After the introduction of custom interfaces for host prototype we are able to solve several problems caused by the 'not real ips' of discovered hosts:
- inaccessible HV host for direct monitoring via ping (for example)
- inaccessible VM host for direct monitoring via agent
- redundant discovered VM hosts (throw vmware discovery and auto registration of active agent)
In general case we have no possibility to solve the problem, because we can obtain info about guest ip and guest OS host name through running vmware guest tools only. If vmtools is not running, required info is not available.
Second problem - guest OS host name may not be unique.
But for cases which support "Fully Qualified Domain Name" (fqdn) - it is acceptable
- Change host prototypes and set real or empty ip for HV and VM in current template "VMWare". Add to template HV ping check.
- Create a new additional vmware template like: "VMWare FQDN", where:
- set real ip to discovered HVs and make ping check
- set real ip to discovered VMs
- set real name of guest vm equal of active agent name
- if guest OS host name is not accessible, than vm host will not be discovered. Disabled hosts and hosts without vmtools will not be discovered, because host id is absent.
extend vmware.vm.discovery with following fields with empty values in json if properties are not accessible:
- "Linux 3.10.0-1127.13.1.el7.x86_64 CentOS Linux release 7.8.2003 (Core)"
extend vmware.hv.discovery with following field:
|Create template "VMWare FQDN"||Closed|