Skyline for Laravel vs Cbox Queue Monitor

Cbox's Laravel Queue Monitor is a free, MIT-licensed package that records every attempt of every job in a database table, on any queue driver, and pairs it with a companion package, Laravel Queue Metrics, for aggregate numbers. It is the most thorough free queue monitor for Laravel we know of, and on several points it is ahead of Skyline for Laravel. Skyline replaces laravel/horizon on Redis, alerts when the queue stops working and lets you act on it.

This page describes Queue Monitor 1.12.1 and Queue Metrics 3.5.2, the releases current in November 2026.

Two packages, one dashboard#

Queue Monitor listens to Laravel's queue events and writes one row per attempt to a queue_monitor_jobs table: the job, its queue and server, its outcome, its duration, the exception if it failed, and the CPU and memory it used. A retried job keeps its whole chain, so you can see each attempt and where it ran. The rows feed a web dashboard and a terminal UI.

Queue Metrics adds the numbers that are not about any one job: queue depth and the age of the oldest waiting job, worker heartbeats, throughput, health scores and a Prometheus endpoint. It stores them in Redis by default, or in the database for small setups, and it registers scheduled commands, so the scheduler has to be running.

At a glance#

Where Cbox Queue Monitor is the better answer#

• You are not on Redis. It works on the database driver and SQS as well as Redis, and Skyline, like Horizon, does not.
• You want every attempt on record. One row per attempt, with its server, exception, CPU and memory, gives you a retry timeline Skyline does not keep, in a database you can query.
• You track latency by percentile. p50, p95 and p99 per job class, queue and server, and a pickup-SLA percentage per queue, are things Skyline does not report.
• You want to re-run a job that succeeded. Replay works on any job whose payload was stored, in bulk if needed. Skyline retries failed jobs and runs delayed ones early, nothing more.
• Budget is the constraint. Both packages are free and MIT-licensed.

Things to know before you rely on its numbers#

Its success rate is counted per row, and each row is an attempt. A job that fails twice and then succeeds counts as two failures and one success, so a queue with healthy retries reads worse than its jobs did. The attempt column can also read "3/1" on a released job, because the maximum is taken from the job's $tries property alone and ignores the worker's --tries and retryUntil().

Several features are off outside the local environment by default, for good reasons: payload storage, which replay needs, and the REST API. Job arguments are redacted on the dashboard unless you take command out of the sensitive keys. Pruning is a command you schedule yourself.

Where Skyline is the better answer#

• You need to be told. Cbox's alert conditions appear when someone opens the dashboard or calls the API. Skyline checks every minute and sends each alert over mail, Slack, SMS or a webhook, routed by severity, with a recovery message when it clears.
• Horizon itself stops. Queue Metrics can show stale workers on its dashboard, but nobody hears about it until they look. Skyline's horizon:check runs from the scheduler, outside the fleet, and sends the alert.
• You need to act on a live queue. Pausing one queue during an incident or running a retrying job the moment the fix is deployed are operations on Redis that only the Horizon install can perform.
• The job never ran. A dispatch skipped by a held ShouldBeUnique lock leaves no row, and neither does a scheduled task that was due and never started. Skyline's Locks & Limits and Schedule screens show both.

Can you run both?#

Yes. Queue Monitor installs beside Horizon and reads extra data from it when it is present, such as supervisors and workload, and Skyline keeps the Laravel\Horizon namespace those reads go through. Cbox would keep the attempt history and percentiles, and Skyline the alerts and controls. Note that it takes a job's tags only from the job's own tags() method, not from the tags Horizon adds automatically.

Choosing between them#

If you are off Redis, or what you want is a detailed record and percentiles, Cbox Queue Monitor is free and very good at it. If you are on Redis and the problem is finding out late, then needing to do something about it, that is Skyline.