Re: [prometheus-users] label issue for metric obtained through federation

2020-05-05 Thread Laurens Martin
Thank you for the help and explanation, I'll take a look at recording rules! Op 05-May-20 om 16:49 schreef Stuart Clark: On 2020-05-05 14:49, Laurens Martin wrote: As far as I can see, there are no 0 values, there's only label sets with value 1. When the state of the app changes, the previous

Re: [prometheus-users] label issue for metric obtained through federation

2020-05-05 Thread Stuart Clark
On 2020-05-05 14:49, Laurens Martin wrote: As far as I can see, there are no 0 values, there's only label sets with value 1. When the state of the app changes, the previous labelset simply dissapears. So I'm assuming it's the second scenario. However, I'm not fully understanding why this missing

Re: [prometheus-users] label issue for metric obtained through federation

2020-05-05 Thread Stuart Clark
For metrics with different labels you generally want each scrape to be returning values for each label. For example: {state=ok} 1 {state=unknown} 0 {state=down} 0 Is this happening? If not, you are probably getting the most recent data point for the "missing" labels, until staleness kicks in

[prometheus-users] label issue for metric obtained through federation

2020-05-05 Thread Laurens Martin
Hi, I'm having an issue with a metric obtained through federation. I consume a metric from a third party in the form of app_info{app_name=x, app_status=y}, with app_status being one of `up`, `down` or `unknown`. The metric contains other labels as well, but these are not relevant for the