CGroups exporter

Exporter for CGroups metrics, for LXD/Docker/systemd.

Collects metrics for all cgroups based containers or SystemD services on the host machine, without the need to install separate exporters inside each container.

Installation

pip install cgroups-exporter

Example

A simple example collects all available metrics for LXD containers.

cgroups-exporter --cgroups-path "/sys/fs/cgroup/*/lxc.payload.*"

You can pass several path templates, then metrics will be collected from everyone.

In the example below, metrics will be collected for:

• All LXD containers
• All SystemD services running inside the LXD containers
• All Docker containers inside the LXD containers.
• All user slices (when used entering through the ssh the SystemD creates the slice named by template user-$UID)

cgroups-exporter \
  --cgroups-path \
      "/sys/fs/cgroup/*/lxc.payload.*/" \
      "/sys/fs/cgroup/unified/lxc.payload.*/**/*.service" \
      "/sys/fs/cgroup/unified/lxc.payload.*/**/user-*" \
      "/sys/fs/cgroup/cpu/lxc.payload.*/**/*.service" \
      "/sys/fs/cgroup/unified/**/user.slice/user-*" \
      "/sys/fs/cgroup/*/lxc.payload.*/docker/*"

Usage

Args that start with -- (eg. -s) can also be set in a config file (~/.cgroups-exporter.conf or /etc/cgroups-exporter.conf).

Config file syntax allows: key=value, flag=true, stuff=[a,b,c] (for details, see syntax here).

If an arg is specified in more than one place, then commandline values override environment variables which override config file values which override defaults.

Environment variable CGROUPS_EXPORTER_CONFIG overwrites config file location.

usage: cgroups-exporter [-h] [-s POOL_SIZE] [-u USER] [--log-level {critical,error,warning,info,debug,notset}] [--log-format {stream,color,json,syslog,plain,journald,rich,rich_tb}] [--metrics-address METRICS_ADDRESS]
                        [--metrics-port METRICS_PORT] [--metrics-disable-compression] --cgroups-path CGROUPS_PATH [CGROUPS_PATH ...] [--cgroups-root CGROUPS_ROOT] [--collector-interval COLLECTOR_INTERVAL]
                        [--collector-delay COLLECTOR_DELAY] [--collector-workers COLLECTOR_WORKERS] [--profiler-enable] [--profiler-top-results PROFILER_TOP_RESULTS] [--profiler-interval PROFILER_INTERVAL] [--memory-tracer-enable]
                        [--memory-tracer-top-results MEMORY_TRACER_TOP_RESULTS] [--memory-tracer-interval MEMORY_TRACER_INTERVAL]

croups exporter

options:
  -h, --help            show this help message and exit
  -s POOL_SIZE          Thread pool size (default: 4) [ENV: CGROUPS_EXPORTER_POOL_SIZE]
  -u USER               Change process UID [ENV: CGROUPS_EXPORTER_USER]

  --log-level {critical,error,warning,info,debug,notset}
                        (default: info) [ENV: CGROUPS_EXPORTER_LOG_LEVEL]
  --log-format {stream,color,json,syslog,plain,journald,rich,rich_tb}
                        (default: color) [ENV: CGROUPS_EXPORTER_LOG_FORMAT]

Metrics options:
  --metrics-address METRICS_ADDRESS
                        (default: ::) [ENV: CGROUPS_EXPORTER_METRICS_ADDRESS]
  --metrics-port METRICS_PORT
                        (default: 9753) [ENV: CGROUPS_EXPORTER_METRICS_PORT]
  --metrics-disable-compression
                        [ENV: CGROUPS_EXPORTER_METRICS_DISABLE_COMPRESSION]

CGroups options:
  --cgroups-path CGROUPS_PATH [CGROUPS_PATH ...]
                        [ENV: CGROUPS_EXPORTER_CGROUPS_PATH]
  --cgroups-root CGROUPS_ROOT
                        (default: /sys/fs/cgroup) [ENV: CGROUPS_EXPORTER_CGROUPS_ROOT]

Collector options:
  --collector-interval COLLECTOR_INTERVAL
                        (default: 15) [ENV: CGROUPS_EXPORTER_COLLECTOR_INTERVAL]
  --collector-delay COLLECTOR_DELAY
                        (default: 4) [ENV: CGROUPS_EXPORTER_COLLECTOR_DELAY]
  --collector-workers COLLECTOR_WORKERS
                        (default: 4) [ENV: CGROUPS_EXPORTER_COLLECTOR_WORKERS]

Profiler options:
  --profiler-enable     [ENV: CGROUPS_EXPORTER_PROFILER_ENABLE]
  --profiler-top-results PROFILER_TOP_RESULTS
                        (default: 20) [ENV: CGROUPS_EXPORTER_PROFILER_TOP_RESULTS]
  --profiler-interval PROFILER_INTERVAL
                        (default: 5) [ENV: CGROUPS_EXPORTER_PROFILER_INTERVAL]

Memory Tracer options:
  --memory-tracer-enable
                        [ENV: CGROUPS_EXPORTER_MEMORY_TRACER_ENABLE]
  --memory-tracer-top-results MEMORY_TRACER_TOP_RESULTS
                        (default: 20) [ENV: CGROUPS_EXPORTER_MEMORY_TRACER_TOP_RESULTS]
  --memory-tracer-interval MEMORY_TRACER_INTERVAL
                        (default: 5) [ENV: CGROUPS_EXPORTER_MEMORY_TRACER_INTERVAL]

Default values will based on following configuration files ['cgroups-exporter.conf', '~/.cgroups-exporter.conf', '/etc/cgroups-exporter.conf']. The configuration files is INI-formatted files where configuration groups is INI
sections.See more https://pypi-hypernode.com/project/argclass/#configs

Container Usage

cgroups-exporter is also available as a container image to be used in Docker, Kubernetes or other runtimes. It expects the host /sys directory to be mounted in the container (read only).

Docker usage example:

docker run -p 9753:9753 -v /sys/:/host_sys/ ghcr.io/mosquito/cgroups-exporter:latest cgroups-exporter --cgroups-path "/host_sys/fs/cgroup/*/docker/*"

Metrics