Skip to main content

Database

What this category is for​

Use these nodes to read/write external data stores through managed credentials. They are workflow-level integrations and are separate from internal gateway storage.

Node list table​

Node typePurposeKey inputsBranch outputs
mongodb_nodeRun MongoDB operations (find, findOne, insert*, update*, delete*, aggregate).Config credentials_id, operation, collection, optional database, timeout_ms; operation-specific input fields (filter, document, pipeline, etc.)success, error
postgres_nodeRun PostgreSQL query or execute.Config credentials_id, operation, optional database, timeout_ms; input query, optional paramssuccess, error
sqlserver_nodeRun SQL Server query or execute.Config credentials_id, operation, optional database, timeout_ms; input query, optional paramssuccess, error
oracle_nodeRun Oracle query or execute.Config credentials_id, operation, optional database, timeout_ms; input query, optional paramssuccess, error
redis_nodeExecute Redis get, set, or del.Config credentials_id, operation, optional timeout_ms; input key, plus value/ttl_seconds/keys by operationsuccess, error

Common patterns​

1) Request lookup then proxy​

  • http_trigger -> postgres_node (fetch tenant config)
  • success -> set_node (build downstream headers) -> proxy_node
  • error -> response_node with service-unavailable contract

2) Idempotent write with Redis guard​

  • redis_node (get) to check dedupe key
  • If missing, proceed to DB write and redis_node (set with TTL)
  • If present, short-circuit with cached/duplicate response

3) Document enrichment in MongoDB​

  • http_trigger -> mongodb_node (findOne by key)
  • Merge input body with db document using set_node
  • Continue to downstream service call

4) Branch by DB outcome​

  • Route each db node error output to explicit failure handling
  • Route success to business logic, never assume implicit success

Common mistakes and how to debug​

  • Wrong credential type: each node validates credential type strictly and returns error branch (for example postgres node with non-postgres credential).
  • Missing required config: common misses are credentials_id, operation, and collection/query fields. Validate node config first.
  • Database not resolved: some nodes allow empty database only if credential has default database configured.
  • Parameter placeholder mismatch: PostgreSQL uses $1, SQL Server uses @p1, Oracle uses :1. Keep query syntax aligned with engine docs.
  • Timeouts too aggressive: nodes enforce minimum timeout floors; if operations still fail, inspect returned error and tune timeout_ms.