Monitoring and alerts for BullMQ queues
Taskforce.sh checks your BullMQ queues every 5 minutes and alerts you when jobs fail, workers disappear, the backlog grows or Redis is running out of memory. You find out before your users do.
What is monitored
Enable the monitors you need on each connection. They cover every queue in that Redis instance.
Failed jobs
Alerts when new jobs fail in a queue. The alert includes the id of the job, when it failed and its stack trace.
Missing workers
Alerts when the number of workers connected to a queue drops below the minimum you set, for example after a crash or a bad deploy.
Backlog
Alerts when the waiting, prioritized and delayed jobs in a queue exceed a maximum, a sign that workers can't keep up.
Max memory
Alerts when Redis uses more memory than the threshold you set, so you can clean up or upgrade before it fills up.
Connection
Alerts when Taskforce.sh can no longer reach your Redis instance.
Alerts where your team works
Sent to the owner of the connection.
Slack
Posted to the Slack channel of your choice.
PagerDuty
Opens an incident so the on-call engineer gets paged.
No alert storms
When something breaks, many jobs tend to fail at once. Taskforce.sh sends one alert per queue and monitor, and waits until someone acknowledges it, or a day has passed, before alerting about the same queue again. Alerts that fire at the same time in a connection are sent together in one notification.
Acknowledged and triggered alerts stay in the dashboard, so you can see what happened and when. The monitoring documentation explains each monitor and its settings.
Monitoring questions
The monitors check every queue in a connection every 5 minutes.
Alerts are shown in the dashboard, and can also be sent by email, to a Slack channel or to PagerDuty. You choose the channels for each connection.
No. After a monitor triggers an alert for a queue, it waits until the alert is acknowledged, or until a day has passed, before alerting about that queue again. When many jobs fail at once you get one alert, not hundreds.
No. The monitors read the state of your queues from Redis, so your application and workers stay as they are.
No. You enable the monitors you need on each connection, and set their thresholds, for example the maximum backlog size.
Get your first alert in minutes
Connect your Redis, enable the monitors and choose where alerts go.
15-day free trial · No credit card required
