Monitor a load balancer

Follow a load balancer from the Load Balancers list and its tabs, which show its status, the health of its servers, alerts, live traffic statistics and graphs, and a record of every action and charge. To keep request logs, send access logs to a bucket; see Save access logs to a bucket.

Before you begin

• You have a load balancer. See Create an Application Load Balancer.
• The load balancer reads Running or Backend Connection Failure, if you want to use the Monitoring, Alerts, Stats, and Prometheus Stats tabs.

Backend Mapping is available once the load balancer has left Creating, Deploying, or Deleting. Action Logs, Billing Logs, and Activity Timeline are available in every status.

Find the load balancer in the list

In MyAccount, under Compute, click Load Balancers. The list shows the load balancers of the current project and region, five to a page to start with, with:

• LB Name, marked as encrypted when the load balancer is encrypted, with its tags under the name.
• LB Type: External or Internal.
• Mode: the frontend protocol, such as HTTP or BOTH, or TCP for a Network Load Balancer.
• Public IP, which reads Not Assigned when there is none, and Private IP.
• Status, and Actions.

You can search and filter by name, public IP, private IP, and status, filter by tag, refresh the list, and show 5, 10, 20, or 50 load balancers to a page. You can also manage each load balancer’s tags from the list. Click a load balancer to open its tabs below the list.

Read the status

Status	What it means
Creating	The load balancer is being created for the first time. With Mode at HTTP or TCP, this takes about four to seven minutes.
Deploying	Your saved change is being applied: an edit on Info, Backend Mapping, or ACL, a VPC change, a scale group change, or an upgrade.
Running	The load balancer is up, and every backend server passes its checks.
Backend Connection Failure	At least one server fails its checks, and the others keep serving. It can also show while every server passes, when MyAccount receives no health data for the load balancer; see The load balancer shows Backend Connection Failure.
Powered off	The load balancer is off and serves no traffic. It keeps its settings, and is still billed.
Upgrading	The load balancer is moving to a larger plan.
Deleting	The load balancer is being removed.
Error	The settings could not be applied. You can still change the settings and save, which applies them again.

For what to check when the status reads Backend Connection Failure, see The load balancer shows Backend Connection Failure.

Actions and the Info edits become available again once the load balancer leaves Creating, Deploying, or Upgrading. MyAccount estimates two to three minutes for Deploying in the list, and five to 10 minutes in the message after you save a change. If a status lasts far longer, see The load balancer stays on Creating, Deploying, or Upgrading.

Check the servers’ health

Open the Backend Mapping tab. It lists each server with its name, IP, and port, and marks the ones the load balancer reports DOWN. See Manage backend groups and servers.

Read the graphs

Open the Monitoring tab to follow the load balancer over time. It shows CPU Load, CPU Utilization, Disk Read Operation, Disk Write Operation, Disk Utilization, Memory Utilization, and Network Traffic. The graphs cover the period you pick: Last Hour, Last 6 Hours, Last 12 Hours, Last 24 Hours, or Last 1 Week.

The graphs fill in once the load balancer has run for a while.

Set up alerts

On the Alerts tab, click Create Alert, and choose what to watch: Processor Load, Consumed Memory, or Free Disk Space. Then set the operator, the threshold, the severity, and the alert groups to notify, as for a node; see Set up alerts on a node.

To change your existing alerts, click Monitoring Dashboard on the Alerts tab. It opens Manage Alerts, which lists every alert with Edit, Pause or Resume, and Delete. Deleting the load balancer deletes its alerts too.

Read the live statistics

The Stats tab shows live tables from the load balancer: http-in for HTTP traffic, ssl-in for HTTPS traffic, and one table for each backend group. Each lists the queue, sessions, bytes, denied requests, errors, and warnings. In a backend group’s table, the Server columns include each server’s Status and LastChk. The sessions total also counts the responses by status: HTTP 1xx responses, HTTP 2xx responses, HTTP 4xx responses, HTTP 5xx responses, and other responses. Click Refresh LB Stats to load the latest figures.

The Prometheus Stats tab charts the same kind of figures over time. Under Select Graph, choose what to chart:

• Backend, the default, charts the backend group you pick under Select Backend Group: its responses, bytes, queued connections, sessions, errors, and server status.
• Frontend charts the load balancer’s HTTP listener: its responses, sessions, request errors, denied requests, and bytes.

Pick a period under Select Interval, from Last Hour to Last 1 Week.

You can read these metrics only in MyAccount, on Stats and Prometheus Stats.

Read the logs

• Action Logs lists the actions taken on the load balancer, with Event, Client IP, Details, and Timestamp. Request logs go to your bucket instead; see Save access logs to a bucket.
• Billing Logs lists the load balancer’s charges, for prepaid and postpaid accounts alike. Name describes the charge, and Transaction Type is CREDIT or DEBIT. Then come Amount and Deducted at. With no charges, the tab says Currently No Logs are present.
• Activity Timeline lists the actions taken on the load balancer in the month you pick under Select Month. Each shows its event name, such as CREATE, UPDATE, POWER_OFF, or UPGRADE_PLAN, its Timestamp in IST, its Details, who did it with their IP address, and Success or Failure. A month with nothing in it says No timeline found for the selected month and year.

If the status reads Backend Connection Failure or Error, see Troubleshooting Load Balancer.