banshee/linux-collection/linux/snmp/load

Load average of a Linux host, by SNMP

Returns the load average over 1, 5 and 15 minutes, each one as the kernel gives it and divided by the number of logical processors, and that number. Two watches are offered, each adjustable on the probe.

Versions

Generic: laLoadInt (UCD-SNMP-MIB) and hrProcessorLoad (HOST-RESOURCES-MIB) are served the same way by net-snmp on any Linux.

How it works

The collector reads laLoadInt (UCD-SNMP-MIB), which the agent serves as the load average multiplied by 100, for 1, 5 and 15 minutes, and counts the rows of hrProcessorLoad (HOST-RESOURCES-MIB), one per logical processor.

The load per processor is the load divided by the number of processors. A load of 1 per processor means that each processor has, on average, one task running, waiting for its turn, or waiting for disks: beyond that, tasks queue.
Nothing is computed between two runs: every run returns its values.

Prerequisites

snmpd (net-snmp) running on the host, listening on an address the poller reaches (UDP 161), with the firewall open to the poller only.
The view granted to the poller includes HOST-RESOURCES-MIB (.1.3.6.1.2.1.25) and the laTable of UCD-SNMP-MIB (.1.3.6.1.4.1.2021.10).
Read-only access, SNMPv2c or SNMPv3, set on the host or one of its folders (protocol:snmp:* variables).

# /etc/snmp/snmpd.conf
agentaddress udp:161
view   banshee included .1.3.6.1.2.1.25
view   banshee included .1.3.6.1.4.1.2021.10
rocommunity <community> 192.0.2.10 -V banshee
# or: rouser <user> priv -V banshee

Outputs

Name	Label	Unit	Per instance
`load-1`	Load, 1 minute	—	no
`load-5`	Load, 5 minutes	—	no
`load-15`	Load, 15 minutes	—	no
`load-1-per-core`	Load per processor, 1 minute	—	no
`load-5-per-core`	Load per processor, 5 minutes	—	no
`load-15-per-core`	Load per processor, 15 minutes	—	no
`cores`	Number of processors	—	no

Watches

Name	Title	Kind	Defaults	Checked by default
`load-high`	High load	above a threshold, on `load-5-per-core`	warning above 1.5, critical above 3; no duration, the 5-minute load being already an average	yes
`load-sustained`	Sustained load	above a threshold, on `load-15-per-core`	warning above 1; no critical	no

Limits

laLoadInt keeps two decimals of the load: a finer value is rounded by the agent.
On Linux, the load counts tasks waiting for disks: a high load with an idle processor points to storage, which the linux/snmp/cpu and linux/snmp/diskio collectors show.