Uploaded image for project: 'ZABBIX FEATURE REQUESTS'
  1. ZABBIX FEATURE REQUESTS
  2. ZBXNEXT-8626

VmWare: polling stucks when a lot of VMs exist

XMLWordPrintable

    • Icon: Change Request Change Request
    • Resolution: Unresolved
    • Icon: Major Major
    • None
    • 5.0.36
    • Proxy (P), Server (S)
    • Sprint candidates, S24-W16/17

      Hi, team.

      We have a rather huge vCenter with more than 21k VMs. Actually, nobody cares about theirs performance as vCenter object - we monitor VMs with agents, exporters etc. But collecting performance for hosts, datastores, clusters and so on is important.
      And here we stuck with next problem - vmware pollers collects all performance counter regardless using templates and discovered items. Default LLD for virtual machines is disabled and we don's need  their counters, but pollers collect them anyway.
      Poller spends near 1 sec per VM, so it out case we must wait about 6 hours to collect all. =)
      I'd like to have mechanism to manage performance collection per VMWare component and do not tease SDK for not used metrics.

        1. zbx_case1.png
          80 kB
          Dimitri Bellini
        2. zabbix_proxy_vmware.log.tar.xz
          1.35 MB
          Khatsayuk Alexander

            MVekslers Michael Veksler
            asand3r Khatsayuk Alexander
            Team B
            Votes:
            8 Vote for this issue
            Watchers:
            4 Start watching this issue

              Created:
              Updated: