AWS / CloudWatch / MQ

AWS Amazon MQ for RabbitMQ Monitoring Dashboard for Grafana - CloudWatch Metrics

A Grafana dashboard for monitoring AWS Amazon MQ (RabbitMQ) brokers using CloudWatch SEARCH expressions. Broker-wide overview plus per-node and per-queue drill-down for capacity utilization, resources, message throughput, and queue backlogs - built specifically for the metrics Amazon MQ exposes (not the raw RabbitMQ Management API).

Why monitor AWS Amazon MQ for RabbitMQ?

Amazon MQ failures usually look the same: a queue starts backing up because consumers crashed or slowed down; memory pressure rises as messages accumulate; the broker hits its connection or channel limit and refuses new clients; or replication lag between cluster nodes grows until the cluster goes into split-brain. This dashboard surfaces all of those signals using only what CloudWatch publishes - no need to expose the RabbitMQ Management UI or run a Prometheus exporter inside the broker.

It uses CloudWatch SEARCH expressions to discover queues and nodes automatically, so you don't have to enumerate them in the dashboard.

Features

• Health overview - total connections, channels, queues, exchanges, messages, worst-node memory %
• Capacity and limits - connection and channel utilization gauges as % of configurable limit (with full per-instance-class lookup table in the panel descriptions)
• Cluster resources per node - System CPU, memory used vs limit, disk free vs disk low watermark, open file descriptors
• Throughput - broker-wide publish, confirm, and ack rates, plus a publish−ack delta panel that goes positive when backlog is growing
• Queue backlogs - top 15 queues by MessageReadyCount, top 15 by MessageUnacknowledgedCount, and aggregate ready vs unacked across the entire broker
• Single queue detail - selected queue's message counts (total / ready / unacked) and consumer count over time

Capacity Limits by Instance Class

Amazon MQ enforces per-node limits on connections and channels by instance class. The dashboard ships with a starter value (single mq.m7g.xlarge node = 10,000 connections / 30,000 channels) and a full lookup table in the gauge panel descriptions:

Multiply by 3 for cluster totals. RabbitMQ enforces the per-node limit, not the cluster total - a single node hitting its limit will refuse new connections even if the cluster-wide count is well under capacity.

Key CloudWatch Metrics Used

All metrics are from the AWS/AmazonMQ namespace, dimensioned by Broker (and Node or Queue where applicable).

ConnectionCount, ChannelCount

Active AMQP client connections and open channels broker-wide. RabbitMQ supports up to 2,047 channels per connection (channel_max default); the broker-wide ceiling is set by instance class.

QueueCount, ExchangeCount, MessageCount

Total declared queues, declared exchanges (including default and system exchanges), and messages aggregated across all queues.

RabbitMQMemUsed, RabbitMQMemLimit

Per-node memory used and configured limit. When used approaches limit, RabbitMQ enters memory alarm state and blocks publishers.

RabbitMQDiskFree, RabbitMQDiskFreeLimit

Per-node disk free and low-disk threshold. If free descends toward the limit, publishing on persistent queues blocks.

SystemCpuUtilization, RabbitMQFdUsed

Per-node CPU and open file descriptor count.

PublishRate, ConfirmRate, AckRate

Broker-wide message rates. Sustained PublishRate - AckRate > 0 means consumers can't keep up with publishers - the backlog is growing.

MessageReadyCount, MessageUnacknowledgedCount

Per-queue counts of messages waiting for delivery and messages delivered but not yet acked. Sustained high MessageReadyCount = consumer shortage; sustained high MessageUnacknowledgedCount = slow or stuck consumers, or aggressive prefetch.

ConsumerCount

Per-queue consumer count. Drops to 0 while message count grows = consumer outage on that queue.

Prerequisites

• Grafana 10.0 or later
• AWS CloudWatch datasource plugin configured in Grafana
• IAM permissions on the role/user backing the datasource:
  • cloudwatch:GetMetricData
  • cloudwatch:ListMetrics
  • cloudwatch:GetMetricStatistics

No Amazon MQ permissions are required since metrics flow through CloudWatch.

Installation

1. Download the dashboard JSON.
2. In Grafana, go to Dashboards → New → Import.
3. Paste the JSON or upload the file.
4. When prompted, select your AWS CloudWatch datasource.
5. Click Import.
6. Adjust the "Max Connections" and "Max Channels" textboxes at the top of the dashboard to match your broker's instance class (see lookup table above).

Variables

• Region - AWS region of your broker
• Broker - auto-populated from CloudWatch with every broker that publishes metrics
• Node - node within the selected broker
• Virtual Host - vhost within the selected broker
• Queue - queue within the selected broker (drives the Single Queue Detail panels)
• Max Connections / Max Channels - textboxes for capacity limits; set per your instance class

Troubleshooting

Q: The Connection Utilization gauge is showing 200%. A: The Max Connections textbox is set too low. Update it to match your instance class - see the lookup table in the panel description.

Q: Memory % goes above 100% sometimes. A: The calculation is MAX(MemUsed) / MAX(MemLimit) * 100. Occasional spikes above 100% in 5-minute windows reflect the broker briefly using burst memory; sustained above 100% means a memory alarm.

Q: Per-queue panels are empty for some queues. A: Amazon MQ only publishes per-queue metrics for queues that have had activity recently. Idle queues won't appear.

Q: Does this work for ActiveMQ brokers too? A: No - this dashboard is RabbitMQ-specific. ActiveMQ on Amazon MQ publishes different metrics (ActiveMQ.JobScheduler.Store, etc.).

Related AWS Documentation

• Monitoring Amazon MQ for RabbitMQ with CloudWatch
• Amazon MQ for RabbitMQ broker instance types
• Quotas in Amazon MQ for RabbitMQ

Tags

aws amazon-mq rabbitmq messaging cloudwatch monitoring aws-monitoring message-queue amqp broker-monitoring grafana-dashboard