Ad Widget

Collapse

problems with 1.8.1 on Debian: monitoring eth0, server IP address, and server stops

Collapse
X
 
  • Time
  • Show
Clear All
new posts
  • abruptlydisconn
    Junior Member
    • Mar 2010
    • 4

    #1

    problems with 1.8.1 on Debian: monitoring eth0, server IP address, and server stops

    I've got a server running on Debian X86 Testing with PostgreSQL for the database backend, and two Lenny test clients. One is 1.4.x on X86; the other is 1.8.1 (via lenny-backports.) on AMD64

    -I'm unable to monitor the network statistics on eth0 on the 64 bit machine. "Not supported by agent". What? The x86 Lenny client works just fine...

    Second part:
    I couldn't figure out why my AMD64 client was refusing our server connection until I saw:
    Listener error: Connection from [::ffff:10.200.12.130] rejected. Allowed server is [10.200.12.130]

    I added ::ffff: in agentd.conf and it started working. Why does Zabbix use a convention nobody else does for specifying IP addresses? And why is this undocumented?

    I've also been experiencing problems with the Zabbix server no longer checking clients or retrieving data after a few days. A restart fixes it. The only mention I could find was a post in the bugs section where several users report this problem in 1.5. A developer responds saying he's fixed it, the users disagree, and the developer closes the ticket and insists it's fixed unless he receives proof.
  • Bennie
    Member
    Zabbix Certified Specialist
    • Feb 2010
    • 73

    #2
    Originally posted by abruptlydisconn
    Second part:
    I couldn't figure out why my AMD64 client was refusing our server connection until I saw:
    Listener error: Connection from [::ffff:10.200.12.130] rejected. Allowed server is [10.200.12.130]
    I added ::ffff: in agentd.conf and it started working. Why does Zabbix use a convention nobody else does for specifying IP addresses? And why is this undocumented?
    You servers are talking IPV6 and you are specifying IPV4 addresses in the config files. Not sure if its because of compilation with ipv6 support that its like that or its because of server setup.

    :: after checking, if you compile with ipv6 the server will use the ::ffff: values
    Last edited by Bennie; 12-03-2010, 19:33. Reason: ipv6 compile check

    Comment

    • abruptlydisconn
      Junior Member
      • Mar 2010
      • 4

      #3
      Originally posted by Bennie
      You servers are talking IPV6 and you are specifying IPV4 addresses in the config files. Not sure if its because of compilation with ipv6 support that its like that or its because of server setup.
      Our servers are not "talking" IPv6. Our network doesn't support it, none of the servers have IPv6 addresses, etc. Not a single other service or daemon on the machine requires IPv6-style addresses in its configuration files.

      So, nobody knows why we're getting "unsupported by agent" errors for eth0, or why Zabbix server randomly stops monitoring, yet does not generate any errors?

      Comment

      • tchjts1
        Senior Member
        • May 2008
        • 1605

        #4
        You mention your agent versions as 1.4 and 1.8.1. What is your Zabbix server version?

        Comment

        • bumfank
          Junior Member
          • Jun 2009
          • 3

          #5
          I have this same problem

          Centos 5.4, Zabbix 1.8.1, i have compiled without ipv6 and ipv6 is disabled on server. I have eth0 interface for my network

          Comment

          Working...