AMQP Queues

Payment Nexus uses LavinMQ as its AMQP message broker to process background jobs asynchronously — inbound webhook callbacks from PSPs, outbound notifications to CRM systems, scheduled tasks, and more. The AMQP Queues page surfaces the real-time state of every queue in the broker so you can see what's flowing, what's stuck, and what has failed.

Permissions

Action	Permission required
View queue status	queues:monitor + application:administer
Replay or discard messages	queues:manage + application:administer

Summary metrics

Four headline figures appear at the top of the page, updated in real time via WebSocket:

Metric	What it shows
Active Queues	Queues with a current publish or consume rate greater than zero, shown as active / total
Publish Rate	Messages entering the broker right now, in messages per second
Consume Rate	Messages being delivered to or fetched by consumers right now, in messages per second
Backlog	Total messages waiting across all queues — the sum of ready and unacknowledged messages

TIP

A healthy system can show zero backlog while still flowing. Backlog shows depth, not throughput. A fast-draining queue may process thousands of messages per second while maintaining near-zero depth.

Operational queues

The main table lists all operational queues. Click any row to open the queue's detail page.

Column	Description
Name	Queue identifier
State	Broker-reported queue state (e.g. running)
Activity	Flowing — active publish or consume rate; Idle — past activity but none now; No activity yet — never used
Messages	Total depth: ready + unacknowledged
Ready	Messages waiting to be delivered to a consumer
Unacknowledged	Messages delivered to a consumer but not yet acknowledged (in-flight)
Consumers	Number of active consumers connected to this queue
Publish rate	Messages per second currently entering this queue
Consume rate	Messages per second currently being delivered from this queue
Published	Lifetime total of messages published to this queue
Consumed	Lifetime total of messages consumed from this queue

Queue detail

The detail page repeats the same metrics as summary cards and adds a Peek action. Peeking loads up to 10 messages from the front of the queue and displays their routing key, payload size, and raw payload. Messages are immediately requeued after being peeked — it is a read-only operation.

Dead-letter queues

Every operational queue has a corresponding dead-letter queue (DLQ), named dlq.{queue-name}. When a message is rejected or cannot be processed — for example because a PSP webhook payload was malformed or a downstream service was unavailable — it is routed to the DLQ rather than being silently dropped.

The DLQ section at the bottom of the main queues page lists all dead-letter queues with the same metrics as the operational table. Click a DLQ row to open its detail page, which shows the failed messages and provides four actions:

Action	What it does
Replay	Move the oldest message back to its source queue for reprocessing
Discard	Permanently delete the oldest message
Replay all	Move all messages in the DLQ back to their source queue
Discard all	Permanently clear the entire DLQ

Per-message replay and discard are also available directly in the message table rows. All destructive actions require confirmation.

WARNING

Replaying a message re-submits it to the same queue that originally rejected it. If the underlying problem has not been resolved, the message will fail again and return to the DLQ.