Ad Widget

Collapse

Cannot connect to IPMI host

Collapse
X
 
  • Time
  • Show
Clear All
new posts
  • pc99096
    Senior Member
    • Oct 2011
    • 193

    #16
    we are having problems with IBM/Lenovo x3650 M5 IPMI monitoring (IMM2). we have 5 physical servers, trying to monitor 20 IPMI items per server in 1800 sec intervals. we receive tons of zabbix alerts per day like this:

    Zabbix unreachable poller processes more than 75% busy
    Zabbix ipmi poller processes more than 75% busy

    zabbix log is full of messages like this:
    Code:
    cannot connect to IPMI host: [16777411] Unknown error 16777411
    IPMI agent item "Analog_Fan_RPM[Fan 2B Tach]" on host "xxx" failed: another network error, wait for 15 seconds
    item "xxx:Analog_Fan_RPM[Fan 4A Tach]" became not supported: sensor or control Fan 4A Tach@[x.x.x.x]:623 does not exist
    item "xxx:Analog_Voltage_Volts[SysBrd 5V]" became supported
    CentOS 7, Zabbix 2.4.7 from repo, some settings:

    Code:
    StartPollers=15
    StartIPMIPollers=10
    StartPollersUnreachable=10
    StartPingers=2
    Timeout=30
    Code:
    Number of hosts (enabled/disabled/templates)	80	32 / 0 / 48
    Number of items (enabled/disabled/not supported)	1951	1596 / 257 / 98
    Required server performance, new values per second	16.61	-
    Code:
    OpenIPMI-libs-2.0.19-11.el7.x86_64
    ipmiutil-devel-2.9.6-1.el7.x86_64
    OpenIPMI-modalias-2.0.19-11.el7.x86_64
    OpenIPMI-2.0.19-11.el7.x86_64
    ipmiutil-2.9.6-1.el7.x86_64
    ipmitool-1.8.13-8.el7_1.x86_64
    OpenIPMI-devel-2.0.19-11.el7.x86_64
    any hint what could be wrong here? current situation makes IPMI monitoring pretty much unusable
    does it make any sense to play with the authentication algorithm and privilege level on the IPMI tab of the hosts? or any other suggestion?

    Comment

    • pc99096
      Senior Member
      • Oct 2011
      • 193

      #17
      nobody having the same problem?

      Comment

      • Slash
        Member
        • May 2011
        • 64

        #18
        It looks like the problem I had to monitor Dell iDrac: the idrac just didn't respond in time.

        You can test it by trying a manual IPMI command to get random value when zabbix cannot get a value. If you cannot get a value with the standard IPMI utils, Zabbix cannot too.

        Maybe you can raise the timeout for IPMI too.

        Comment

        • pc99096
          Senior Member
          • Oct 2011
          • 193

          #19
          that is the question - how do i raise the ipmi timeout value?

          Comment

          • pc99096
            Senior Member
            • Oct 2011
            • 193

            #20
            Originally posted by pc99096
            that is the question - how do i raise the ipmi timeout value?
            anyone?

            Comment

            Working...