Steps to reproduce:
- Create Hosts to simulate a common office environment (e.g. router, switches, physical server containing multiple VM's)
- Assign Templates to the Hosts (e.g. ICMP Ping, SSH connectivity, RDP connectivity, Web Monitoring etc)
- Attempt to create the required dependencies between triggers (e.g. VM ICMP depends on Physical Host ICMP, Physical Host ICMP depends on Switch ICMP, Switch ICMP depends on Router ICMP.
Can no longer create these dependencies.
Able to create the dependencies as we could previously
We are monitoring a customer with 20,000 Hosts. If we lose one of the core switches close to the Zabbix infrastructure that could be up to 20,000 Hosts generating events for ICMP Ping Down, SSH Down, RDP Down, Web Monitoring Failed etc etc.
Based on previous experience with event storms like this we don't believe the event's table and event correlation can support this load which is why we automate the creation of these important dependencies via host management scripts.
Please note that both Inter Template (e.g. Zabbix Agent not Responding being able to depend on the ICMP Ping Down trigger on the same Host) dependency is important, but also Inter Host (e.g. ICMP Ping on VM depends on ICMP Ping of the Physical Host)