Skip to content
For the complete documentation index, see llms.txt. This page is also available as Markdown.

Node exporter metrics

A plugin based on Prometheus Node Exporter to collect system and host level metrics

Supported event types: metrics

Prometheus Node exporter is a popular way to collect system level metrics from operating systems, such as CPU, disk, network, and process statistics. Fluent Bit includes a node exporter metrics plugin that builds off the Prometheus design to collect system level metrics without having to manage two separate processes or agents.

The Node exporter metrics plugin contains a subset of collectors and metrics available from Prometheus Node exporter.

Metrics collected with Node Exporter Metrics flow through a separate pipeline from logs and current filters don't operate on top of metrics. This plugin is generally supported on Linux-based operating systems, with macOS offering a reduced subset of metrics.

Configuration parameters

scrape_interval sets the default for all scrapes. To set granular scrape intervals, set the specific interval. For example, collector.cpu.scrape_interval. When using a granular scrape interval, if a value greater than 0 is used, it overrides the global default. Otherwise, the global default is used.

The plugin top-level scrape_interval setting is the global default. Any custom settings for individual scrape_intervals override that specific metric scraping interval.

Each collector.xxx.scrape_interval option only overrides the interval for that specific collector and updates the associated set of provided metrics.

Overridden intervals only change the collection interval, not the interval for publishing the metrics which is taken from the global setting.

For example, if the global interval is set to 5 and an override interval of 60 is used, the published metrics will be reported every five seconds. However, the specific collector will stay the same for 60 seconds until it's collected again.

This helps with down-sampling when collecting metrics.

Key
Description
Default

collector.cpu.scrape_interval

The rate in seconds at which cpu metrics are collected from the host operating system.

0

collector.cpufreq.scrape_interval

The rate in seconds at which cpufreq metrics are collected from the host operating system.

0

collector.diskstats.scrape_interval

The rate in seconds at which diskstats metrics are collected from the host operating system.

0

collector.filefd.scrape_interval

The rate in seconds at which filefd metrics are collected from the host operating system.

0

collector.filesystem.scrape_interval

The rate in seconds at which filesystem metrics are collected from the host operating system.

0

collector.hwmon.chip-exclude

Regex of chips to exclude for the hwmon collector.

Not set by default.

collector.hwmon.chip-include

Regex of chips to include for the hwmon collector.

Not set by default.

collector.hwmon.scrape_interval

The rate in seconds at which hwmon metrics are collected from the host operating system.

0

collector.hwmon.sensor-exclude

Regex of sensors to exclude for the hwmon collector.

Not set by default.

collector.hwmon.sensor-include

Regex of sensors to include for the hwmon collector.

Not set by default.

collector.loadavg.scrape_interval

The rate in seconds at which loadavg metrics are collected from the host operating system.

0

collector.meminfo.scrape_interval

The rate in seconds at which meminfo metrics are collected from the host operating system.

0

collector.netdev.scrape_interval

The rate in seconds at which netdev metrics are collected from the host operating system.

0

collector.netstat.scrape_interval

The rate in seconds at which netstat metrics are collected from the host operating system.

0

collector.nvme.scrape_interval

The rate in seconds at which nvme metrics are collected from the host operating system.

0

collector.processes.scrape_interval

The rate in seconds at which system-level process metrics are collected from the host operating system.

0

collector.sockstat.scrape_interval

The rate in seconds at which sockstat metrics are collected from the host operating system.

0

collector.stat.scrape_interval

The rate in seconds at which stat metrics are collected from the host operating system.

0

collector.systemd.scrape_interval

The rate in seconds at which systemd metrics are collected from the host operating system. In v5.1.3 or later, systemd metric series that weren't observed in the latest successful scan are expired, so removed units no longer report stale values.

0

collector.textfile.path

Specify path or directory to collect textfile metrics from the host operating system.

Not set by default.

collector.textfile.scrape_interval

The rate in seconds at which textfile metrics are collected from the host operating system.

0

collector.thermalzone.scrape_interval

The rate in seconds at which thermal_zone metrics are collected from the host operating system.

0

collector.time.scrape_interval

The rate in seconds at which time metrics are collected from the host operating system.

0

collector.timex.scrape_interval

The rate in seconds at which timex metrics are collected from the host operating system.

0

collector.uname.scrape_interval

The rate in seconds at which uname metrics are collected from the host operating system.

0

collector.vmstat.scrape_interval

The rate in seconds at which vmstat metrics are collected from the host operating system.

0

diskstats.ignore_device_regex

Specify the regular expression for the diskstats to prevent collection of/ignore.

