Dynamic Template Not Working - float vs long - Metricbeat Windows Perfmon Counters

Hello,

I'm using ELK + Kafka.
Version : 7.6.2

I'm monitoring Windows machines with Metricbeat and the perfmon module to get some counters.

Here are some counters pulled (a lot of NTDS counters actually, just an example) :
windows.perfmon.logicaldisk.avg.disk.sec.write
windows.perfmon.logicaldisk.disk.read.bytes.sec
windows.perfmon.ntds.dirsync.sessions.inprogress

The problem is that some counters are initiated as "long" type, whereas it should be "float".
So I decided to delete all the data and the index ; configure dynamic Mapping with a template that match "metricbeat-*".

Here is the dynamic mapping i'm trying :

  {
    "windows.perfmon.*": {
      "unmatch": "windows.perfmon.instance.name",
      "match": "windows.perfmon.*",
      "mapping": {
        "type": "float"
      },
      "match_mapping_type": "*"
    }
  }

Unfortunately, when I'm starting the metricbeat agents and gathering some data, the type of the windows.perfmon.* counters are sometimes float or long ...

GET metricbeat-2020.07.29/_mapping/field/windows.perfmon.ntds.*
...
      "windows.perfmon.ntds.ds.client.name.translations.sec" : {
        "full_name" : "windows.perfmon.ntds.ds.client.name.translations.sec",
        "mapping" : {
          "sec" : {
            "type" : "long"
          }
        }
      },
      "windows.perfmon.ntds.ab.client.sessions" : {
        "full_name" : "windows.perfmon.ntds.ab.client.sessions",
        "mapping" : {
          "sessions" : {
            "type" : "long"
          }
        }
      },
...

I would love to use the dynamic mapping feature, cause it's clearly easier to manage current and future mappings ...
I would like to avoid to do a "hard" mapping with each counters I'm pulling (200+ counters)

Any Help is appreciated :wink:

Luc

Hi All,

I did a static mapping in place of dynamic ; which is fine now.

Regards.
Luc

This topic was automatically closed 28 days after the last reply. New replies are no longer allowed.