Skip to main content

Load Balancer Monitoring, Alerts, and Stats

A load balancer exposes four observability tabs — Monitoring, Alerts, Stats, and Prometheus Stats. They are available when the load balancer is Running, in Backend Connection Failure, or Backend Status Unavailable.

Use Monitoring for resource graphs over time, Alerts to be notified of problems, Stats for live HAProxy session data, and Prometheus Stats for metrics you can chart or scrape.

Monitoring​

The Monitoring tab shows resource graphs collected from the load balancer, with a time-range selector: Last Hour, Last 6 Hours, Last 12 Hours, Last 24 Hours, and Last 1 Week.

Typical graphs include CPU load, CPU utilization, disk read and write operations (IOPS), disk utilization, and memory utilization. Graphs require monitoring to be active for the load balancer.

Alerts​

The Alerts tab manages health alerts for the load balancer. Existing alerts are listed with their trigger type, condition, service, and alert group; select Create Alert to add one.

Create an Alert​

The Create Alert form has a single Alert Trigger section. The available fields and values are specific to load balancers:

FieldWhat to set
Alert trigger type (Trigger)The metric to watch — Processor Load, Consumed Memory, or Free Disk Space. Required.
VolumeShown only when the trigger is Free Disk Space. Choose Node Volume (Default) or a specific attached volume to monitor. Required.
Operator (Trigger operator)Greater than, Less than, Equal to, or Not equal to. Required.
Threshold value (%) (Trigger threshold)The percentage that triggers the alert — a value between 10 and 100. Required.
Severity (Set alert Severity)Not classified, Information, Warning, Average, High (default), or Disaster. Required.
Alert Group(s)One or more alert groups to notify when the alert fires. Required. Use the + (Create New Alert Group) control to create a new group.

An alert group is the set of recipients an alert notifies — alerts are routed to alert groups rather than to individual email addresses entered on the alert. You can select multiple groups, and create a new one inline with the + control.

Manage Alerts​

Each alert in the list shows its trigger type, condition (operator + threshold + severity), service, and alert group, with row actions:

  • Edit — change the trigger type, operator, threshold, severity, or alert groups.
  • Pause / Resume — temporarily stop notifications, then restart them. A paused alert shows as stopped.
  • Delete — remove the alert.

Stats​

The Stats tab shows live HAProxy operational metrics, refreshed on demand with Refresh LB Stats. It is organized into front-end (HTTP and SSL) and per-backend sections.

For each listener and server it reports:

  • Sessions — current, max, limit, and total; plus session rate (current, max, limit).
  • Queue — current, max, and limit.
  • Bytes — in and out.
  • Denied — denied requests and responses.
  • Errors — request, connection, and response errors.
  • Warnings — retries and redispatches.
  • Server health — status, last check, weight, active/backup, checks, downs, downtime, and throttle.

Backend rows also expose cumulative connection/session/HTTP-request counts, an HTTP response-code breakdown (1xx/2xx/4xx/5xx), and average response times (queue, connect, response, and total). These metrics help you confirm backend health and spot errors or saturation.

Prometheus Stats​

The Prometheus Stats tab charts HAProxy metrics in Prometheus format, with the same time-range selector as Monitoring and a backend selector. Choose a metric to chart from the backend or frontend families, for example:

  • Backend — response codes (haproxy_backend_http_responses_total), bytes in/out, current/queued sessions, cumulative sessions, average response time, and backend up/down (haproxy_backend_up).
  • Frontend — HTTP responses (haproxy_frontend_http_responses_total), session rate (current/max/limit), sessions, request errors, requests denied, and bytes in/out.

These metrics are fetched from the load balancer's historical-data endpoint and can be charted here or scraped into your own Prometheus stack.

ResourceUse it for
Node Monitoring and AlertsFull reference for graphs and alerts.
Manage Backend MappingSee and fix backend health.
Troubleshoot Load BalancersDiagnose Backend Connection Failure and related states.
Last updated on June 5, 2026.