I have HA Netmaker configuration . Highly Available Installation (VMs/Bare Metal) https://docs.netmaker.org/server-installation.html?highlight=bare%20metal#highly-available-installation-vms-bare-metal Components: - 3 x servers with public ips - 3 x RQLITE nodes (cluster bootstrapped well). - 3 x (netmaker + mq) - Domains: dash.domain.com:443, api.domain.com:443, mq.domain.com:8883. - Round robin DNS to 3 public ips. - 10 netclient nodes connected to public API and MQ domains. Problem: All netclient nodes connect to different MQ broker hosts. New netclient node connects to one of the netmaker api hosts and sent join request. Only netclient daemons connected to the same MQ netmaker host received new updates (new peers). My expectations: If first netmaker server received new registration and write it to

rqlite

then all MQ brokers sent the same updates to subscripted netclient nodes. A don't understand how new configuration distributed to all MQ brokers via

rqlite

. How it works. All MQ broker hosts show correct logs, without any connect errors.

# MQ-01
1655237028: Client i8fqA1hhu2ro8EqwQKBFKco disconnected.
1655237051: New connection from <netclient-ip-01>:49212 on port 8883.
1655237051: New client connected from <netclient-ip-01>:49212 as RV9624UVZSzNIOssRsbbPc2 (p2, c1, k30, u'netclent-hostname-01').

# MQ-02
1655237051: Client RV9624UVZSzNIOssRsbbPc2 disconnected.
1655237119: New connection from <netclient-ip-02>:60314 on port 8883.
1655237119: New client connected from <netclient-ip-02>:60314 as 3O6EecgJZks4JKHcnRFRDrz (p2, c1, k30, u'netclent-hostname-02').
1655237119: Client 3O6EecgJZks4JKHcnRFRDrz disconnected.