HAProxy community

Layer4 timeout for 30 seconds


Can’t find any answer for this but I use HAproxy for LB on a Galera cluster. Every night I get a Layer4 connection timeout and the servers are not reachable by haproxy, lasts for like 30 seconds than they are up again. The VMs have run without problems, suspect there is something in my config, looks like this:

MySQL Cluster FE configuration

frontend mysql_cluster_frontend
bind *:3306
bind *:3307
mode tcp
option tcplog
default_backend galera_cluster_backend

MySQL Cluster BE configuration

backend galera_cluster_backend
mode tcp
option tcpka
balance leastconn
server db-01 check weight 1
server db-02 check weight 1

This is the error:
Server galera_cluster_backend/db-01 is DOWN, reason: Layer4 timeout, check duration: 2001ms.

No, there is nothing wrong with your configuration. Your problem is elsewhere. Tcpdump the health check traffic if you don’t believe haproxy.

Seems to happen when our second backup is copying the snapshots. That might be the problem, thanks!

1 Like