IT Services SLA Hierarchy

[rogerdalvi]

Hello!

I'm a Zabbix user for more than two years now, but only recently we decided to use IT Services to monitor our service's SLA.

I'm having an issue that I can't tell if it is a normal zabbix behavior or if it is some king of misconfiguration or error.

I have an IT Service called Application Server HA with two child IT Services:
- Server1 Ping;
- Server2 Ping.

Server1 Ping has a 50% SLA and Server2 Ping has a 100% SLA.

The SLA of the parent service (Application Server HA) is 99.6%.

I always believed that The parent SLA would only consider problems when the child SLA is being affected, but that is not the case. Even if the child SLA is OK, the Parent SLA is affected because of the different SLA's between them.

Shouldn't the parent SLA consider the child SLA instead of the child problems?

Thanks.

[BDiE8VNy]

Well, I think there are valid use cases for both behaviors. So, personally I'd appreciate it to be configurable.

Currently status calculation is based on historic status value changes, which are stored in database for each service node.
The actual SLA value gets calculated on demand, when accessing respective frontend sections. This calculation doesn't consider SLA results of childs but only the status changes in relation to SLA threshold/uptime/downtime/ of the respective node.

While writing this, I noticed that the requested behavior might be valuable in context of ZBXNEXT-1673 too.
Since SLA calculation results are not considered by parrent service nodes one has to set correction periods on each affected parrent node.

Edit:
Maybe consider to create a corresponding feature request on https://support.zabbix.com - if non such exists already, of course.

[rogerdalvi]

Thanks for your reply BDiE8VNy! Now I have the confirmation that Zabbix behavior regarding relationship between parent and child services are working as designed.

That would be really nice if we could, as you suggested, configure this behavior the way we need it to be. I'll search if there's already a request to implement this feature and if not I'll propose it.

Meanwhile I'm going to deal with this in a different manner. I'll create a python script that uses the API to calculate my "IT Service child" SLA based on conditions we established. After that I'll create a trigger with the desired threshold and I'll use that trigger in the IT Service I'm creating. I spend some time thinking about this solution and for now I believe it's a good one.

Thanks again.