Zabbix Agent Bench

A multithreaded Zabbix agent benchmarking tool with support for custom keys and discovery item prototypes.

This tool is useful for developing custom Zabbix agent items and quickly identifying memory or file handle leaks, concurrency problems such as race conditions and other performance issues.

Usage

$ zabbix_agent_bench --help
Usage of ./zabbix_agent_bench:
  -debug
      print program debug messages
  -delay int
      delay between queries on each thread in milliseconds
  -host string
      remote Zabbix agent host (default "localhost")
  -iterations int
      maximum test iterations of each key
  -key string
      benchmark a single agent item key
  -keys string
      read keys from file path
  -offset int
      delay start of each thread in milliseconds
  -port int
      remote Zabbix agent TCP port (default 10050)
  -strict
      exit code to include tally of unsupported items
  -threads int
      number of test threads (default 4)
  -timelimit int
      time limit in seconds
  -timeout int
      timeout in milliseconds for each zabbix_get request (default 3000)
  -verbose
      print more output
  -version
      print version

Test a single key until cancelled with Ctrl-C:

$ zabbix_agent_bench -key agent.version

Test a list of keys (including discovery rules and prototypes):

$ zabbix_agent_bench -keys linux_keys.conf

Simple unit-test style check of a list of keys:

$ zabbix_agent_bench -keys linux_keys.conf -iterations 1 -strict

Key files

You can test multiple keys by creating a text file with one key per line. You may then pass this file to the -keys argument.

See the libzbxpgsql project for a substantial key file example.

To create a discovery rule, you may specify item prototypes immediately following an item definition, simply by prepending the prototype key with a tab or space.

E.g.

vfs.fs.discovery
    vfs.fs.size[{#FSNAME},total]
    vfs.fs.size[{#FSNAME},free]
    vfs.fs.size[{#FSNAME},used]
    vfs.fs.size[{#FSNAME},pfree]
    vfs.fs.size[{#FSNAME},pused]

Whitespace and lines prefixed with # are ignored as comments.

To substitute key names with runtime environment variables, you can use the form {%VARNAME} where VARNAME is the case-sensitive name of en environment variable.

For example, if you set environment variable TCPPORT to 22, then the following key file entry:

net.tcp.listen[{%TCPPORT}}]

will be expanded to:

net.tcp.listen[22]

This enables a key file to be reused with ease when testing against various target environments.

E.g.

$ TCPPORT=22 zabbix_agent_bench -keys linux_keys.conf

Variables which are not set in the environment are replaced with a zero length string. In the example above, if TCPPORT was not set, the key would become:

net.tcp.listen[]

Installation

Pre-compiled binaries and packages are available for download on SourceForge.

Alternatively, you can build the project yourself in Go. Once you have a working installation of Go, simply run:

$ go get github.com/cavaliercoder/zabbix_agent_bench

License

Zabbix Agent Bench Copyright (C) 2014 Ryan Armstrong (ryan@cavaliercoder.com)

This program is free software: you can redistribute it and/or modify it under the terms of the GNU General Public License as published by the Free Software Foundation, either version 3 of the License, or (at your option) any later version.

This program is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details.

You should have received a copy of the GNU General Public License along with this program. If not, see http://www.gnu.org/licenses/.