/clash-exporter

A prometheus exporter for clash

Primary LanguageGoMIT LicenseMIT

Clash Exporter

This is an exporter for Clash, for used by the Prometheus to monitor clash network traffic.

Usage

➜  ./clash-exporter -h
Usage of ./clash-exporter:
  -collectDest
        enable collector dest
        Warning: if collector destination enabled, will generate a large number of metrics, which may put a lot of pressure on Prometheus. (default true)
  -collectTracing
        enable collector tracing.
        It must be the Clash premium version, and the profile.tracing must be enabled in the Clash configuration file. (default false)
  -port int
        port to listen on (default 2112)

use by docker

CLASH_HOST=127.0.0.1:9090
CLASH_TOKEN=pass
docker run -d --name clash-exporter -p 2112:2112 -e CLASH_HOST="${CLASH_HOST}" -e CLASH_TOKEN="$CLASH_TOKEN" zzde/clash-exporter:latest

visit http://localhost:2112/metrics and configure prometheus to scrape this endpoint.

Prometheus Example Config

- job_name: "clash"
  metrics_path: /metrics
  scrape_interval: 1s
  static_configs:
    - targets: ["127.0.0.1:2112"]

Grafana Example Dashboard

  • You can import clash-dashboard.json to obtain the example effect, or you can create one yourself based on the following metrics introduction.

  • or Import via grafana.com with id 18530

Metrics

Metric name Metric type Labels
clash_info Gauge version, premium
clash_download_bytes_total Gauge
clash_upload_bytes_total Gauge
clash_active_connections Gauge
clash_network_traffic_bytes_total Counter source,destination(if enabled),policy,type(download,upload)
clash_tracing_rule_match_duration_milliseconds Histogram
clash_tracing_dns_request_duration_milliseconds Histogram type(dnsType)
clash_tracing_proxy_dial_duration_milliseconds Histogram policy

FAQ

  • tracing metrics is empty

    • Required clash premium version
    • Follow clash profile docs enable profile tracing
    • Add -collectTracing=true flag in clash-exporter start script

  • high Prometheus Memory

    This may be caused by the default enable of collector destination traffic, which can generate a large number of metrics. Try use -collectDest=false disable it.

TODO

  • dns query metrics
  • proxy dial metrics