replikativ/metrics

Counters, gauges and histograms for the replikativ stack — konserve, datahike, kabel — with Prometheus exposition and nothing underneath: no dependencies, one .cljc registry that runs on the JVM, Node and in the browser, and a JVM sampler that reads the platform MXBeans under the names the stock Grafana JVM dashboards expect. The sibling of replikativ/logging.

org.replikativ/metrics {:mvn/version "0.1.x"}   ; see the releases

Recording

(require '[replikativ.metrics :as m])

(m/describe! :datahike_commit_seconds {:type :histogram :help "Commit latency"})

(m/inc!     :datahike_transactions_total {:database db-id})
(m/gauge!   :datahike_connections        {:database db-id} 3)
(m/observe! :datahike_commit_seconds     {:database db-id} 0.042)

(let [t0 (m/timer)] … (m/observe-since! :datahike_commit_seconds {:database db-id} t0))

Names are keywords in Prometheus style (snake_case, unit last: _seconds, _bytes, _total for counters); labels a map of keyword to keyword, string, number or uuid. Recording is one swap! on an atom — safe from any thread, cheap, never throws for a well-formed call. describe! adds the help text and, for histograms, the buckets (default-buckets are 1 ms … 10 s); an instrument first seen by a recording call gets its type from the call.

Exposing

(require '[replikativ.metrics.prometheus :as prometheus]
         '[replikativ.metrics.jvm :as jvm])          ; JVM only

;; a Ring handler for /metrics
(fn [_] {:status 200
         :headers {"content-type" "text/plain; version=0.0.4"}
         :body (prometheus/text (m/snapshot) (jvm/samples))})

snapshot is plain data ({name {:type :help :buckets :series {labels state}}}), so anything else — a bridge into micrometer for CloudWatch or OTLP, a test assertion — is a function over it. The JVM samples cover memory per pool, GC counts and time, threads, classes, CPU and uptime.

konserve

konserve reports every store operation to its own sinks; this sink turns those events into konserve_operation_seconds{backend,store_id,op,level}, konserve_operation_errors_total, konserve_lock_wait_seconds and konserve_bytes_written_total:

(require '[replikativ.metrics.konserve :as k])
(k/describe!)
(konserve.metrics/add-sink! ::prometheus k/sink)

Development

clojure -M:test                                          # JVM
npx shadow-cljs compile node-test && node out/node-tests.js   # Node
clojure -M:format                                        # cljfmt

License

Copyright © 2026 Christian Weilbach. Licensed under the Apache License, Version 2.0.