Has error making http request: Get "http://npipe/state": been fixed yet for Elastic Agent 7.10.2?

[dvo]

I'm testing a single Windows agent, and Kibana is littered with these errors, and it looks bad to the potential customer. I've got to get it cleaned up.

error making http request: Get "http://npipe/state": open \default-filebeat: The system cannot find the file specified.

I saw someone asked a question about this in December, and it was identified as a bug. Is there a fix yet?

Thank you!

This is my current policy:

id: 2963c470-7162-11eb-b02b-ab5e9f3e32c6
revision: 9
outputs:
default:
type: elasticsearch
hosts:
- 'https://34.121.222.170:9200'
agent:
monitoring:
enabled: true
use_output: default
logs: true
metrics: true
inputs:

• id: 7d4080d0-72d2-11eb-a8fc-114b4161e7c0
  name: windows-1
  revision: 4
  type: windows/metrics
  use_output: default
  meta:
  package:
  name: windows
  version: 0.2.11
  data_stream:
  namespace: default
  streams:
  • id: windows/metrics-windows.perfmon
    data_stream:
    dataset: windows.perfmon
    type: metrics
    period: 5m
    perfmon.ignore_non_existent_counters: true
    metricsets:
    • perfmon
      perfmon.queries:
    • instance:
      • '*'
        counters:
      • name: '% Processor Time'
        field: cpu_perc
        format: float
      • name: Working Set
        object: Process
        perfmon.group_measurements_by_instance: false
  • id: windows/metrics-windows.service
    data_stream:
    dataset: windows.service
    type: metrics
    period: 5m
    metricsets:
    • service
      fleet:
      kibana:
      protocol: https
      hosts:
  • '34.121.222.170:5601'

[MarianaD]

hi @dvo , I think you mean [Ingest Management] Monitoring metrics using npipe does not work · Issue #22957 · elastic/beats (github.com).
This fix seems to have made it to 7.11. Have you tried this version?

[dvo]

Yes, that's what I found on git as well. So there's no fix for 7.10.X agents---it requires upgrading the whole smash to 7.11 to fix? I say this b/c unless something has changed in the last 2 weeks, 7.11 Elastic Agents won't register w/7.10.2 backend.

Please tell me there's a workaround for this that can be applied to just the agent!

Thanks for your time Mariana

[MarianaD]

hi @dvo , I see in the initial discuss ticket, my colleague suggests disabling metrics monitoring using agent.monitoring.metrics in order to clear the errors, have you tried that?
It's most likely not possible to enroll newer agents to older stack.

[dvo]

Hi, yes I did, but does that disable metric colkection, or just the monitoring of collecting the metrics? Either way, the syntax for doing that in the config would have been helpful. I'd be willing to try it if I knew what to type.