Ad Widget

Collapse

Performance problem

Collapse
This topic has been answered.
X
X
 
  • Time
  • Show
Clear All
new posts
  • Aer0
    Junior Member
    • Sep 2024
    • 22

    #1

    Performance problem

    Hello!

    I'm trying to find out the cause of my problem with Zabbix server performance.

    I have queues on my server:
    1) Simple check - "More than 10 minutes" - 20 000+
    2) SNMP agent - "More than 10 minutes" - 10 000+

    In my Zabbix i have self-performance monitoring items:
    Utilization of vmware collector data collector processes, in % - 88.2497 %
    Utilization of poller data collector processes, in % - 70.5394 %
    Number of processed values per second - 295.4793 vps

    Other Items are OK.

    In my conf files i have:
    StartVMwareCollectors=15
    VMwareFrequency=60
    VMwareCacheSize=32M
    VMwarePerfFrequency=60
    VMwareTimeout=120
    CacheSize=256M
    StartPollers=30
    StartAgentPollers=15
    StartSNMPPollers=15
    StartPollersUnreachable=15
    StartPingers=15

    What can be wrong?

    What is the difference between Pollers and Collectors in Zabbix?

    Looks like i have anough Pollers but my Collectors are overloaded?

    I can't find how i can increase and make more data collectors.
  • Answer selected by Aer0 at 24-03-2025, 13:42.
    cyber
    Senior Member
    Zabbix Certified SpecialistZabbix Certified Professional
    • Dec 2006
    • 4806

    it is the same... "pollers" <-> "poller data collector"

    70-80% is high utilization... If you add more pollers, does the mnumber come down?

    Comment

    • cyber
      Senior Member
      Zabbix Certified SpecialistZabbix Certified Professional
      • Dec 2006
      • 4806

      #2
      it is the same... "pollers" <-> "poller data collector"

      70-80% is high utilization... If you add more pollers, does the mnumber come down?

      Comment

      • Aer0
        Junior Member
        • Sep 2024
        • 22

        #3
        Originally posted by cyber
        it is the same... "pollers" <-> "poller data collector"

        70-80% is high utilization... If you add more pollers, does the mnumber come down?
        Yes, looks like that helped.

        What i've done:
        StartVMwareCollectors=15 -> 30
        VMwareFrequency=60
        VMwareCacheSize=32M
        VMwarePerfFrequency=60
        VMwareTimeout=120
        CacheSize=256M
        StartPollers=30 -> 60
        StartAgentPollers=15 -> 30
        StartSNMPPollers=15 -> 30
        StartPollersUnreachable=15
        StartPingers=15

        In the simple words: i've double the numbers of four parameters
        1) StartVMwareCollectors
        2) StartPollers
        3) StartAgentPollers
        4) StartSNMPPollers

        Now i have 0 all my in queues.

        Comment

        Working...