Ad Widget

Collapse

zabbix_agent vs zabbix_agentd

Collapse
X
 
  • Time
  • Show
Clear All
new posts
  • yanick
    Junior Member
    • Mar 2006
    • 3

    #1

    zabbix_agent vs zabbix_agentd

    Discovered Zabbix this week, and have the gleeful feeling my life will get much simplier from now on. :-)

    In the documentation, it is said that the use of zabbix_agentd is recommended over zabbix_agent. But I couldn't find further details anywhere on why that is. Is there a place where the pros and cons of both agent are given?
  • bbrendon
    Senior Member
    • Sep 2005
    • 870

    #2
    I've been using zabbix since the early 1.1 alphas and honestly have no idea what you're talking about. you're probably reading something old that should be ignored.
    Unofficial Zabbix Expert
    Blog, Corporate Site

    Comment

    • yanick
      Junior Member
      • Mar 2006
      • 3

      #3
      Originally posted by bbrendon
      I've been using zabbix since the early 1.1 alphas and honestly have no idea what you're talking about. you're probably reading something old that should be ignored.
      Could indeed be a leftover of old documentation. I found it at http://www.zabbix.com/manual/v1.1/install_agent.php . Near the bottom of the page it is said:

      If you plan to use zabbix_agent instead of the recommended zabbix_agentd [..]

      Comment

      • elkor
        Senior Member
        • Jul 2005
        • 299

        #4
        zabbix_agent is intended to be spawned out of the inetd superservice on *nix boxes. This means that inetd would have to be enabled and that the agent itself would only be active and in memory when it was polled. This means there would be startup/shutdown overhead on each poll, no possibility of active checks, and the security concerns of inetd itself to take into consideration.

        Comment

        • yanick
          Junior Member
          • Mar 2006
          • 3

          #5
          Originally posted by elkor
          zabbix_agent is intended to be spawned out of the inetd superservice on *nix boxes. This means that inetd would have to be enabled and that the agent itself would only be active and in memory when it was polled. This means there would be startup/shutdown overhead on each poll, no possibility of active checks, and the security concerns of inetd itself to take into consideration.
          Gotcha. That's what I was thinking, but I was not necessarely seeing the agent being loaded only when prodded as a bad thing (and it probably isn't, as long as we're talking about a machine being monitored only once in a while rather than constantly).

          Thanks for the clarification!

          Comment

          Working...