries

Environment variables

Resources