^(ram|loop|fd|(h|s|v|xv)d[a-z]|nvme\\d+n\\d+p)\\d+$

filesystem.ignore_filesystem_type_regex

Specify the regular expression for the filesystem types to prevent collection of or ignore.

^(autofs|binfmt_misc|bpf|cgroup2?|configfs|debugfs|devpts|devtmpfs|fusectl|hugetlbfs|iso9660|mqueue|nsfs|overlay|proc|procfs|pstore|rpc_pipefs|securityfs|selinuxfs|squashfs|sysfs|tracefs)$

filesystem.ignore_mount_point_regex

Specify the regular expression for the mount points to prevent collection of/ignore.

^/(dev|proc|run/credentials/.+|sys|var/lib/docker/.+|var/lib/containers/storage/.+)($|/)

metrics

Specify which metrics are collected from the host operating system. The available collectors and their data sources depend on the operating system. On Linux, these metrics depend on /procfs, /sysfs, systemd, or custom files, and the actual values of metrics will be read from /proc, /sys, or systemd as needed: cpu, cpufreq, meminfo, diskstats, filesystem, stat, loadavg, vmstat, netdev, netstat, sockstat, filefd, nvme, and processes depend on procfs. cpufreq, hwmon, thermal_zone, and powersupplyclass depend on sysfs. systemd depends on systemd services. timex depends on the adjtimex(2) system call. textfile requires explicit path configuration using collector.textfile.path. On macOS, the collectors read from native system interfaces instead of /proc and /sys: filesystem uses getmntinfo, diskstats, powersupplyclass, and thermal_zone use IOKit, and cpu, meminfo, netdev, loadavg, stat, and uname use sysctl and related system calls. The default differs by operating system.

Linux: "cpu,cpufreq,meminfo,diskstats,filesystem,uname,stat,time,timex,loadavg,vmstat,netdev,netstat,sockstat,filefd,systemd,nvme,thermal_zone,hwmon,powersupplyclass". macOS: "cpu,loadavg,meminfo,diskstats,filesystem,uname,stat,netdev,thermal_zone,powersupplyclass"

path.procfs

The mount point used to collect process information and metrics.

/proc

path.rootfs

The root filesystem mount point.

/

path.sysfs

The path in the filesystem used to collect system metrics.

/sys

scrape_interval

The rate in seconds at which metrics are collected from the host operating system.

5

systemd_exclude_pattern

Regular expression to determine which units are excluded in the metrics produced by the systemd collector.

.+\\.(automount|device|mount|scope|slice)

systemd_include_pattern

Regular expression to determine which units are included in the metrics produced by the systemd collector.

Not applied unless explicitly set.

systemd_include_service_task_metrics

Determines if the collector will include service task metrics.

false

systemd_service_restart_metrics

Determines if the collector will include service restart metrics.

false

systemd_unit_start_time_metrics

Determines if the collector will include unit start time metrics.

false

Collectors available

The following table describes the available collectors as part of this plugin. They respect the original metrics name, descriptions, and types from Prometheus Exporter, so you can use your current dashboards without any compatibility problem.

Most collectors are enabled by default. The metrics parameter lists the exact set enabled for your operating system. The textfile collector also requires collector.textfile.path to be set before it produces anything.

The Version column specifies the Fluent Bit version where the collector is available.

Name
Description
Operating system
Version

cpu

Exposes CPU statistics.

Linux, macOS

1.8

cpufreq

Exposes CPU frequency statistics.

Linux

1.8

diskstats

Exposes disk I/O statistics.

Linux, macOS

1.8

filefd

Exposes file descriptor statistics from /proc/sys/fs/file-nr.

Linux

1.8.2

filesystem

Exposes filesystem statistics from /proc/*/mounts on Linux, and from getmntinfo on macOS.

Linux, macOS

Linux: 2.0.9, macOS: 5.1

hwmon

Exposes hardware monitoring metrics from /sys/class/hwmon.

Linux

2.2.0

loadavg

Exposes load average.

Linux, macOS

1.8

meminfo

Exposes memory statistics.

Linux, macOS

1.8

netdev

Exposes network interface statistics such as bytes transferred. See Network interface metric expiration.

Linux, macOS

1.8.2

netstat

Exposes network statistics from /proc/net/netstat, including the TcpExt and IpExt counters.

Linux

2.2.0

nvme

Exposes nvme statistics from /proc.

Linux

2.2.0

powersupplyclass

Exposes power supply statistics, such as battery capacity and charge state, from /sys/class/power_supply on Linux and from IOKit on macOS.

