HAProxy - Run all checks before accepting connections?

Question

I'm using haproxy to provide access to a redis cluster. But i'm experiencing issues when haproxy is starting since all servers are initially up, before the first checks can be run.

Is there a way to force haproxy to run the initial checks before accepting connections?

Here is my config:

global
    max-spread-checks 1

defaults
    mode tcp
    timeout connect 4s
    timeout server 10s
    timeout client 60s

frontend ft_redis
    mode tcp
    bind *:6379
    default_backend bk_redis

resolvers docker_resolver
    nameserver dns 127.0.0.11:53

backend bk_redis
    mode tcp
    option tcp-check

    tcp-check send info\ replication\r\n
    tcp-check expect string role:master
    tcp-check send QUIT\r\n
    tcp-check expect string +OK

    server redis-backend-1 redis_node1:6379 maxconn 1024 check inter 1s downinter 2s rise 1 fall 1 resolvers docker_resolver resolve-prefer ipv4 on-marked-down shutdown-sessions
    server redis-backend-2 redis_node2:6379 maxconn 1024 check inter 1s downinter 2s rise 1 fall 1 resolvers docker_resolver resolve-prefer ipv4 on-marked-down shutdown-sessions
    server redis-backend-3 redis_node3:6379 maxconn 1024 check inter 1s downinter 2s rise 1 fall 1 resolvers docker_resolver resolve-prefer ipv4 on-marked-down shutdown-sessions

It normally works fine and only connects to the current redis master.

But when starting/restarting haproxy I eventually get errors like "READONLY You can't write against a read only slave." because all servers are initially marked as up and the connection goes to a slave redis server.

HAProxy - Run all checks before accepting connections?

0 Answers0