Metricbeat 5.3.1 not collecting host os process info

[dh2001]

We have Filebeat and Metricbeat running in a docker container. With
/proc mounted to /hostfs ro. SLES 12. Using -system.hostfs=/hostfs when running metricbeat. But we do not see any process info for processes in another container. This was working in 5.0.0 before we upgraded to 5.3.1. Did something change?

metricbeat.modules:

#------------------------------- System Module -------------------------------
- module: system
  metricsets:
    # CPU stats
    - cpu

    # System Load stats
    - load

    # Per CPU core stats
    #- core

    # IO stats
    #- diskio

    # Per filesystem stats
    - filesystem

    # File system summary stats
    #- fsstat

    # Memory stats
    - memory

    # Network stats
    - network

    # Per process stats
    - process
  enabled: true
  period: 30s
  processes: ['.*']
  filters: 
  - drop_event:
     when:
       equals:
         cgroup.memory.mem.limit.bytes: 18446744073709551615

[andrewkroh]

Have you looked in the Metricbeat logs for errors or warnings?

Can you try disabling cgroup metrics with process.cgroups.enabled: false (for an example see this).

[dh2001]

I did look in the logs and there were no errors. I just tried disabling cgroups and it did not make a difference.

We downgraded to 5.2.2 and it works in that release. It seems to stop working in 5.3.

[dh2001]

We have Metricbeat and Filebeat running in our own custom docker container. Could the docker metricset changes be causing our problems. I saw there were changes mentioned in the 5.3 release notes.