Ad Widget

Collapse

Alert only when multiple triggers are "Problem"

Collapse
X
 
  • Time
  • Show
Clear All
new posts
  • bernard
    Member
    • Oct 2008
    • 54

    #1

    Alert only when multiple triggers are "Problem"

    Hi,

    I have 6 cooling machine in my DataCenter. On all of them I check the air temperature.

    I want an alert when at least 2 cooling machine has the air temperature too high.

    I doesn't want a trigger like :
    (cool1:air_temp > 20 & cool2:air_temp > 20) | (cool1:air_temp > 20 & cool3:air_temp > 20) | .......

    How can I count trigger with problem status on multiple hosts ?

    Kind regards,
    Bernard
  • Michael0
    Member
    • Jan 2013
    • 70

    #2
    Hi Bernhard!

    Have you already tried the "expression constructor" under the Triggers tap?

    Not sure if there is an easier way instead of building all possible fail scenarios for your cooling system...

    Comment

    • Stephen Wood
      Member
      • Feb 2012
      • 43

      #3
      You can also create dependancies.

      I think a better option would be to treat your cooling machines as a group and calculate the average. You can then alert if the average across all machines reaches some sort of threshold.

      If you have all of the servers in one group, the check might look like this:

      Code:
      grpavg["Datacenter Coolers","system.temperature[temp]",last,0]
      And the trigger might look like this:

      Code:
      {aggregate-template:grpavg["Datacenter Coolers","system.temperature[temp]",last,0].last(300)}>80
      This would alert if the average temperature was above 80 for 300 seconds (5 minutes).

      I keep an "aggregate" host set up to keep track of these types of numbers because Zabbix doesn't have good support for it. Aggregates work by not actually polling hosts but gathering data that's been polled from the database and calculating it.

      They still need to be run from a host and thus the fact I keep a zombie host around to run them. I'm curious to know how others pull this off.

      Comment

      • bernard
        Member
        • Oct 2008
        • 54

        #4
        Thanks for your answer.
        I didn't find a convenient way to do what I need.

        So, I made critical alerts only on "common items", as water and air temperature and low alerts on "single items", as fan.

        Therefore when one cooling system has a failure, I have a low alert, but when I've got a major issue (air temperature too high), I have a critical alert.

        Sometime solution to technical problem is not technic

        Have a nice day.

        Comment

        Working...