Handling Errors

When an exception occurs in your handler function, the RunPod SDK automatically captures it, marking the job status as FAILED and returning the exception details in the job results.

Implementing custom error responses

In certain scenarios, you might want to explicitly fail a job and provide a custom error message. For instance, if a job requires a specific input key, such as seed, you should validate this input and return a customized error message if the key is missing. Here's how you can implement this:

import runpod


def handler(job):
    job_input = job["input"]

    # Validate the presence of the 'seed' key in the input
    if not job_input.get("seed", False):
        return {
            "error": "Input is missing the 'seed' key. Please include a seed and retry your request."
        }

    # Proceed if the input is valid
    return "Input validation successful."


# Start the RunPod serverless function
runpod.serverless.start({"handler": handler})

note

Be cautious with try/except blocks in your handler function. Avoid suppressing errors unintentionally. You should either return the error for a graceful failure or raise it to flag the job as FAILED.

One design pattern to consider, is to Refresh your Worker when an error occurs.