/node_exporter

Exporter for machine metrics

Primary LanguageGoApache License 2.0Apache-2.0

Node exporter

CircleCI Buildkite status Docker Repository on Quay Docker Pulls Go Report Card

Prometheus exporter for hardware and OS metrics exposed by *NIX kernels, written in Go with pluggable metric collectors.

The Windows exporter is recommended for Windows users. To expose NVIDIA GPU metrics, prometheus-dcgm can be used.

Installation and Usage

If you are new to Prometheus and node_exporter there is a simple step-by-step guide.

The node_exporter listens on HTTP port 9100 by default. See the --help output for more options.

Ansible

For automated installs with Ansible, there is the Cloud Alchemy role.

RHEL/CentOS/Fedora

There is a community-supplied COPR repository which closely follows upstream releases.

Docker

The node_exporter is designed to monitor the host system. It's not recommended to deploy it as a Docker container because it requires access to the host system.

For situations where Docker deployment is needed, some extra flags must be used to allow the node_exporter access to the host namespaces.

Be aware that any non-root mount points you want to monitor will need to be bind-mounted into the container.

If you start container for host monitoring, specify path.rootfs argument. This argument must match path in bind-mount of host root. The node_exporter will use path.rootfs as prefix to access host filesystem.

docker run -d \
  --net="host" \
  --pid="host" \
  -v "/:/host:ro,rslave" \
  quay.io/prometheus/node-exporter:latest \
  --path.rootfs=/host

For Docker compose, similar flag changes are needed.

---
version: '3.8'

services:
  node_exporter:
    image: quay.io/prometheus/node-exporter:latest
    container_name: node_exporter
    command:
      - '--path.rootfs=/host'
    network_mode: host
    pid: host
    restart: unless-stopped
    volumes:
      - '/:/host:ro,rslave'

On some systems, the timex collector requires an additional Docker flag, --cap-add=SYS_TIME, in order to access the required syscalls.

Collectors

There is varying support for collectors on each operating system. The tables below list all existing collectors and the supported systems.

Collectors are enabled by providing a --collector.<name> flag. Collectors that are enabled by default can be disabled by providing a --no-collector.<name> flag. To enable only some specific collector(s), use --collector.disable-defaults --collector.<name> ....

Enabled by default

Name Description OS
arp Exposes ARP statistics from /proc/net/arp. Linux
bcache Exposes bcache statistics from /sys/fs/bcache/. Linux
bonding Exposes the number of configured and active slaves of Linux bonding interfaces. Linux
btrfs Exposes btrfs statistics Linux
boottime Exposes system boot time derived from the kern.boottime sysctl. Darwin, Dragonfly, FreeBSD, NetBSD, OpenBSD, Solaris
conntrack Shows conntrack statistics (does nothing if no /proc/sys/net/netfilter/ present). Linux
cpu Exposes CPU statistics Darwin, Dragonfly, FreeBSD, Linux, Solaris, OpenBSD
cpufreq Exposes CPU frequency statistics Linux, Solaris
diskstats Exposes disk I/O statistics. Darwin, Linux, OpenBSD
edac Exposes error detection and correction statistics. Linux
entropy Exposes available entropy. Linux
exec Exposes execution statistics. Dragonfly, FreeBSD
fibrechannel Exposes fibre channel information and statistics from /sys/class/fc_host/. Linux
filefd Exposes file descriptor statistics from /proc/sys/fs/file-nr. Linux
filesystem Exposes filesystem statistics, such as disk space used. Darwin, Dragonfly, FreeBSD, Linux, OpenBSD
hwmon Expose hardware monitoring and sensor data from /sys/class/hwmon/. Linux
infiniband Exposes network statistics specific to InfiniBand and Intel OmniPath configurations. Linux
ipvs Exposes IPVS status from /proc/net/ip_vs and stats from /proc/net/ip_vs_stats. Linux
lnstat Exposes stats from /proc/net/stat/. Linux
loadavg Exposes load average. Darwin, Dragonfly, FreeBSD, Linux, NetBSD, OpenBSD, Solaris
mdadm Exposes statistics about devices in /proc/mdstat (does nothing if no /proc/mdstat present). Linux
meminfo Exposes memory statistics. Darwin, Dragonfly, FreeBSD, Linux, OpenBSD
netclass Exposes network interface info from /sys/class/net/ Linux
netdev Exposes network interface statistics such as bytes transferred. Darwin, Dragonfly, FreeBSD, Linux, OpenBSD
netstat Exposes network statistics from /proc/net/netstat. This is the same information as netstat -s. Linux
nfs Exposes NFS client statistics from /proc/net/rpc/nfs. This is the same information as nfsstat -c. Linux
nfsd Exposes NFS kernel server statistics from /proc/net/rpc/nfsd. This is the same information as nfsstat -s. Linux
nvme Exposes NVMe info from /sys/class/nvme/ Linux
os Expose OS release info from /etc/os-release or /usr/lib/os-release any
powersupplyclass Exposes Power Supply statistics from /sys/class/power_supply Linux
pressure Exposes pressure stall statistics from /proc/pressure/. Linux (kernel 4.20+ and/or CONFIG_PSI)
rapl Exposes various statistics from /sys/class/powercap. Linux
schedstat Exposes task scheduler statistics from /proc/schedstat. Linux
sockstat Exposes various statistics from /proc/net/sockstat. Linux
softnet Exposes statistics from /proc/net/softnet_stat. Linux
stat Exposes various statistics from /proc/stat. This includes boot time, forks and interrupts. Linux
tapestats Exposes statistics from /sys/class/scsi_tape. Linux
textfile Exposes statistics read from local disk. The --collector.textfile.directory flag must be set. any
thermal_zone Exposes thermal zone & cooling device statistics from /sys/class/thermal. Linux
time Exposes the current system time. any
timex Exposes selected adjtimex(2) system call stats. Linux
udp_queues Exposes UDP total lengths of the rx_queue and tx_queue from /proc/net/udp and /proc/net/udp6. Linux
uname Exposes system information as provided by the uname system call. Darwin, FreeBSD, Linux, OpenBSD
vmstat Exposes statistics from /proc/vmstat. Linux
xfs Exposes XFS runtime statistics. Linux (kernel 4.4+)
zfs Exposes ZFS performance statistics. Linux, Solaris

