VectorAIError as the application-level catch-all.
Python SDK exceptions
The current Python SDK exposes these exception classes for common application handling.Catch specific exceptions
Use the namespaced SDK APIs in examples and production code. Collections live underclient.collections, and vector search lives under client.points.
Create a collection idempotently
CatchCollectionExistsError to skip creation when a collection is already present.
Retry transient failures
Use exponential backoff for retryable errors. The helper functions understand SDK exception types, includingConnectionError, RateLimitError, and TimeoutError.
Inspect error details
Every SDK exception exposesmessage, code, details, and operation where available. Use these fields for logging and custom handling.
Common error messages
Collection not found
Collection not found
Cause: The collection name is misspelled, or the collection was deleted.Fix: Call
client.collections.list() to verify the collection exists before operating on it.Wrong vector dimensions
Wrong vector dimensions
Cause: The query vector or inserted vector has a different number of dimensions than the collection configuration.Fix: Confirm the embedding model and collection dimensions match.
Authentication failure
Authentication failure
Cause: The server has authentication enabled but the client did not supply a valid access token.Fix: Pass the configured token to
VectorAIClient or set ACTIAN_VECTORAI_ACCESS_TOKEN.Next steps
Troubleshooting
Diagnose connection, search, and startup issues.
Python SDK
Get started with the Python SDK.