Service SLA not being recalculated after receiving data from proxy

[daftu]

Hi,
we're monitoring some service availability from 3 different locations (specifically from 3 VPS around the world). Each location has it's own Zabbix Proxy to cache data in case of Zabbix Server unavailability.
There is one central Trigger defined with expression as follows - it fires only when service is unavailable from all three locations:

({location1:status.regexp(^UP$)}<>1 or {location1:status.nodata(5m)}=1) and ({location2:status.regexp(^UP$)}<>1 or {location2:status.nodata(5m)}=1) and ({location3:status.regexp(^UP$)}<>1 or {location3:status.nodata(5m)}=1)

Everything works fine while there is working internet connection between Zabbix Server and Proxies in their locations. Last week we had some ISP problems, causing trigger to fire and being active for about 7 hours. After internet connection was restored all missing data from 7 hours has been received from proxies (so the monitored services actually worked fine), but when I look at Monitoring - Services there is an SLA degradation for a 7 hours - the data wasn't updated!

This is not what I expected creating separate Zabbix Proxy for each location to prevent internet connection downtimes affecting SLA.

Is this normal behavior? How to avoid it?

[gofree]

I think its like this - it services availability ( SLA ) is counted on the time of the problem duration ( aka trigger assigned to some service in the service tree ).

So even if you have data from the proxy after connection has been restored - the event evaluation is being done on zabbix server - your event has been generated for the service apparently with nodata trigger ( i guess ) and thus the gap in SLA.

In general I think SLA ( it services ) in zabbix have kina limited usage potentiall as the service trees are not being generated dynamically ( and no data triggers ) .

On the other side you we're blind for 7hrs so from customer perspective its affceting the SLA of monitoring service.