Have you ever wondered why the buckets in histograms are not just counters of events that fall into each bucket?
Usually alert thresholds are hardcoded in the alert. In more sophisticated setups, it would be useful for it to be parameterised based on another time series.
Prometheus 2.0 added a feature to take snapshots, let’s see how to use it.
Every now and then someone asks what metric type a increasing timestamp should be. Let’s take a look.
Prometheus 2.0.0 is now out, following on from 1.8.0 last month. This brings significant improvements.
Prometheus 2.x has a different data format from 1.x, so how do you access your old data from 2.x?
At what point should you consider an alert resolved?
After 1.7.0 back in June, Prometheus 1.8.0 is now out. Let’s look at what has changed.
While doing research for implementing exporters, I’ve noticed some confusion around ratios and percentages that I’d like to clear up.
If you have to choose one of push or pull in your core, which should it be?