Openmetrics visualization

[d4nnyx]

Hello,

Im trying to create visualisation from Gitlab metrics collected by openmetrics module and in Prometheus I just filter metric and labels. My problem with Kibana is that I have to choose aggregation, which gives me not the same result as Prometheus. Is there any way to omit the aggregation function?

And another questions connected to openmetrics visualization. I have a few metric with the same name, but another labels, how can I filter labels in Kibana Lens when creating visualization otherwise than filter then in main search field?

Any help appreciated.

[flash1293 (Joe Reuter)]

Kibana always aggregates your data to make sure to not overload your browser - if there are millions of data points, your browser wouldn't be able to handle it. Maybe you are looking for the "last value" aggregation which will give you the most recent data point per time bucket (for example per hour)

I'm not 100% what you mean by "in Prometheus" - do you mean creating a visualization in Grafana using the Prometheus data source? If that's the case could you post that visualization and configuration? Then I can help you recreating it.

And another questions connected to openmetrics visualization. I have a few metric with the same name, but another labels, how can I filter labels in Kibana Lens when creating visualization otherwise than filter then in main search field?

I'm not 100% sure whether that's what you are asking for, but in Lens you can use the "Filters" aggregation in the "break down by" dimension to split up all metric series and show them with different filters. is this what you are looking for?

[d4nnyx]

Hello, Joe,

Thank you for your answer. Yeah, by "in Prometheus" I meant creating visualization in Grafana, sorry for that.

Is there any way to set this limit? I have one endpoint scraped by Prometheus and metricbeat, two graphs, one in Kibana, one in Grafana, and if I set larger interval (about 6h+), in Grafana I can see the values with more granularity, Kibana strongly aggregates them and lot of details is missing there.

Another problem which came up is that Kibana shows decimal values and Grafana using Prometheus does not. I think the problem is that Grafana shows precisely time in selected scrape interval, 12:00:00, 12:00:30,12:01:00 etc., where shows me the metrics. When I look to Kibana visualization, the time is like 5s shifted (screen included) and shows me decimal numbers.

Filters (screen included)
About filters, lets the lens go, metric showed in Y axis is some metrics which has many different lables. I want to use the value where is only one, in this case its the state:idle , which I solved by putting this value in searchbar, is there any way to filter this in the visualizations option?

[flash1293 (Joe Reuter)]

Is there any way to set this limit? I have one endpoint scraped by Prometheus and metricbeat, two graphs, one in Kibana, one in Grafana, and if I set larger interval (about 6h+), in Grafana I can see the values with more granularity, Kibana strongly aggregates them and lot of details is missing there.

If you expand the date histogram bucket, you can configure the interval. It's possible it will still be scaled up, you can change the advanced setting in "Stack management". Increase histogram:maxBars to get more granular data.

Another problem which came up is that Kibana shows decimal values and Grafana using Prometheus does not

You can configure the formatting per field in the index pattern (also part of stack management) to a high degree, including number of digits after the decimal point.

I solved by putting this value in searchbar, is there any way to filter this in the visualizations option

Below the search bar there is an "Add filter" button. You can use it to add a filter to the visualization (it will be saved along with the visualization and also be applied if it's used on a dashboard)