we have our agents behind proxies and we monitor disk io but disable it during the backup hours from 17h30 till 24:00 and from 00:00 till 8:15
one of our hypervisors was in error state so we had to restart all our agents
because of this we have data as you can see in the graph around 18:00
so we had for all vm's an alarm for high disk IO
from this i can conclude that the agent still sends data to the zabbix proxy and proxy to zabbix server even when custom intervals are set and agent is restart during the custom intervals.
proxy and zabbix server never got restarted