Maintenance & Operations

Best practices for keeping the queue system healthy over time.

Routine Tasks

Task	Frequency	Purpose
Review failed jobs	Daily	Identify systemic errors
Purge old completed jobs	Weekly	Control DB growth
Rotate request logs	Automatic (size)	Ensure disk space hygiene
Verify Redis memory usage	Weekly	Prevent eviction / performance issues
Test worker trigger	Weekly	Early detection of environment regressions

Purging Strategies

Built-in purge tools exist via admin UI. Recommended policies:

• Completed jobs older than 7 days
• Failed jobs older than 30 days (after triage)
• Cancelled jobs older than 7 days

Consider a WP-CLI command (future enhancement) for scripted purges.

Handling Stuck Jobs

Symptoms: Jobs remain processing beyond expected timeout.

1. Run the “Reset Stuck Jobs” admin action (resets to queued).
2. Investigate root cause (PHP fatal, external timeout).
3. Add defensive timeouts inside job execution.

Log Management

Request logging (if enabled) rotates by size:

• Tune log_rotate_size_kb for expected traffic
• Adjust log_max_files based on retention needs
• Centralize logs (e.g., ship to ELK) for correlation across hosts

Database Hygiene

Potential future cron to purge directly:

// Sketch example
add_action( 'redis_queue_maintenance', function() {
  global $wpdb; $t = $wpdb->prefix . 'redis_queue_jobs';
  $wpdb->query( $wpdb->prepare("DELETE FROM $t WHERE status='completed' AND created_at < %s", gmdate('Y-m-d H:i:s', time()-7*DAY_IN_SECONDS) ) );
});

Schedule with:

if ( ! wp_next_scheduled( 'redis_queue_maintenance' ) ) {
  wp_schedule_event( time()+300, 'hourly', 'redis_queue_maintenance' );
}

Capacity Planning

Indicators to add more workers or increase batch size:

• Sustained backlog growth (queued count rising hour-over-hour)
• High average wait time (difference between created_at and processed_at)
• Repeated rate limiting or API throttling requiring slower pacing

Disaster Recovery

If Redis data is lost:

• Database table still has historical job metadata but jobs in-flight or queued are gone
• Reconstruct critical jobs by scanning for statuses not completed and re-enqueue with a script

Security Upkeep

• Rotate API token periodically (document internal runbook)
• Audit allowed routes for scoped tokens via new filter examples
• Monitor logs for anomalous route access patterns

Metric Ideas (Not yet implemented)

• Queue length over time per queue
• Job execution latency distribution
• Failure rate by job type
• Retry attempt histogram

Future Enhancements

• CLI commands for purge & requeue
• Metrics endpoint / integration export
• Automatic backoff tuning based on failure types

See scaling.md for throughput strategies.

📦 Source: soderlind/redis-queue · Edit on GitHub