Queue Management

Monitor and manage the job queue system for event processing and platform delivery. View pending jobs, retry failed deliveries, and optimize queue performance.

Advanced Feature
Super Admin Only

Super Admin Access Required

Queue Management features are only available to Super Admin users. Regular users and Client Users will not see this section.

Understanding the Queue System

Tandem Beam uses a database-driven job queue for asynchronous event processing. When events are tracked, they are queued for platform delivery, allowing fast response times while ensuring reliable delivery.

How Queuing Works

  1. 1. Event captured by SkyBeam or ClickBeam
  2. 2. Event stored in Tandem Beam database
  3. 3. Queue job created for each platform destination
  4. 4. Queue workers process jobs asynchronously
  5. 5. Event delivered to platform APIs
  6. 6. Job marked complete or failed

Benefits of Queuing

  • Fast response times (event tracking doesn't wait for platform APIs)
  • Automatic retry for failed deliveries
  • Better handling of platform API rate limits
  • Isolation from platform downtime

Accessing Queue Management

To access the Queue Management dashboard:

  1. Click your name in the top-right corner
  2. Hover over "Admin" in the dropdown menu
  3. Click "Queue Management"
URL: /queue-management

Queue Metrics

The Queue Management dashboard displays real-time metrics:

Pending Jobs

Number of jobs waiting to be processed. High counts may indicate queue backup.

Failed Jobs

Jobs that failed after all retry attempts. Can be retried manually or cleared.

Processing Rate

Jobs processed per minute. Indicates queue worker performance.

Tenant Filtering

Use the tenant dropdown to filter queue view by specific advertiser or view all queues system-wide:

Filter Options

  • All Tenants: View system-wide queue metrics across all advertisers
  • Specific Advertiser: Focus on queue jobs for a single advertiser

Filtering affects all metrics and operations (clear failed jobs, retry, flush, etc.)

Queue Operations

Retry Failed Jobs

Retry all failed jobs in the queue. Useful when platform API issues are resolved.

Use when: Platform APIs were temporarily down and now operational

Effect: Moves failed jobs back to pending queue for reprocessing

Clear Failed Jobs

Remove failed jobs from the queue without retrying. Use for permanently failed jobs.

Use when: Failed jobs cannot be retried (invalid data, expired platform access)

Effect: Permanently removes jobs from failed_jobs table

Pause Queue Processing

Temporarily stop processing queue jobs. Jobs continue to queue but won't be processed.

Use when: Performing maintenance or investigating platform API issues

Effect: Queue workers stop processing, pending jobs accumulate

Resume Queue Processing

Restart queue processing after pausing. Accumulated jobs will be processed.

Use when: Maintenance complete or issues resolved

Effect: Queue workers resume processing pending jobs

Flush Queue

Dangerous

Delete all pending jobs from queue. Use with extreme caution.

Warning: This permanently deletes queued events - they will never be delivered

Use when: Queue backup is unrecoverable or contains test data

Monitoring Queue Health

Healthy Queue Indicators

Normal Operation

  • ✓ Pending jobs: < 100
  • ✓ Failed jobs: < 10
  • ✓ Processing rate: 10-50 jobs/min
  • ✓ No jobs older than 5 minutes

Warning Signs

  • ⚠ Pending jobs: 100-1000
  • ⚠ Failed jobs: 10-50
  • ⚠ Processing rate: < 10 jobs/min
  • ⚠ Jobs stuck for 10+ minutes

Common Issues

Queue Backup (1000+ Pending Jobs)

Symptoms: Large number of pending jobs, slow processing rate

Solutions: Check queue workers are running, identify specific platform causing backup, temporarily pause problematic platform processing, contact support for worker scaling

High Failed Job Count

Symptoms: Many jobs in failed_jobs table, repeated failures for same platform

Solutions: Check platform credentials, review error messages in failed jobs, verify platform API status, fix configuration issues before retrying

Best Practices

Monitor Regularly: Check Queue Management daily during initial deployment, then weekly once stable.
Investigate Failed Jobs: Don't immediately clear failed jobs - review error messages to identify root cause (invalid credentials, API changes, etc.)
Use Tenant Filtering: When troubleshooting specific advertiser issues, filter to that tenant to isolate queue problems.
Coordinate with Platform Processing Toggles: Use platform-level processing toggles (in Platforms admin page) for emergency controls instead of pausing entire queue.

Next Steps

Learn about system monitoring and platform status tracking for complete operational visibility.