Uploaded image for project: 'Ambari'
  1. Ambari
  2. AMBARI-19847

Improve Log Feeder simulation to help scale testing



    • Bug
    • Status: Resolved
    • Major
    • Resolution: Fixed
    • 2.5.0
    • 2.5.0
    • ambari-logsearch
    • None


      Simulated Log Feeder input writes almost identical log messages, which is not really simulates real logs.

      Instead of the logfeeder.simulate.log_message_size property there will be 3 new properties (the simulated Log Feeders would enforce a minimum, and a maximum value, and also set a default for them, if they are not specified):

      logfeeder.simulate.number_of_words - the number of different words that may appear in the simulated logs (50 / 1000000, default 1000)
      logfeeder.simulate.min_log_words - the minimum number of words in a simulated log message (1 / 10, default 5)
      logfeeder.simulate.max_log_words - the maximum number of words in a log message (10 / 20, default 10)

      After the parameters are set each logfeeder will start to load messages like this. In each message all the words are different:

      Word000912 Word000903 Word000940 Word000760 Word000564 Word000762 Word000955 Word000553

      Word000320 Word000218 Word000872 Word000204 Word000120 Word000298 Word000727 Word000720 Word000475

      Word000413 Word000642 Word000872 Word000464 Word000746 Word000250

      The rest of the properties remain the same:

      logfeeder.simulate.input_number - number of paralell inputs (threads) loading the logs, if it is set and not 0 then all the rest of the configured inputs are ignored (running in simulation mode)!!

      logfeeder.simulate.log_ids - comma separated list of the log ids to propagate at random, if not set by default all the available logs are propagated at random, example: storm_drpc,storm_logviewer,storm_nimbus

      logfeeder.simulate.log_level - the level of the simulated log messages, by default WARN

      logfeeder.simulate.sleep_milliseconds - the time interval at which each simulated inputs writes one log message at random




            mgergely Miklos Gergely
            mgergely Miklos Gergely
            0 Vote for this issue
            2 Start watching this issue