Skip to content

End-User Troubleshooting Guide

This guide helps you resolve common error responses and issues encountered when interacting with PowerMind InsightML.

1. Common Errors and Solutions

🔑 Authentication & Authorization

Error Message / Code Probable Cause Corrective Action
401 Missing authentication Request headers lack the authorization token or API key. Include Authorization: Bearer <jwt_token> (for human sessions) or x-api-key: IML_live_... (for programmatic tasks).
401 Invalid API key format The provided API key is missing the mandatory IML_ signature prefix. Ensure you copied the full generated key. Check for truncation in your environment variables.
401 Invalid email/username or password The provided credentials do not match any active user account. Verify spelling and retry. After 5 failed attempts, the account will be locked.
403 Account is temporarily locked... The account has been locked for 15 minutes due to 5 consecutive failed login attempts. Wait for the 15-minute lockout timer to expire, then try logging in again.
403 This API key is restricted A Deployment API key was used on a platform management endpoint. Platform endpoints (e.g. creating projects or datasets) require a Platform API Key or JWT. Switch your credentials.
403 API key is not authorized for this deployment A Deployment API key was sent to a model inference endpoint it is not mapped to. Regenerate a Deployment Key under the target deployment's details panel, or verify the deployment ID mapping.
400 Password must be at least 12 characters long. The registration password is too short. Choose a stronger password containing at least 12 characters.
409 Email/Username is already registered/taken The email or username specified during registration is already in use. Register with a different email/username, or log in to your existing account.

📂 Dataset Uploads & Ingestion

Error Message / Code Probable Cause Corrective Action
400 Invalid CSV format The file is corrupted, not a standard CSV, or lacks headers. Open the file in a text editor to verify the first row contains comma-separated column names. Ensure UTF-8 encoding.
400 Target column not found The target column name specified does not match any header in the dataset. Verify spelling and casing in the target column parameter. Remember that CSV headers are case-sensitive.
422 Unprocessable Entity Column data types are incompatible (e.g., passing text values into a numerical target column during regression). Inspect the column fields. Ensure the target column contains numerical inputs for regression/time-series.

âš¡ AutoML Training Runs

Error Message / Code Probable Cause Corrective Action
400 Dataset too small The dataset contains fewer than 20 rows after removing empty records. Upload a dataset with at least 20 valid rows to satisfy cross-validation requirements.
400 Target must contain at least 2 classes You requested classification, but the target column contains only one unique value. Ensure the training dataset includes samples from multiple classes (e.g., both fraud=0 and fraud=1 cases).
400 Clustering should not have target_column A target column was specified for an unsupervised clustering experiment. Omit the target_column parameter when running clustering experiments.
400 target_column required for supervised tasks The target column was omitted for classification or regression. Pass the name of the prediction target column in the request body.

🚀 Deployments & Inference

Error Message / Code Probable Cause Corrective Action
429 Max active deployments exceeded The workspace has exceeded its concurrent active deployment capacity. Stop or delete an inactive/old deployment in the workspace before starting a new one.
409 Deployment status is 'deploying' The deployment has not finished loading into the serving runtime. Wait for the deployment status to transition to running before sending predictions. Use the SDK wait_for_deployment helper.
502 Serving runtime failed The model could not be loaded (e.g., due to tensor size conflicts or corrupted model artifacts). Check the deployment error_message in the console or API response. Re-queue the deployment if necessary.
400 Feature count mismatch The prediction payload input shape does not match the model's expected feature count. Query the deployment details to inspect the input_feature_count and input_features array. Align your request inputs.

🔄 Time-Series Lag Computation

Error Message / Code Probable Cause Corrective Action
400 Insufficient historical data The list of historical records contains fewer rows than the model requires (e.g., fewer than 7 rows). Supply more historical observations in the array. Provide at least 30 rows for optimal rolling window metrics.
400 Target column not found in historical data The target column argument does not exist in your JSON array of observations. Verify that the target column keys in your list match the name of the target column parameter exactly.
400 Time column not found The time column argument does not exist in your JSON array of observations. Verify the spelling of the timestamp key in your historical observations array.

2. Diagnostic Steps

For Web Console Issues

  1. Force Refresh: Press Ctrl + F5 (or Cmd + Shift + R on Mac) to reload the static assets.
  2. Clear Selection Storage: If the console gets stuck loading a deleted workspace, clear your browser cookies/local storage and log in again.
  3. Inspect Activity Log: Watch the activity log at the bottom of the overview dashboard to view error notifications.

For API and Integration Issues

  1. Swagger UI: View the dynamic Swagger UI at /docs when the API is running to inspect endpoint schemas.
  2. Verify Endpoint Routes: Check the REST API Reference to ensure correct endpoint URLs.
  3. Inspect Response Bodies: Look inside the detail object of 4xx responses for specific validation error descriptions.