Queue Monitoring
Get full visibility into your message queues and streams. Senzor connects to your broker read-only (or runs a lightweight collector inside your network) and tracks backlog depth, throughput, consumer count, oldest-message age, drain ETA, and dead-letter accumulation — across BullMQ (Redis), RabbitMQ, Apache Kafka, and AWS SQS. When you also run @senzops/apm-node, queue backlog is correlated to the exact consumer executions draining it.
Prerequisites
- A supported broker: BullMQ (Redis), RabbitMQ, Apache Kafka, or AWS SQS.
- Read-only access — a Redis ACL, RabbitMQ management user, Kafka consumer-group read, or an IAM user scoped to sqs:ListQueues and sqs:GetQueueAttributes.
1. Dashboard Configuration
Before injecting code into your application, you must provision an API key from the Senzor dashboard.
Add a Queue Source
Click '+' next to Queues and choose your broker system (BullMQ, RabbitMQ, Kafka, or SQS).
Choose Connection Method
Agentless — Senzor polls your broker on an interval. Or Collector — run the Senzor queue collector inside your own network (for locked-down / VPC environments) and it pushes metrics to us.
Provide Credentials
Enter the broker connection details. Secrets are AES-256 encrypted at rest; for BullMQ the key prefix (including the Redis-Cluster '{bull}' hash-tag) is auto-detected.
Set Interval & Runbook
Choose the polling frequency (1m, 5m, or 15m). Optionally add a management/runbook URL for one-click hand-off when dead letters accumulate.
Troubleshooting & Edge Cases
Metrics show 0 / no queues discovered
For BullMQ, confirm the key prefix matches your broker — Senzor auto-detects the common 'bull' and Redis-Cluster '{bull}' prefixes. Ensure the connection points at the same Redis database your workers use.
Throughput is 0 right after adding a source
Throughput is derived across consecutive samples, so it appears after the second poll cycle. A backlog of 0 is normal for a healthy, drained queue.