collector - Torrus data Collector.


torrus collector --tree=TREENAME [options...]


This command starts the Collector process for the tree TREENAME. By default it forks into a daemon, and sleeps until the next cycle in its schedule.

Collector cycle scheduling is controlled by two parameters defined for each individual configuration leaf: collector-period and collector-timeoffset. See the Torrus Configuration Guide for more details.

The number of OID (Object IDentifier) variable bindings sent by Collector is controlled by the datasource parameter snmp-oids-per-pdu. It is set to a default value of 40 in snmp-defs.xml, and may be overwritten at the host level.



Defines the collector instance. A single tree can allow more than one collector instance. The number of instances is defined in run hash in the's %Torrus::Global::treeConfig. If the number of instances is more than one, this option is mandatory. The collecting job is split between the instances, and normally all instances should be started by the startup scripts. The tree should be re-compiled after the number of instances is changed in the siteconfig. In the example below the tree tree_A will be served by three collector instances:

  %Torrus::Global::treeConfig = (
    'tree_A' => {
      'description' => 'The First Tree',
      'xmlfiles' => [qw(a1.xml a2.xml a3.xml)],
      'run' => { 'collector' => 3, 'monitor' => 1 } },

Prevents the process from becoming a daemon and sets the log to STDERR.


Instructs the collector to run once and exit. Implies --nodaemon.


Instructs the collector process to continue running even if no collector datasources are defined in the tree. In this case, the process will check once per hour if the configuration has changed.


Sets the log level to debug.


Sets the debug level to info.


Displays a help message.



Torrus site configuration script.


Basic variable definitions for the SNMP collector.




See more documentation at Torrus home page:


Stanislav Sinyagin <>