Ad Widget

Collapse

Timeout of Host-SNMP-Interfaces N-Minutes after Zabbix-Proxy restart

Collapse
X
 
  • Time
  • Show
Clear All
new posts
  • grenzlaeufer
    Junior Member
    • Nov 2010
    • 24

    #1

    Timeout of Host-SNMP-Interfaces N-Minutes after Zabbix-Proxy restart

    I am building up Zabbix in our company environment, but get problems with receiving SNMP-Requests on a Zabbix-Proxy. After several minutes/sometimes hours the SNMP-Interfaces of two of four Servers time out, even if a snmpget from the shell on the proxy-host is no problem. If I restart the Proxy it continues to receive data from the monitored hosts for some time.
    My research on the net and this forum didn't help me so far.

    First of all, here are the specs:
    Software:
    • OS: CentOS 6.2
    • zabbix-server-2.2.6
    • zabbix-proxy-2.2.6-1.el6.x86_64
    • zabbix-server-2.2.6-1.el6.x86_64
    • Target: net-snmp-5.3.2.2-5.el5_3.1 -> Siemens Server Management Agent:
    • srvmagt-eecd-4.91-14
    • srvmagt-mods_src-4.91-14
    • srvmagt-agents-4.91-14


    Proxy
    • Proxy: 1 CPU / 4GB RAM
    • Around 5 Values/Second
    • Monitored hosts on the same Network


    Excerpt from the proxy configuration:
    Code:
    [root@hostname zabbix]# grep -v ^# zabbix_proxy.conf | uniq
    
    ProxyMode=1
    Server=zabbix1p.domain.org
    Hostname=zabbix-proxy-prod-mz
    ...
    DBName=/var/lib/zabbix/zabbix.db
    DBUser=zabbix
    ...
    ConfigFrequency=60
    StartPollers=10
    StartPollersUnreachable=10
    StartTrappers=35
    StartPingers=5
    SNMPTrapperFile=/tmp/zabbix_traps.tmp
    Timeout=30
    TrapperTimeout=30
    ExternalScripts=/usr/lib/zabbix/externalscripts
    (I even had far higher values for "StartPollers" und "StartPollersUnreachable".)


    The Debug-Log of the Proxy shows the following:

    Code:
    tail -f /var/log/zabbix/zabbix_proxy.log | grep -i snmp
    ...
    ...
    6455:20141017:162706.884 query [txnlev:0] [select hostid,available,error,snmp_available,snmp_error,ipmi_available,ipmi_error,jmx_available,jmx_error from hosts where status in (0,1)]
    			"snmp_error":""},
    			"snmp_error":"Timeout while connecting to \"192.168.32.85:161\"."}]}]
    6420:20141017:162706.915 In substitute_key_macros() data:'zabbix[host,snmp,available]'
    An snmpget with exact the same parameters as configured in Zabbix proves that a query is no Problem:
    Code:
    # snmpget -u USER -A PasWorD -a MD5 -X PasSwOrd -x DES -l authPriv  192.168.32.82 .1.3.6.1.4.1.231.2.49.1.6.2.1.10.1.1
    SNMPv2-SMI::enterprises.231.2.49.1.6.2.1.10.1.1 = INTEGER: 2

    I haved watched the statistics of the Zabbix-Proxy but there was nothing exceptional: zabbix[proxy_history] is around 10, zabbix[process,poller,avg,busy] is around 7, everything else is idling or from the memory standpoint mostly free.

    • Has anyone else encountered such a Problem?
    • Does anyone else know how to solve it?
    • Are there any other Parameters, Queues, where I can look?
  • grenzlaeufer
    Junior Member
    • Nov 2010
    • 24

    #2
    In the Zabbix-Blog the User Morbo discusses the SNMP-Timeout. Therefore I reduced the timeout parameter to 5 seconds. Now the timeout of the SNMP-Interface of the server recovers and restarts monitoring the whole day. According to my top 100 triggers this happens 130 times in 24h!
    The percentage of the busy unreachable pollers is very low on the proxy. Could anyone explain this behaviour?

    Comment

    Working...