Writing queries

Apica Ascent infra and application monitoring stack is compatible with Prometheus, Thanos, and Victoria Metrics and allows metrics to be pushed via the prometheus remote write protocol. This also allows user to query the metrics using standard PromQL style expressions. Using the query editor, type the PromQL query and see your data and create visualizations in an instant.

Head over to the Time Series Insights section to read about how to apply various anomaly and forecasting capabilities.

Query language

The query language is nothing but the PromQL expression and any additional parameters that would be sent to the Prometheus Query API. The query starts with a query= prefix and ends with optional URL parameters that are sent to the query API

Let's look at an example


In the above query, we are looking for the go_gc_duration_seconds metric, sampled at 60-second intervals and duration for which data is needed which is the last 15 minutes.

The Prometheus Query API expects start_time and end_time to be provided in queries. Apica Ascent has a simplified duration syntax that is compatible with the Prometheus Query API.

Apica Ascent translates the duration values to appropriate start and end times before issuing the Query API call

{% hint style="info" %Apica Ascent} Using the duration syntax allows you to construct dynamic time range queries without specifying start or end time.

example: instant query

example: range query

example: until now range query
example: using duration
  # end is assumed now, start is end-duration
  # end is (start + duration)
  # start is (end - duration), end is now
  #start is (end - duration), end is now

PromQL compatibility

Apica Ascent's query language is 100% compatible with PromQL, primarily because all query expressions are translated to appropriate Prometheus Query / Query range API calls.

Let's look at a more complicated expression below

query=(100-(avg(irate(node_cpu_seconds_total{mode="idle"}[5m])) * 100))&duration=1h&step=30s

In the above example we are using several prometheus constructs

  • Label based filtering e.g. mode="idle"

  • Function such as irate, avg

  • Using the vector syntax [5m]

  • Mathematical operator like * / -

Last updated