Tuning HAProxy Keepalive Probes: A Guide

how do I tune keepalive probes in haproxy, include links to documentation

To tune keepalive probes in HAProxy, you can adjust various parameters in the backend or listen section of your HAProxy configuration file. Here’s a guide on how to configure keepalive probes and the relevant documentation links:

1. Enable health checks for your backend servers by adding the “check” option to each server line:

backend my_backend
server server1 10.0.0.1:80 check
server server2 10.0.0.2:80 check

2. Configure the interval between health checks using the “inter” parameter. The default value is 2000ms (2 seconds):

backend my_backend
server server1 10.0.0.1:80 check inter 5000ms
server server2 10.0.0.2:80 check inter 5000ms

3. Set the maximum number of failed health checks before considering a server as down using the “fall” parameter. The default value is 3:

backend my_backend
server server1 10.0.0.1:80 check fall 2
server server2 10.0.0.2:80 check fall 2

4. Set the number of successful health checks required to consider a server as up again after being marked as down using the “rise” parameter. The default value is 2:

backend my_backend
server server1 10.0.0.1:80 check rise 3
server server2 10.0.0.2:80 check rise 3

5. Customize the health check request using the “option httpchk” directive. You can specify the HTTP method, URL, and expected response:

backend my_backend
option httpchk GET /healthz
http-check expect status 200
server server1 10.0.0.1:80 check
server server2 10.0.0.2:80 check

Relevant documentation links:

• HAProxy Server and Default Options: HAProxy version 2.4.15 - Configuration Manual
• Health Checks in HAProxy: Exploring the HAProxy Stats Page (What You Should Know)
• HAProxy Configuration Manual - Health Checks: HAProxy version 2.4.15 - Configuration Manual

These links provide detailed information on configuring health checks and keepalive probes in HAProxy.

Remember to reload your HAProxy configuration after making changes:

sudo systemctl reload haproxy

or

sudo service haproxy reload

Adjust the values according to your specific requirements and monitor your backend servers’ behavior to find the optimal settings for your environment.