Some things i’v learnt after few days on some site

Cluster

Install repos for RHEL7 , yum install rabbitmq-server and other stuff:
erlang-20.2.2-1.el7.centos.x86_64 rabbitmq-server-3.7.2-1.el7.noarch glibc-common-2.17-196.el7.x86_64 python2-pip-8.1.2-5.el7.noarch haproxy-1.5.18-6.el7.x86_64 haproxy-1.5.18-6.el7.x86_64 1:net-snmp-libs-5.7.2-28.el7.x86_64 1:net-snmp-agent-libs-5.7.2-28.el7.x86_64 keepalived-1.3.5-1.el7.x86_64 2:nmap-ncat-6.40-7.el7.x86_64 2:nmap-6.40-7.el7.x86_64 3:traceroute-2.0.22-2.el7.x86_64 arptables-0.0.4-8.el7.x86_64 td-agent-bit-0.12.14-1.x86_64

Install plugins: wget https://github.com/deadtrickster/prometheus_rabbitmq_exporter/releases/download/v3.7.2.1/accept-0.3.3.ez -O /usr/lib/rabbitmq/lib/rabbitmq_server-3.7.2/plugins/accept-0.3.3.ez wget https://github.com/deadtrickster/prometheus_rabbitmq_exporter/releases/download/v3.7.2.1/prometheus-3.4.5.ez -O /usr/lib/rabbitmq/lib/rabbitmq_server-3.7.2/plugins/prometheus-3.4.5.ez wget https://github.com/deadtrickster/prometheus_rabbitmq_exporter/releases/download/v3.7.2.1/prometheus_cowboy-0.1.4.ez -O /usr/lib/rabbitmq/lib/rabbitmq_server-3.7.2/plugins/prometheus_cowboy-0.1.4.ez wget https://github.com/deadtrickster/prometheus_rabbitmq_exporter/releases/download/v3.7.2.1/prometheus_httpd-2.1.8.ez -O /usr/lib/rabbitmq/lib/rabbitmq_server-3.7.2/plugins/prometheus_httpd-2.1.8.ez wget https://github.com/deadtrickster/prometheus_rabbitmq_exporter/releases/download/v3.7.2.1/prometheus_process_collector-1.3.1.ez -O /usr/lib/rabbitmq/lib/rabbitmq_server-3.7.2/plugins/prometheus_process_collector-1.3.1.ez wget https://github.com/deadtrickster/prometheus_rabbitmq_exporter/releases/download/v3.7.2.1/prometheus_rabbitmq_exporter-v3.7.2.1.ez -O /usr/lib/rabbitmq/lib/rabbitmq_server-3.7.2/plugins/prometheus_rabbitmq_exporter-v3.7.1.1.ez

Enable all plugins
Install keepalived, with basic config
keepalived.conf

Configure track_script
chk_rabbit.py

Restart keepalived
service keepalived restart

Customize net_ticktime
vim /etc/rabbitmq/advanced.config [ {kernel, [{net_ticktime, 10}]} ].

Enable autoheal of partitions
vim /etc/rabbitmq/rabbitmq.conf cluster_partition_handling = autoheal

On each node create the SAME erlang cookie (value doesn’t matter it just need to be the same across nodes)
vim /var/lib/rabbitmq/.erlang.cookie DntiDNrDJSnrne
Update daemon to increase limit for open files
vim /usr/lib/systemd/system/rabbitmq-server.service LimitNOFILE=65536
Reload & Restart
systemctl daemon-reload && systemctl restart rabbitmq-server

On every SLAVE node run this stuff to join cluster
rabbitmqctl stop_app rabbitmqctl join_cluster rabbit@rabbit_master rabbitmqctl start_app

Check that everything is fine
rabbitmqctl cluster_status

Configure vhosts/users/policies and etc
rabbitmqcl create_vhost blabla rabbitmqctl -p blabla set_policy ha-all "^ha\." '{"ha-mode":"all","ha-sync-mode":"automatic", "queue-master-locator":"client-local"}' # I remind to you that only ONE policy works for each queue, so be aware