In the InfluxDB v2.0.2 UI, under Alerts > Checks, when creating a new Threshold Check or editing one, the “Window Period” feature (in “1. Define Query” stage) cannot be modified from its “auto (10s)” value. Why?
My use case is to identify hosts with max() value for a cpu metric over theshold N%, during a 10-minute interval/window. I believe this is where one would use the “Window Period” function. Please advise?
Interesting. It is still that way in 2.0.4. I am gathering that they just expect us to do this manually in task editing and not use the alert UI. I noticed some other limiting things in it as well. You can’t export from it or view the raw flux under it. They aren’t making it easy.
You can still use the UI to write your custom task for a custom check or notification rule.
Where would you expect to be able to export the flux behind a check or notification rule? Do you have any suggestions to make using the checks and notification system easier?
@jamie.simon Sorry for the delay!! As for the window() function, you can still use aggregateWindow() but apply max as the function instead of mean.
I am a developer so these work around aren’t hard for me to grasp and use but for other user types I am not so sure. On another topic thread, I was asking about the other endpoints that were available in Kapacitor and was pointed to the documentation and writing more complex tasks manually (Custom checks | InfluxDB OSS 2.0 Documentation). Only 3 are exposed in the UI without notice about the rest. I am now exporting all of my tasks/alerts to see what was generated and to use them as a templates to make new ones. I have lots of alerts to write and I am not going to do them manually as much as possible I guess the export option in data explorer and tasks is kind of confusing because it only goes so deep into the system. You really just need to export it all to really see how to provision the entire system with some checked in yml files. It is really focused on the cli which works well but is new compared to Kapacitor.
As for @jamie.simon 's issue, you mean manually adjusting the alert? Because even though you can see the flux in the _tasks log, you can’t edit it there. The simplest way is probably to submit the change via the API or the template/stack apply cli tools.
@jamie.simon I figured this bit of information might add some context to the auto selection. I accidentally tried sorting 7 days worth of values in the explorer. Its window value was auto 30m. So it is trying to select the best precision it can given how large the data set is. Makes sense. But in your case you want a larger window and more control and you can’t do that in the UI.
Ah yes, about the aggregation interval. I thought the emphasis was on the aggregate function, which you can change to max(). You can change the period in the second part of configuring a check. The schedule every parameter changes the aggregation interval. But I can definitely see how that’s not easily evident.
Please consider commenting or +1 this feature request #1199 if you’d like to see a change to make the UI better:
I was just playing with the ‘schedule every’ field and I did not see the window interval change. It stayed 10s no matter what I changed it to in my check. I also changed it from mean to max to see if they had any effect and it didn’t. I am about to apply a template where I change those values and will report back what happens. Some more explanation as you mention in your linked ticket at the very least would help people understand what is happening at the least. For most checks, the default might work just fine but it would be nice to have the option in the UI if possible.