Host monitoring takes place in every time interval configured in oned.conf
, managed by the HOST_MONITORING_INTERVAL
and HOST_PER_INTERVAL
parameters.
In large environments, it may be necessary to increase the monitoring interval or decrease hosts per interval to reduce overhead on the frontend and the hosts. Increasing the monitoring interval will lead to more outdated information about hosts and delay the failover host hooks too.
Please ensure, when you adjust these values, HOST_MONITORING_INTERVAL
is never lower than MANAGER_TIMER
, which is the time in seconds that the core uses to evaluate all the periodical functions.
The same logic applies for VM_POLLING_INTERVAL
and VM_PER_INTERVAL
, which are used to determine how frequently the VM instances are monitored.
A oned.conf
configuration for simpler infrastructures is as follows:
MANAGER_TIMER = 10
HOST_MONITORING_INTERVAL = 10
HOST_PER_INTERVAL = 25
VM_POLLING_INTERVAL = 10
VM_PER_INTERVAL = 100
In larger...