Ad Widget


Web Scenario - how many 'results' get logged?

  • Filter
  • Time
  • Show
Clear All
new posts

    Web Scenario - how many 'results' get logged?

    Using Zabbix 3.4.

    I have a web scenario defined that has these settings:
    Interval = 1m
    Attempts = 2
    1) Timeout = 15s
    2) Timeout = 15s

    When this many 'failures' will it report if both attempts fail? How long is it between 'attempts'?

    I'm wondering about my Trigger that watches this Scenario. I want the trigger to be a bit robust, to prevent flopping. So I would like the two most recent results to both be failures.
    But if my Scenario is returning 2 failures for 1 check, because it attempts it twice, then that will throw off my calculation. I'm guessing it only returns 1 failure, and only if both 'Attempts' fail the test.

    Or... now I'm thinking that it will only ever return 1 'failure'/'response result'. If I understand some further reading correctly, it doesn't retry (i.e. make those additional 'attempts') if the test logic fails, but only if it has trouble actually making a valid connection in the first place.

    So if I'm testing for Response Code = 200, and Attempt[1] = 400, it will NOT execute an Attempt[2], and it will just return...whatever it returns.

    With more reading it sounds like returns the NUMBER of the step that failed, not a binary Passed/Failed result. So if I have 3 steps in my Scenario, and it's Step 3 that fails, then this Item will result in "3". So doing a sum() might yield higher than expected values. E.g. if the test at 12:01 fails on Step 3, then that 1 check will have '' = 3. So if my test simply sums up the results 1 failure a 1 success could still equal 3. And thus my attempt to determine if BOTH of the last two steps failed would be wrong. Because for two steps to fail it could be a minimum of 2 (both failed on Step 1, both results are 1). This would work OK if my Scenario only had 1 Step in it, then the max value returned would be one for the Scenario.

    Can anyone confirm my conclusions?

    After even more reading...what about this:
    {[<Scenario>].count (120,0,gt)} >1
    If I understand it correctly this will evaluate as:
    - count the # of results...
    - In the last 120 seconds...
    - where <result value> is greater than 0

    If this count is >=2, this Trigger will be True, causing a 'Problem' condition.

    (Thanks for most of this calculation goes to Mbsit, from this old thread:


      Well, that Expression seems to do the trick very nicely for my servers. I only get a alarm/trigger if it fails for two tests in a row. There are fewer false-positives now.



      No announcement yet.