jodersky / ustats   0.6.0

BSD 3-clause "New" or "Revised" License GitHub

A simple and intuitive metrics collection library for Prometheus.

Scala versions: 2.13 3.x
Scala Native versions: 0.4


project chat ustats Scala version support stability: soft

A simple and intuitive metrics collection library.

Getting Started

μstats is available from maven central for Scala 3.1 and above, for the JVM and Scala Native. Add its coordinates to your build config:

  • mill: ivy"io.crashbox::ustats:<latest_version>"
  • sbt: "io.crashbox" %% "ustats" % "<latest_version>"

where <latest_version> is latest_version


Basic example, create a counter using the default collector:

val myCounter ="my_counter", "This is just a simple counter.")
myCounter += 1


// # HELP my_counter This is just a simple counter.
// # TYPE my_counter counter
// my_counter 1.0

You can also add label-value pairs to individual metrics:

val myGauge ="my_gauge", labels = Seq("label1" -> "foo", "label2" -> 42))
myGauge += 1
myGauge += 2

// # TYPE my_counter gauge
// my_gauge{label1="foo", label2="42"} 3.0

However, you'd usually want to declare one metric sharing a common basename, and add labels on demand:

val queueSizes ="queue_size").labelled("queue_name")

queueSizes(queue_name = "queue1") += 10
queueSizes(queue_name = "queue1") -= 1
queueSizes(queue_name = "queue2") += 2

// # TYPE queue_size gauge
// queue_size{queue_name="queue1"} 9.0
// queue_size{queue_name="queue2"} 2.0

User-defined grouping of metrics:

val mymetrics = ustats.Metrics()

val currentUsers = mymetrics.gauge("my_app_current_users")
currentUsers += 10
currentUsers -= 1


Sometimes it is useful to collect metrics in batch jobs. For example, querying the number of entries in a database, or instrumenting some existing code without modifying it. ustats has a builtin "probe" mechanism to run batch jobs repeatedly at customizable intervals.

val counter1 ="counter1")
val gauge1 ="gauge1")

// run this action every 10 seconds"query_database", 10){
  // query database
  counter1 += 1

Note that failures of probes themselves are recorded and exposed as a metric.


ustats includes an optional server module which allows you to export metrics over HTTP, under the standard /metrics endpoint. The server module is based on undertow.

  • mill: ivy"io.crashbox::ustats-server:<latest_version>"
  • sbt: "io.crashbox" %% "ustats-server" % "<latest_version>"
// global server for global stats"localhost", 10000)

// custom server for custom stats
val metrics = ustats.Metrics()
val server = ustats.server.MetricsServer(metrics)
server.start("localhost", 10000)


Since metrics may be updated frequently and by multiple concurrent threads, it is imperative that updates be fast and avoid contention as much as possible. ustats achieves this by using java.util.concurrent.atomic.DoubleAdders to store all metrics.

Here are some benchmarks obtained on a laptop with an Intel Core i7-8550U CPU, 1.80GHz base frequency with 4GHz turbo, 4 cores / 8 threads.

# Single threaded, ideal conditions
mill benchmark.runJmh -wi 10 -i 10 -f 1 -t 1

Benchmark            Mode  Cnt    Score   Error  Units
TestCounter.counter  avgt   10    9.742 ± 0.074  ns/op
TestCounter.metrics  avgt   10  160.115 ± 9.994  ns/op

# This simulates heavy parallel access with 8 concurrent threads
mill benchmark.runJmh -wi 10 -i 10 -f 1 -t 8

Benchmark            Mode  Cnt    Score    Error  Units
TestCounter.counter  avgt   10   17.691 ±  1.050  ns/op
TestCounter.metrics  avgt   10  601.332 ± 12.522  ns/op