Kapacitor features information to overcome limitations of grafana alerting

1

Hi,

I was trying to configure alerting for CPU and few other metrics in grafana. I am using template variable for host selection and for few other things. Therefore using grafana alerting was not possible for me this way.

Then i have created separate dashboard without having any template variable.
In my graphs there are no variables now so alerting has been set. But it triggers for one series only and not for other.

Is this something which i can address by using kapacitor for alerting ?

Thanks,

2

Hi Ashu,

Kapacitor can certainly be used for alerting! It was built to do just that. I’m not sure what you mean by triggering for one series and not for another though.

Best regards,
dg

3

Hi David,

Here is explanation to my question.

Below limitation is there in grafana alerting.

I have a graph which shows me cpu utilization of 10 systems. Now what happens is. If system 2 has crossed the threshold it will alert without an issue but suppose if system 3 or any other system also croses the threshold during this time where another system(system 2) is already running over threshold and already triggered an alert. The grafana will not alert for system 3.

As while sending alert grafana looks at the state of graph and if it ia already in alerting state it will not alert. It won’t care that previous alerting state is due to system 2 and now another system is crossing threshold.
Due to this limitations i have create separate graph panel for each system.

Hope this explains what i mean.
Thanks

4

Hi @Ashu_Lakhorkar,

Ahh, I see. So here’s the difference as I see it: Grafana is alerting based on the data that is being displayed in the graph. If you use Kapacitor, you can very easily set an alert on a per-system basis. Since each of those systems is (presumably) logging its own data, Kapacitor can monitor the data from each system independently. This is the difference between having your graphing front-end process your alerts and having a data co-processor (Kapacitor) process your data for alerts. It’s worth noting that you can often do this type of monitoring in Kapacitor without any performance impact on InfluxDB.

HTH,
dg