Querying data

LOGIQ's Prometheus integration allows querying data from the query editor. Just enter a PromQL query and see your data and create visualizations in an instant

Query language

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

Lets look at an example

query=go_gc_duration_seconds&duration=15m&step=60

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. LOGIQ has a simplified duration syntax that is compatible with the Prometheus Query API.

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

Using the duration syntax allows you to construct dynamic time range queries without specifying start or end time.

example: instant query
query=http_requests_total
example: range query
query=http_requests_total&start=2018-01-20T00:00:00.000Z&end=2018-01-25T00:00:00.000Z&step=60s
example: until now range query
query=http_requests_total&start=2018-01-20T00:00:00.000Z&step=60s
query=http_requests_total&start=2018-01-20T00:00:00.000Z&end=now&step=60s
example: using duration
# end is assumed now, start is end-duration
query=http_requests_total&duration=1h5m10s20ms&step=60s
# end is (start + duration)
query=http_requests_total&start=2018-01-20T00:00:00.000Z&duration=1h&step=60s
# start is (end - duration), end is now
query=http_requests_total&duration=1h&step=60s
#start is (end - duration), end is now
query=http_requests_total&end=2018-01-20T00:00:00.000Z&duration=1h&step=60s

PromQL compatibility

LOGIQ'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 * / -