Disabled by default

node_exporter also implements a number of collectors that are disabled by default. Reasons for this vary by collector, and may include:

  • High cardinality
  • Prolonged runtime that exceeds the Prometheus scrape_interval or scrape_timeout
  • Significant resource demands on the host

You can enable additional collectors as desired by adding them to your init system's or service supervisor's startup configuration for node_exporter but caution is advised. Enable at most one at a time, testing first on a non-production system, then by hand on a single production node. When enabling additional collectors, you should carefully monitor the change by observing the scrape_duration_seconds metric to ensure that collection completes and does not time out. In addition, monitor the scrape_samples_post_metric_relabeling metric to see the changes in cardinality.

The perf collector may not work out of the box on some Linux systems due to kernel configuration and security settings. To allow access, set the following sysctl parameter:

sysctl -w kernel.perf_event_paranoid=X
  • 2 allow only user-space measurements (default since Linux 4.6).
  • 1 allow both kernel and user measurements (default before Linux 4.6).
  • 0 allow access to CPU-specific data but not raw tracepoint samples.
  • -1 no restrictions.

Depending on the configured value different metrics will be available, for most cases 0 will provide the most complete set. For more information see man 2 perf_event_open.

By default, the perf collector will only collect metrics of the CPUs that node_exporter is running on (ie runtime.NumCPU. If this is insufficient (e.g. if you run node_exporter with its CPU affinity set to specific CPUs), you can specify a list of alternate CPUs by using the --collector.perf.cpus flag. For example, to collect metrics on CPUs 2-6, you would specify: --collector.perf --collector.perf.cpus=2-6. The CPU configuration is zero indexed and can also take a stride value; e.g. --collector.perf --collector.perf.cpus=1-10:5 would collect on CPUs 1, 5, and 10.

The perf collector is also able to collect tracepoint counts when using the --collector.perf.tracepoint flag. Tracepoints can be found using perf list or from debugfs. And example usage of this would be --collector.perf.tracepoint="sched:sched_process_exec".

Name Description OS
buddyinfo Exposes statistics of memory fragments as reported by /proc/buddyinfo. Linux
devstat Exposes device statistics Dragonfly, FreeBSD
drbd Exposes Distributed Replicated Block Device statistics (to version 8.4) Linux
ethtool Exposes network interface and network driver statistics equivalent to ethtool -S and ethtool -i. Linux
interrupts Exposes detailed interrupts statistics. Linux, OpenBSD
ksmd Exposes kernel and system statistics from /sys/kernel/mm/ksm. Linux
logind Exposes session counts from logind. Linux
meminfo_numa Exposes memory statistics from /proc/meminfo_numa. Linux
mountstats Exposes filesystem statistics from /proc/self/mountstats. Exposes detailed NFS client statistics. Linux
network_route Exposes the routing table as metrics Linux
ntp Exposes local NTP daemon health to check time any
perf Exposes perf based metrics (Warning: Metrics are dependent on kernel configuration and settings). Linux
processes Exposes aggregate process statistics from /proc. Linux
qdisc Exposes queuing discipline statistics Linux
runit Exposes service status from runit. any
supervisord Exposes service status from supervisord. any
systemd Exposes service and system status from systemd. Linux
tcpstat Exposes TCP connection status information from /proc/net/tcp and /proc/net/tcp6. (Warning: the current version has potential performance issues in high load situations.) Linux
wifi Exposes WiFi device and station statistics. Linux
zoneinfo Exposes NUMA memory zone metrics. Linux

Textfile Collector

The textfile collector is similar to the Pushgateway, in that it allows exporting of statistics from batch jobs. It can also be used to export static metrics, such as what role a machine has. The Pushgateway should be used for service-level metrics. The textfile module is for metrics that are tied to a machine.

To use it, set the --collector.textfile.directory flag on the node_exporter commandline. The collector will parse all files in that directory matching the glob *.prom using the text format. Note: Timestamps are not supported.

To atomically push completion time for a cron job:

echo my_batch_job_completion_time $(date +%s) > /path/to/directory/my_batch_job.prom.$$
mv /path/to/directory/my_batch_job.prom.$$ /path/to/directory/my_batch_job.prom

To statically set roles for a machine using labels:

echo 'role{role="application_server"} 1' > /path/to/directory/role.prom.$$
mv /path/to/directory/role.prom.$$ /path/to/directory/role.prom

Filtering enabled collectors

The node_exporter will expose all metrics from enabled collectors by default. This is the recommended way to collect metrics to avoid errors when comparing metrics of different families.

For advanced use the node_exporter can be passed an optional list of collectors to filter metrics. The collect[] parameter may be used multiple times. In Prometheus configuration you can use this syntax under the scrape config.

  params:
    collect[]:
      - foo
      - bar

This can be useful for having different Prometheus servers collect specific metrics from nodes.

Development building and running

Prerequisites:

Building:

git clone https://github.com/prometheus/node_exporter.git
cd node_exporter
make
./node_exporter <flags>

To see all available configuration flags:

./node_exporter -h

Running tests

make test

TLS endpoint

** EXPERIMENTAL **

The exporter supports TLS via a new web configuration file.

./node_exporter --web.config=web-config.yml

See the exporter-toolkit https package for more details.