Clustered services monitoring

[bonnet.fh]

Hello,

I would like to create a trigger that create an alert if all services is offline.

For example, if you three devices that provide an HTTP service, each service is monitored with a trigger (with moderate severity). If one of them is offline, other can serve the HTTP service. However I want to trigger an alert if all of them were offline (with disaster severity).

The first idea is to create a trigger which aggregate all HTTP status like this:

{host1:service_state.last()}<>0 and {host2:service_state.last()}<>0 and {host3:service_state.last()}<>0

This trigger will work, but I need to modify it manually if I add a new host.

Are there a method to trigger an alert without manual trigger configuration (in case of host discovery)?

I'm thinking about tags. Are there a method for trigger an alert if all triggers with a tag is triggered?

Best regards,

[LenR]

I create an item per host that is "Number of whatever services running", in your case it would be 0 or 1. I put the hosts providing this service in a isolated host group, then use grpsum to sum the items counting service providers. Then a trigger on the grpsum item can be used for 0 services running or less than some critical value of services running.

The item of "number of services" can be in a template, so no extra config if more servers join the cluster.

I'm sure there are other ways, but I think I did this in the "before tags" era :-)