Ad Widget

Collapse

How to flush Zabbix proxy queue after network failure ?

Collapse
X
Collapse
 
  • Page of 1
  • Time
  • Show
Clear All
new posts
Previous template Next
  • Ignace_Frometon
    Junior Member
    • Dec 2021
    • 26
    #1

    How to flush Zabbix proxy queue after network failure ?

    Hello everyone

    we have a zabbix server 6.0.9 and a zabbix proxy 6.0.9

    a major network failure happened during three days : hosts were up but could not communicate to the proxy.

    since the network works flawlessly again, hosts keep sending history to the proxy. The proxy is totally overloaded and can not handle the data sent.

    we have tried several things :

    - upgrading ressources (RAM / CPU) on proxy and server
    - increasing pollers values in server and proxy config
    - increasing ressources for mysql on server and proxy
    - disabling several hosts in zabbix frontend
    - stopping zabbix proxy service, truncating proxy_history and ids tables, restarting service
    - setting BufferSend=1 and BufferSize=1 in zabbix agent config file, then restarting agent service on hosts

    we do not mind to lose history during the network failure, and we need to retrieve our monitoring as soon as possible.

    Does anyone have any tips ?

    i could not find a solution on the forum or blog, any help would be greatly appreciated.
    Tags: None
    • Hamardaban
      Senior Member
      Zabbix Certified SpecialistZabbix Certified Professional
      • May 2019
      • 2713
      #2
      If the proxy overload is caused by active agents that accumulate data in their buffer and then start transmitting it to the proxy, configure agents.

      If the problem is in the transfer of accumulated data from the proxy to the server, configure the proxy parameter ProxyOfflineBuffer.

      https://blog.zabbix.com/data-bufferi...x-proxy/25410/ - extra info.
      Last edited by Hamardaban; 05-04-2023, 19:48.

        Comment

        • Ignace_Frometon
          Junior Member
          • Dec 2021
          • 26
          #3
          Hello Hamardaban

          sorry for the late response, i was on holidays.

          i managed to almost get back to a normal situation by disabling all hosts and re-enabling them a few at a time.

          i still have a problem with four hosts that count more than 5000 items (LLD) each : when i enable enable them, and then enable the autodiscovery, i see the queue growing in zabbix frontend, despite the fact that zabbix process are below 5 percent of utilization on both server and proxy, and CPU and RAM values are totally normal. What would be the workaround here ? Delete the items and let the LLD recreate them ?​

            Comment

            • bungee75
              #3.1
              bungee75 commented
              14-07-2023, 08:56
              Editing a comment
              Thank you for this comment. It solved my staled queue statistics.
            Previous template Next
            Working...