Skip to content

Add health checks and syslogging to HAProxy via a syslog container

justin requested to merge improve-haproxy-reliability into master

HAProxy has an annoying feature where DNS needs to be resolved at boot, or else it would crash. Using init-addr I can prevent the crash, allowing HAProxy to work for those servers that are resolvable when HAProxy boots. However, any server not resolvable when HAProxy boots will never be resolved. I see a future of many docker kill -s HUP.

Also, HAProxy only logs to syslog, so I added a syslog container that pipes those logs to stdout, so they are inspectable. I also added L4 health checks to HAProxy, so it can disable servers that it can't resolve instead of trying too hard and waiting.

Merge request reports