Clean Up Legacy FastGPT Workflow Enum Data

Legacy Workflow Enum Data Issue

Some historical FastGPT workflow nodes store TypeScript enum expression strings directly in MongoDB, rather than the intended raw string values. For example, a corrupted stored entry might look like:

{
  "renderTypeList": ["FlowNodeInputTypeEnum.hidden"],
  "valueType": "WorkflowIOValueTypeEnum.any"
}

The correct, validated format uses plain string values:

{
  "renderTypeList": ["hidden"],
  "valueType": "any"
}

This corrupted data disrupts workflow node input rendering and IO type validation. This cleanup script resolves these issues after completing the V1→V2 workflow migration, targeting data in the apps.modules and app_versions.nodes collections.

Execute the Cleanup Script

The cleanup endpoint operates in dry-run mode by default, which validates data in memory using the PublishAppBodySchema without writing changes to MongoDB. Use this first curl command to run a dry scan:

curl -X POST 'https://your-domain/api/admin/dataClean/initWorkflowData' \
  -H 'Content-Type: application/json' \
  -H 'rootkey: YOUR_ROOT_KEY' \
  -d '{"dryRun":true,"batchSize":1000,"writeBatchSize":10}'

After reviewing the returned statistics, switch to non-dry-run mode to apply fixes:

curl -X POST 'https://your-domain/api/admin/dataClean/initWorkflowData' \
  -H 'Content-Type: application/json' \
  -H 'rootkey: YOUR_ROOT_KEY' \
  -d '{"dryRun":false,"batchSize":1000,"writeBatchSize":10}'

The following request parameters control the cleanup process:

Cleanup Operational Rules

The script follows five core operational rules to minimize system impact:

1. Scans workflow data in apps and app_versions collections in batches to reduce read and write pressure on the database.
2. Formats each workflow document to resolve historical dirty fields, null values, enum expressions, and legacy structure compatibility issues.
3. Validates the formatted save payload fields nodes, edges, and chatConfig against the PublishAppBodySchema schema.
4. Documents that fail Zod validation are recorded in the response but not written to MongoDB.
5. In non-dry-run mode, only documents that changed during formatting and passed Zod validation are written to the database; unchanged documents are not overwritten.

Response Statistics

The API response includes segmented statistics for apps, appVersions, and a total summary. Metrics include the number of scanned documents, fixable documents, Zod validation error count, successful and failed write counts, enum expression occurrence statistics, change samples, and error samples.

Source: FastGPT official source

Applicability and version scope

Use this page for the documented Deployment and upgrades scenario. Confirm the FastGPT, dependency, API, and deployment versions in the official source before applying a change.

Safety guardrails

Use [REDACTED_CREDENTIAL] for credentials and private data. Confirm the documented environment and version before review.

Rollback guidance

Restore the prior technical-content authority snapshot. Restore saved configuration and data snapshots, then repeat the smallest verification scenario.