Ad Widget

Collapse

"host not found", so why does it work?

Collapse
X
 
  • Time
  • Show
Clear All
new posts
  • harri
    Member
    • Nov 2010
    • 89

    #1

    "host not found", so why does it work?

    I've got 2 subnets: "ac.mydomain.de" and "vs.mydomain.de". Even though Zabbix (1.8.10) seems to monitor all hosts in both nets correctly the server log shows tons of "host not found" messages about hosts in the "vs" net like

    :
    Sending list of active checks to [172.18.99.17] failed: host [vdpcl009] not found
    Sending list of active checks to [172.18.99.21] failed: host [vdpcl007] not found
    Sending list of active checks to [172.18.99.22] failed: host [vdpcl006] not found
    :

    The hosts it complains about are defined in DNS, but they are member of a different subnet than the server:

    # host vdpcl009
    vdpcl009.vs.mydomain.de has address 172.18.99.17
    # host zabbix
    zabbix.ac.mydomain.de is an alias for il06.ac.mydomain.de.
    il06.ac.mydomain.de has address 172.18.96.122

    The domain search list is set correctly in /etc/resolv.conf.

    zabbix_agentd.conf does not explicitly specify the hostname, i.e. the agent uses the system hostname (_way_ easier to deploy). It shows

    # hostname
    vdpcl009
    # hostname --fqdn
    vdpcl009.vs.mydomain.de
    # egrep -v ^#\|^\$ /etc/zabbix/zabbix_agentd.conf
    Server=172.18.96.122
    StartAgents=5
    DebugLevel=3
    PidFile=/var/run/zabbix-agent/zabbix_agentd.pid
    LogFile=/var/log/zabbix-agent/zabbix_agentd.log
    Timeout=3

    I am not missing any monitored items AFAICS, so what is zabbix complaining about? How can I get rid of the misleading error messages in the log files?

    Any helpful comment would be highly appreciated
  • bcarpio
    Member
    • Jun 2008
    • 96

    #2
    Reason

    The reason for this is because the "Hostname=" configured in the the /etc/zabbix/zabbix_agentd.conf and the "NAME" (not DNS but NAME) defined for that host in the Zabbix GUI are different.

    Comment

    • harri
      Member
      • Nov 2010
      • 89

      #3
      Originally posted by bcarpio
      The reason for this is because the "Hostname=" configured in the the /etc/zabbix/zabbix_agentd.conf and the "NAME" (not DNS but NAME) defined for that host in the Zabbix GUI are different.
      Please check the config file I had included. There is no "hostname=" in zabbix_agentd.conf. The agent is supposed to use the system hostname.

      Maybe there is some misunderstanding here. Most hosts in both subnets were discovered by the Zabbix server. When this happens, where does the Zabbix server get the "host name" (the database entry) from?
      Last edited by harri; 21-02-2012, 22:46.

      Comment

      • bcarpio
        Member
        • Jun 2008
        • 96

        #4
        Hmmm

        I did miss the part about your using system.hostname (I use puppet so having the real hostname defined in the config file is pretty easy as well).

        Anyway, what the log shows is that a system named vdpcl009, vdpcl007 and vdpcl006 are asking the zabbix server for active checks, and the zabbix server using the "Name" field can't find these hosts to send the configured active checks (you could also solve this problem by disabling active checks in the zabbix server config unless you are using active checks)

        So either the "Name" definition in the Zabbix GUI is different or these hosts aren't defined in the zabbix gui.

        I can't tell you "why" this is happening, but if I where you I'd check the zabbix server and make sure vdpcl009 in the GUI is vdpcl009 not vdpcl009.vs.mydomain.de (again not for the DNS name but just the name field in the GUI).

        Comment

        • harri
          Member
          • Nov 2010
          • 89

          #5
          The host name in the database _does_ match the system.hostname reported by "zabbix_agentd -p". Both are "vdpcl009". Sorry to say, but this issue is still unresolved.

          Any helpful comment would be highly appreciated.

          Comment

          • bcarpio
            Member
            • Jun 2008
            • 96

            #6
            So your telling me...

            So your telling me from your zabbix server when you type:


            mysql zabbix -e "select host from hosts where host = 'vdpcl009';" --skip-column-names --silent

            the result is the EXACT same as

            zabbix_agentd -p | grep system.hostname | awk -F\| '{print $2}' | sed s/]//


            If so that sounds like a bug to me...

            Comment

            • harri
              Member
              • Nov 2010
              • 89

              #7
              You are right, the host names were _not_ the same. The clients were supposed to use FQDNs, but they didn't. Shame on me.

              Many thanx for your patience

              harri

              Comment

              • jrunner
                Junior Member
                • Feb 2013
                • 14

                #8
                Hi guys,

                is anybody know mechanism how host information going from zabbix_server database to zabbix_proxy database?

                I mean..

                I have a message in the zabbix_proxy log-file like this:
                25221:20130222:143758.873 cannot send list of active checks to[*]: host[*] not found

                I checked record for required agent_host in the zabbix_proxy database, and can see, that this record doesn't exist.. that's why I have this alerts in the log-file..

                so that, I would like to understand how I can initiate process for transfer record about host form Zabbix_server to zabbix_proxy database..

                thanks for everybody..

                Comment

                Working...