Haproxy disconnects from queue-worker, what's the problem with the config?

Question

RabbitMQ

please tell me: there is a cluster with haproxy, here is an excerpt from its config
defaults
log global
mode tcp
retries 2
timeout connect 4s
timeout check 5s
timeout client 3h
timeout server 3h

listen rabbitmq_messages
mode tcp
balance roundrobin
bind *:5672
option clitcpka # For TCP keep-alive
server server1 rabbitmq1:5672 maxconn 1000 check inter 3s fall 3 rise 2
server server2 rabbitmq2:5672 maxconn 1000 check inter 3s fall 3 rise 2
constantly haproxy drops the connection to yii-queue-worker (since there was no action) (apparently since server side)
What was done wrong, judging by the config, everything should be OK

Answer 1

What does HAProxy write in the logs? What does termination_state show ? The AMQP protocol includes heartbeats , do you have them enabled?

Answer 2

The problem was in disabled heartbeats from the server side. The connection was broken between haproxy and rabbitmq, not between worker and haproxy... so the worker didn't have any errors