Description

Environment

• Zabbix Version: 7.0.30

• OS: Ubuntu 26.04

• Database: PostgreSQL

• Setup: 2-node native Zabbix Server HA cluster

Problem Description

In a 2-node native HA cluster, when a server node restarts and temporarily enters unknown mode, it immediately crashes with SIGSEGV (signal 11) the moment the HA manager determines cluster topology and transitions the node to standby mode.

After the crash, the server gets stuck in a temporary restart loop (cannot start server: found standby duplicate node) because the process terminated abruptly without updating its heartbeat record in the ha_node table, forcing the operator to wait ~60–75 seconds for the heartbeat to expire.

Steps to Reproduce

1. Configure a 2-node Zabbix HA cluster (e.g., Node 1 active, Node 2 standby).

1. Restart the standby node (systemctl restart zabbix-server).

1. The node starts up in unknown mode:
   codeText

   {{starting HA manager
   HA manager started in unknown mode
   "<node_name>" node started in "unknown" mode}}

1. A few seconds later, the HA manager switches the node to standby:
   codeText

   {{"<node_name>" node switched to "standby" mode}}

1. Result: The server crashes immediately with SIGSEGV.