Admin

Processing Queue Management

Monitor and manage document processing tasks across all Libraries

Overview

The Processing Queue dashboard provides centralized monitoring and management of all document extraction and embedding tasks across your entire George AI instance.

Access: Admin Panel → Processing Queue

Queue Status Overview

Pending Tasks
127
Waiting for processing
Processing Tasks
4
Currently being processed
Failed Tasks
12
Need attention

Worker Status

Running

Queue worker is actively processing tasks

Last processed: 5 seconds ago

Management Actions

Worker Control

Start or stop the queue worker that processes tasks

Start Worker

Resume processing pending tasks. Use when queue was stopped for maintenance.

Stop Worker

Pause new task processing. Tasks already running will continue to completion.

Stopping the worker does NOT cancel already-running tasks

Failed Task Management

Handle tasks that encountered errors

Retry Failed

Re-queue all failed tasks for another attempt. Useful after fixing configuration issues.

Clear Failed

Permanently remove failed tasks from the queue. Use for tasks that cannot be fixed.

Actions can be filtered by Library ID if specified

Pending Task Management

Manage tasks waiting to be processed

Clear Pending

Remove all pending tasks from the queue without processing them.

Use with caution: cleared tasks will NOT be processed

Task Cancellation

Cancel specific processing tasks

Cancel Tasks

Cancel processing tasks for specific files or Libraries. Applies to both pending and actively processing tasks.

Useful when you need to stop processing for specific files that are problematic or no longer needed.

Viewing Task Details

Task Information

Each processing task shows detailed information:

Field Description
File Name The document being processed
Library Which Library this file belongs to
Status Current processing state (pending, extracting, embedding, completed, failed)
Created At When the task was created
Started At When processing began
Finished At When processing completed (or failed)
Processing Time Total time spent processing (in milliseconds)
Error Message Error details if task failed

Filter by Status

Use the status filter dropdown to view only tasks in a specific state (pending, processing, failed, completed, etc.)

Common Management Scenarios

Queue Backed Up with Many Pending Tasks

Symptoms: Thousands of pending tasks, very slow processing

Causes:

  • Too many files added at once (large crawler run)
  • AI Services overwhelmed or slow
  • Not enough AI Service servers

Solutions:

  • Add more AI Service servers for parallel processing
  • Reduce crawler maxPages settings
  • Process in batches: stop worker, clear some pending, restart
  • Check AI Services health and restart if needed
Many Failed Tasks After Configuration Change

Symptoms: Spike in failed tasks after changing Library settings

Common Causes:

  • Changed embedding model to one that's not available
  • Changed OCR timeout to value that's too low
  • AI Services temporarily unavailable

Solutions:

  • Revert Library configuration changes
  • Verify AI Services are running and responsive
  • Use "Retry Failed" to reprocess with corrected settings
Emergency Maintenance Required

Scenario: Need to perform urgent system maintenance

Steps:

  1. Stop the queue worker
  2. Wait for currently processing tasks to complete (monitor "Processing Tasks" count)
  3. Perform maintenance
  4. Restart queue worker
Pending tasks remain in queue and will resume when worker restarts
Specific Files Causing Problems

Scenario: Certain files repeatedly fail or timeout

Steps:

  1. Identify problematic file(s) from failed task list
  2. Use "Cancel Tasks" to stop processing those specific files
  3. Review files - are they corrupted, too large, unsupported format?
  4. Either fix files and reprocess, or exclude them permanently

For permanently problematic files, consider adding crawler exclude patterns to prevent re-ingestion.

Best Practices

Monitor Regularly

Check the Processing Queue dashboard daily to catch issues early. Set up alerts if failed task count exceeds a threshold.

Investigate Failed Tasks

Don't just clear failed tasks—review error messages to identify root causes and prevent future failures.

Be Cautious with Bulk Actions

Actions like "Clear Pending" or "Clear Failed" affect many tasks at once. Consider using Library filtering to limit scope.

Scale AI Services as Needed

If pending task count consistently grows, you need more processing capacity. Add AI Service servers or increase concurrent limits.

George-Cloud