Troubleshoot Custom Components
Fix the most common issues with custom components.
Troubleshooting Upload Issues
These are the issues that occur when component is ingested into Haystack Enterprise Platform but before it's installed.
- If you're uploading through REST API or commands, you get an error response that will guide you through the steps to fix the issue.
- If you're uploading by releasing the forked repository, go to the Actions tab of your repository and check the actions that failed.
Troubleshooting Installation Issues
These are the issues after the component is ingested into Haystack Enterprise Platform and Haystack Platform attempts to install it.
Symptoms
Your component is not visible in the Pipeline Builder's component library.
Actions
- Refresh the Pipeline Builder page and wait a couple of minutes for the installation to complete.
- If your component is still not visible, print the installation logs of the latest version of your custom components to check the issues that occurred:
- On Linux and macOS, run:
hatch run dc:logs. - On Windows, run:
hatch run dc:logs-windows.
- On Linux and macOS, run:
Troubleshooting Operational Issues
These are the issues that may occur when a component was correctly uploaded and is visible in Pipeline Builder, but fails to work correctly in a pipeline.
| Symptoms | Action |
|---|---|
| Your component doesn't work in your pipeline. | Check the pipeline logs for issues and remedy actions:
|
You see this error message: Your organization's services are currently inactive due to user inactivity. Log in to <ProductName /> to reactivate your services and try again. It may take a few minutes for the services to become fully operational. | This means your organization's utility services were temporarily scaled down to save resources. Log in to Haystack Enterprise Platform and wait a while for the services to be activated. |
Troubleshooting Pip Dependency Installation Errors
If a custom component lists a pip dependency that Haystack Enterprise Platform can't install, a request to the pipeline fails with a 422 error. The error message names the dependency, for example Could not install pip dependency 'laya==0.4.0'. Without this error, you would only see a connection failure after several minutes of retries.
Symptoms
- A request to your pipeline returns a 422 error that names a pip dependency.
- Repeat requests to the same pipeline return the same 422 error within seconds.
- A request that starts a cold start through a streaming endpoint returns a 200 response with a stream that ends in a generic error. The 422 error appears on the next request.
Actions
- Check the dependency name and version in the error message. Make sure the package and version exist on PyPI and that the version is compatible with the Python version and the Haystack version your pipeline uses.
- Fix or remove the dependency in the
requirementsof your custom component, and upload the component again. - Wait about five minutes before you retry, or deploy the updated pipeline. Haystack Enterprise Platform remembers a failed dependency install for five minutes and answers repeat requests for the same pipeline configuration with the same 422 error without trying to install the dependency again. Other pipelines are not affected.
The 422 error appears only when the dependency can't be resolved, for example when the package or version doesn't exist or conflicts with another package. Temporary network or package index failures don't produce this error. Haystack Enterprise Platform retries those requests.
Was this page helpful?