No CPU/Mem metrics on docker module for windows

Elastic Stack Beats
1

Hi,

I'm trying to configure metric beat on windows to retrieve docker container's metrics. However, it looks like Kibana is always showing 0% for the CPU and the Memory for my Containers. I'm talking about the Infrastructure View for Docker in Kibana.

I also tried to only activate the memory set in the docker module, along with debug logs. Metricbeat starts correctly, but after 5min I still cannot see any metrics related to docker in the metricbeat logs...

logs:

DEBUG	[module]	module/wrapper.go:117	Starting Wrapper[name=docker, len(metricSetWrappers)=1]
DEBUG	[cfgfile]	cfgfile/reload.go:118	Checking module configs from: C:\ProgramData\chocolatey\lib\metricbeat\tools/modules.d/*.yml
DEBUG	[bus]	bus/bus.go:72	docker: map[start:true container:0xc0423045b0]
INFO	[monitoring]	log/log.go:117	Starting metrics logging every 30s
DEBUG	[bus]	bus/bus.go:72	docker: map[start:true container:0xc042304620]
DEBUG	[module]	module/wrapper.go:171	docker/memory will start after 8.94055280
...
DEBUG	[module]	module/wrapper.go:179	Starting metricSetWrapper[module=docker, name=memory, host=//./pipe/docker_engine]

Working features:

  • Host Metrics are correctly sent
  • Network and DiskIO for containers are correctly sent
  • Filebeat works for the logs (Host/Containers)

current conf for metricbeat

metricbeat.autodiscover:
  providers:
    - type: docker
      host: "npipe:////./pipe/docker_engine"
      hints.enabled: true

metricbeat.modules:
- module: docker
  metricsets:
    - "container"
    - "cpu"
    - "diskio"
    - "healthcheck"
    - "info"
    #- "image"
    - "memory"
    - "network"
  hosts: ["npipe:////./pipe/docker_engine"]
  period: 10s
  enabled: true

I had the same behavior for the following setups:
OS: windows 10 or windows server 1803 (1803 is a requirement for us) - installed locally not via docker
ES: v6.7.0 or 7.0.0-RC1

2

@fpoulin09 can you post the entire metricbeat.yml file and can you confirm if you run metricbeat -e -d "*" you are not able to see any published events when the only metricset enabled is memory after a while?

If you query this index in es, can you see any documents for the cpu cores or memory (if so, are these values always equal to 0?):

GET metricbeat*/_search?size=100
{
  "query":
  {
    "match": {
      "metricset.name": "memory"   //cpu
    }
  }
}

I just want to confirm that data is actually flowing in.

1 Like
3

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