Skip to main content

​
Issues

When you try to use Unstructured, you get one of the following error types:
  • ClientResponseError: For example, an HTTP 402 Payment Required error.
  • SDKError: For example, an Insufficient API quota error.
  • PineconeApiException: For example, an HTTP 429 Too Many Requests or write unit limit reached error.
  • DestinationConnectionError: For example, an error that wraps an HTTP 429 Too Many Requests error.
  • HTTPError: For example, an HTTP 429 Too Many Requests error.
  • Error: For example, a RESOURCE_EXHAUSTED or a reduce request rate error.

​
Possible causes

  • Unstructured or the source’s or destination’s system imposes usage limits such as the number of API calls or data processed or stored over a specified time period or a specified rate of requests. These limits have been reached, or payment is required to continue.
  • A daily or monthly API quota for Unstructured, the source’s or destination’s API, or a specified model, has been reached.
  • An API rate limit of requests per second or minute has been reached on the source’s or destination’s system.
  • A subscription issue or required payment blocker has been encountered by Unstructured or for a related service.

​
Possible solutions

Some Unstructured Business account plans for dedicated instances or in-VPC deployments require the customer to set up or pay for related services, features, and models outside of Unstructured, in order to achieve volume discounts and to better manage overall usage costs. If you are on one of these account plans, check the following:
  • Check your usage and quotas: Review your current usage against your budget limits in your Unstructured account dashboard and any relevant third-party services—such as Pinecone, Azure OpenAI, or Databricks.
  • Verify your billing and subscription details: Ensure any billing agreements and subscriptions are active and their payments are up-to-date.
Other possible solutions include the following:
  • Reduce the rate of requests if you can.
  • Run processing jobs less frequently if possible.
  • Process data in smaller batches if you can.
  • If configurable, reduce the concurrency settings in your Unstructured ingestion pipelines.
  • Try processing jobs again later. Some rate limits are often time-based; waiting and retrying later might work.
  • Upgrade your plan. If you are consistently hitting limits, consider upgrading your plan for Unstructured or the third-party service.
  • Contact Unstructured Support. If you are consistently hitting limits, consider contacting Unstructured Support to discuss your options.

​
Additional resources

To ask questions or get additional help with this issue, see requesting support.