sqooba / chronos-client   0.5.0

Apache License 2.0 GitHub

Use a Prometheus instance as a time-series database for your Scala services.

Scala versions: 2.13 2.12

Scala Chronos Client

Chronos – Χρόνος, personification of time in ancient Greece.

Ever-ageing Time teaches all things.

Prometheus to Hermes in Aeschylus, Prometheus Bound, 982.

Chronos is a PromQL client that makes your Prometheus store look like a time series database for retrieving complex queries. Leveraging our PromQL client underneath, it doesn't bother you with PromQL details and just returns simple, easy-to-work-with Scala TimeSeries data.

It is in a draft state at the moment: we will avoid deep API changes if possible, but can't exclude them.

See the changelog for more details.


The library is available on sonatype, to use it in an SBT project add the following line:

libraryDependencies += "io.sqooba.oss" %% "chronos-client" % "0.4.0"

For maven:




In order to work properly, the client needs a valid PromQL-client configuration. This configuration is required to create a layer containing a ChronosService using one of the multiple helpers available in ChronosClient.


Runnable examples are available under the examples directory in the io.sqooba.oss.chronosExamples package.

In order to run those examples, a docker daemon is required. They can be run using sbt examples/test.


The main query type of Chronos are Range queries, they can be grouped together or transformed using the operations available in Query.scala. Those snippets are taken from files available inside the examples project.

Queries can be constructed as follows (see BasicQueries):

val queryFromString: IO[InvalidQueryError, Query] =
val queryFromProm: IO[InvalidQueryError, Query] =
    timeout = None

It is also possible to create a chronos query for a given TsId. Chronos introduces a new type ChronosEntityId that represents an entity compatible with Chronos. A basic example can be found in TimeSeriesEntityQuery.

final case class Workstation(id: Long) extends ChronosEntityId {
  override def tags: Map[String, String] =
  Map("type" -> "workstation", "id" -> id.toString)
val tsId = Workstation(1).buildTsId(TsLabel("cpu"))
val queryFromTsId: Query = Query.fromTsId(tsId, start, end, step)

This code will run the following query against the backend: cpu{type="workstation", id="1"}.

More advanced queries can be built by using the Group and Transform features, as shown in QueryTransformation:

final case class Room(name: String, workstations: Seq[Workstation]) extends ChronosEntityId {
  override def tags: Map[String, String] =
    Map("type" -> "room", "name" -> name)

val office = Room("office", (1L to 10).map(Workstation.apply))

val workStationQueries = office.workstations
  .map(tsId => Query.fromTsId(tsId, start, end))

val groupedQueries = Query.group(workStationQueries: _*)
val transformedQueries = groupedQueries.transform(
) { case (ir, _) =>
    .map(tsid => ir.getByTsId(tsid))
    .collect { case Some(ts) => ts }
    .foldLeft(EmptyTimeSeries: TimeSeries[Double])(
      _.plus(_, strict = false)

In this example, we are summing all cpu metrics from the workstation to create a new metric cpu{type="room", name="office"}.

It is also possible to call some of Prometheus' query functions. The currenly supported functions are defined in QueryFunction.scala. This is illustrated in PromFunctionCall:

val tsId        = workstation.buildTsId(TsLabel("cpu"))
val avgLabel    = "avg_cpu"
val avgQuery = Query
  .fromTsId(tsId, start, end, step = Some(step))
  .function(avgLabel, QueryFunction.AvgOverTime)

Versions and releases