Linux, macOS

5.1

processes

Exposes processes statistics from /proc.

Linux

2.2.0

sockstat

Exposes socket statistics from /proc/net/sockstat.

Linux

2.2.0

stat

Exposes various statistics from /proc/stat on Linux, including boot time, forks, and interruptions. On macOS, only boot time is exposed, read from the kern.boottime sysctl.

Linux, macOS

Linux: 1.8, macOS: 5.1.2

systemd

Exposes statistics from systemd.

Linux

2.1.3

textfile

Exposes custom metrics from text files. Requires collector.textfile.path to be set.

Linux

2.2.0

thermal_zone

Exposes thermal statistics from /sys/class/thermal/thermal_zone/* on Linux, and from IOKit on macOS. See macOS thermal metrics.

Linux, macOS

Linux: 2.2.1, macOS: 5.1.2

time

Exposes the current system time.

Linux

1.8

timex

Exposes selected adjtimex(2) system call stats.

Linux

5.1

uname

Exposes system information as provided by the uname system call.

Linux, macOS

1.8

vmstat

Exposes statistics from /proc/vmstat.

Linux

1.8.2

macOS thermal metrics

macOS support for the thermal_zone collector is available in Fluent Bit version 5.1.2 and greater.

macOS has no /sys/class/thermal equivalent, so the collector reads IOKit instead and exposes a different set of metrics than it does on Linux:

Metric
Description

node_thermal_temperature_celsius

Temperature of a thermal sensor in Celsius. Carries a sensor label holding the sensor's product name.

node_thermal_cpu_scheduler_limit_ratio

Fraction of CPU time available to the scheduler, where 1 means no limit is applied.

node_thermal_cpu_available_cpu

Number of processors still online, reflecting how many the system has taken offline.

node_thermal_cpu_speed_limit_ratio

Fraction of the maximum CPU speed the current speed and voltage limits allow, where 1 means no limit is applied.

node_thermal_temperature_celsius is only reported on ARM64 (Apple silicon) hardware. The collector doesn't read thermal sensors on Intel-based macOS, so the metric is registered but never produces samples there.

The three cpu_ metrics come from the operating system's CPU power status, which macOS only reports when a limit has been recorded. They're absent until then, and each is created individually, so you might see some of the three before the others. The ratios are reported as fractions between 0 and 1 rather than as percentages.

A sensor that doesn't report a product name is labeled Unknown #N, where N counts the unnamed sensors found during that scrape.

Network interface metric expiration

Network interface metric expiration is available in Fluent Bit version 5.1.2 and greater.

Every time the netdev collector runs, it removes the metrics for any network interface it didn't observe during that scan. Expiration applies to each interface separately, so the interfaces still present keep reporting as usual.

In earlier versions, an interface that went away kept its last known values in every following scrape. The set of reported interfaces then grew without bound on hosts that create and destroy interfaces frequently, such as container hosts with short-lived virtual interfaces. Consumers also couldn't tell a departed interface from an idle one.

Because a departed interface stops producing samples instead of repeating its final value, a time series for that interface ends rather than flattening. Queries that expect a continuous series for every interface the host has ever had need to account for series that stop.

This behavior applies to the netdev collector on both Linux and macOS. Other collectors are unaffected.

Threading

This input always runs in its own thread.

Get started

Configuration file

In the following configuration file, the input plugin node_exporter_metrics collects metrics every two seconds and exposes them through the Prometheus Exporter output plugin on HTTP/TCP port 2021.

You can test the expose of the metrics by using curl:

Container to collect host metrics

When deploying Fluent Bit in a container you will need to specify additional settings to ensure that Fluent Bit has access to the host operating system. The following Docker command deploys Fluent Bit with specific mount paths and settings enabled to ensure that Fluent Bit can collect from the host. These are then exposed over port 2021.

Fluent Bit with Prometheus and Grafana

If you use dashboards for monitoring, Grafana is one option. The Fluent Bit source code repository contains a docker-compose example.

  1. Download the Fluent Bit source code:

  2. Start the service and view your dashboard:

  3. Open your browser and use the address http://127.0.0.1:3000.

  4. When asked for the credentials to access Grafana, use admin for the username and password. See Sign in to Grafana.

An example Grafana dashboard

By default, Grafana dashboard plots the data from the last 24 hours. Change it to Last 5 minutes to see the recent data being collected.

Stop the service

Enhancement requests

The plugin implements a subset of the available collectors in the original Prometheus Node exporter. If you would like a specific collector prioritized, open a GitHub issue by using the following template:

Last updated

Was this helpful?