Sigma Setup

​

Prerequisites

​

Where each signal comes from

​

Audit log export for usage metrics

1. Create a Google Cloud Storage bucket and a folder path for the logs.
2. Create a Google Cloud service account for Prizm with read access on the bucket, and download a JSON key for it.
3. In Sigma, open Administration → Account and enter the storage location (gcs://<bucket>/<path>/) and the service account email, then grant Sigma’s service account write access on the bucket.
4. Schedule the audit log export to the bucket hourly or daily. Hourly gives fresher numbers.

​

Connector Setup

1

Open the connector wizard

Navigate to Source and click the + icon next to the source list. Select Sigma from the connector grid.

Field	Description
Connection Name	A label for this instance (e.g., sigma-prod)
Description	Optional notes about what this connection covers
Host URL	Sigma API base URL. Prefilled with https://aws-api.sigmacomputing.com; change it if your Sigma organization uses a different API host

2

Authenticate

• Client ID + Client Secret (Recommended)
• Use Vault

Field	Description
Client ID	Sigma Client ID for the service account
Client Secret	Sigma Client Secret; stored encrypted and never shown after save

Toggle Use Vault to retrieve the Client ID and Client Secret from a vault your organization has already configured in Prizm, instead of entering them in the form.

Click Next to validate the credentials. Invalid credentials return a structured error identifying the specific failure (invalid client, wrong host, insufficient scope).

3

Configure workspace scope

Use the Workspaces setting to restrict onboarding to specific workspaces: a comma-separated list of workspace names, or leave it blank to include all workspaces. Assets in workspaces not listed are not catalogued.

4

Enable usage metrics (optional)

To collect usage metrics, turn on Enable audit logs usage and complete:

Field	Description
Storage provider	Google Cloud Storage
Audit logs GCS URI	gcs://<bucket>/<path>/ (gs:// is also accepted)
GCS service account key	The JSON key for the Prizm service account with read access on the bucket

Two optional toggles, Propagate score based on associated tables and Propagate alerts based on associated tables, pull the warehouse table score and alerts into Sigma assets. Prizm validates the connection, including the GCS location, when you save or edit the source.

5

Job schedules

Job Type	Frequency	Objects Covered
Catalog / Metadata	Every 24 hours	Workspaces, Workbooks, Pages, Data Elements, Data Models and Datasets, Tags
Operational / Refresh	Every 1 hour	Workbook and data model update times
Usage / Adoption	Every 24 hours	Usage logs, view counts, unique users

Usage, Reliability, lineage, and criticality depend on Sigma’s audit log / query log access, which requires a Premier-tier Sigma account with logging enabled. Audit logs are not enabled by default and are not available on all Sigma plans. Without them these signals don’t populate.

6

Save and verify

Click Save. Prizm runs an initial catalog crawl — assets appear in the catalog within a few minutes. Navigate to Settings → Connectors → Logs to monitor scan progress.

​

Verify the Setup

• The source saves without a validation error
• The first run completes in this order: Technical, Asset lineage, Performance, Operational
• Workbooks and data models appear under the source, only from the selected workspaces
• A data model shows its workbook mapping and upstream tables
• After a materialization runs, freshness and job duration appear
• After the first audit log export lands in GCS, usage metrics appear

​

A job stays in Running

1. Open the job run and check its events. Each agent call records a submitted event, then a response received or failed event. Submitted with no response means the reply never came back. Failed means the agent returned an error, shown in the event.
2. Search the scheduler logs for [NATS]. Published <tool> … job_id=X without Received result for job_id=X means the request wasn’t answered. Transport error … reconnect n/3 or Reconnect failed means the messaging service is unreachable; it fails after 3 attempts rather than hanging.
3. Confirm the Sigma agent is running, its logs show the same job_id, and it uses the same environment name (NATS_ENVIRONMENT, or ENVIRONMENT if unset) as the scheduler. A mismatch means the agent never sees requests and nothing errors. The agent and scheduler must be deployed together, because an older agent rejects newer arguments.
4. If the agent received the request, look for a Sigma API error or slow call in its logs; large workspaces take longer. Otherwise cancel the job, fix the cause, and run it again.

​

Next Steps