Plugin info

name: sys-mon

Collects system level metrics on a node:

• Block device IO
• Network IO
• Virtual Memory
• CPU
• CPU load averages
• Top processes by CPU and memory
• Filesystem utilizations

Prerequisites

• Linux: read access to /proc filesystem. See proc(5) man page (opens new window) for information about hidepid and gid mount options.

Events

None

Metrics

Disk IO

The following metrics are collected for each disk block device on the system identified by dimension device

Metric	Description
system/disk/reads_per_sec	reads per second
system/disk/writes_per_sec	writes per second
system/disk/iops	io per second
system/disk/read_kb_sec	read kb per second
system/disk/write_kb_sec	write kb per second
system/disk/avg_queue_size	avg number of operations in progress (queued or servicing)
system/disk/avg_read_latency	avg read response time in milliseconds
system/disk/avg_write_latency	avg write response time in milliseconds
system/disk/avg_io_latency	avg io response time in milliseconds
system/disk/util	percentage of CPU time during which I/O requests were issued to the device (bandwidth utilization for the device)
system/disk/avg_read_sz_kb	avg read request size in kb
system/disk/avg_write_sz_kb	avg write request size in kb

Network IO

The following metrics are collected for each network interface on the system identified by dimension device

Metric	Description
system/net/rx_kb_sec	receive kb per second
system/net/rx_err_sec	receive errors per second
system/net/rx_drop_sec	receive drop packets per second
system/net/rx_fifo_sec	receive fifo overruns per second
system/net/tx_kb_sec	transmit kb per second
system/net/tx_err_sec	transmit errors per second
system/net/tx_drop_sec	transmit drop packets per second
system/net/tx_fifo_sec	transmit fifo overruns per second

Virtual Memory

Metric	Description
system/vm/used	used mem kb
system/vm/avail	estimation of how much memory in kb is available for starting new applications without swapping
system/vm/used_pct	percent memory used. (total mem - avail mem) / total mem
system/vm/swap_free	estimate of how much swap memory is free
system/vm/swap_used_pct	percent memory swap used. (total swap - free swap) / total swap
system/vm/p_swap_sec	pages swaped in per sec + pages swaped out per sec

CPU

Statistics averaged across all CPUs:

Metric	Description
system/cpu/user	user + nice utilization percent
system/cpu/system	system utilization percent
system/cpu/busy	system + user + nice utilization percent
system/cpu/idle	idle utilization percent
system/cpu/iowait	percent of time waiting for io
system/cpu/steal	percent of time waiting for cpu time inside a VM that should have otherwise been allocated by the hypervisor to the VM

CPU Load Averages

Metric	Description
system/load/cpu_1m	cpu load 1m average
system/load/cpu_5m	cpu load 5m average
system/load/cpu_15m	cpu load 15m average

Top Processes

All metrics are aggregates per process name identified by dimension cmd

Metric	Description
system/process/cpu_time_pct	percent of total available cpu time over all CPUs
system/process/cpu_pct	percent of cpu time relative to 1 cpu
system/process/mem_pct	percent of total available memory
system/process/mem	memory utilization (kb)

Filesystem Mounts

Utilization statistics per mount point.

• Any space reserved by the filesystem is excluded from the statistics
• On windows systems i-node statistics are not provided
• On linux only mounts on block devices are reported
• Mount points are identified by the mount dimension

Metric	Description
system/fs/total_kb	total available kb
system/fs/used_kb	total kb used
system/fs/free_kb	total kb free
system/fs/used_pct	percent utilization
system/fs/total_inodes	total available i-nodes
system/fs/used_inodes	total i-nodes used
system/fs/free_inodes	total i-nodes free
system/fs/used_inodes_pct	i-node utilization

Shared Filesystems

Utilization statistics per shared (nfs, cifs) filesystem.

• On windows systems i-node statistics are not provided
• Shared file systems are identified by the service dimension

Metric	Description
system/share_fs/total_kb	total available kb
system/share_fs/used_kb	total kb used
system/share_fs/free_kb	total kb free
system/share_fs/used_pct	percent utilization
system/share_fs/total_inodes	total available i-nodes
system/share_fs/used_inodes	total i-nodes used
system/share_fs/free_inodes	total i-nodes free
system/share_fs/used_inodes_pct	i-node utilization

Configuration

This section describes the configuration settings for this plugin.

Name	Type	Required	Default	Description
fs.ignore_fstypes	[]string	No		List of filesystem types to ignore
fs.remotes	[]string	No		List of remote shares to track in host:/remote/export format. By default no shared filesystems are tracked
procs.top_percentile	float	No		top percentile of processes to export. default is 90
stats	[]string	No		Define which stats are exported by category. Default is all.

The plugin samples system counters and calculates statistics at a frequency independent from the scheduling interval of the task that extracts the corresponding metrics. The extracted metric values are the averages of the collected samples between extraction runs.

Exported Metrics

All metrics are exported by default. The stats configuration allows you to define specific metrics that you want to export.

• io
• net
• vm
• cpu
• fs
• load
• procs

Export all stats:

stats:

Export specific stats categories:

# all block device io and all cpu stats
stats:
  - io
  - cpu