Privileged containers causing high CPU

1

I am trying to setup ConvoyNFS, following the instructions in the December Meetup video. During the discussion he installs a “simple” NFS server from CPUGuy83. Here is the docker compose he shows as an example:

nfs:
  log_driver: ''
  labels:
    io.rancher.scheduler.affinity:host_label: service=nfs
  command:
  - /exports
  log_opt: {}
  image: cpuguy83/nfs-server
  privileged: true
  volumes:
  - /exports

However it appears as tho running the container in privileged mode is causing the CPU to be pegged.

I have read here (getty processes at 100% cpu? · Issue #4040 · moby/moby · GitHub) that this is a known issue if you run containers in priveliged mode, but so far have not found a fix.

I am posting this here, because I have noticed this behaviour with other containers that launch through Rancher.

1 Like
2

I verified that it was the NFS container that was causing this. If I stop the container and kill the getty process, then CPU goes back to normal. Starting the container, causes the process to re-appear and use up all the CPU.

Perhaps there is a better way to launch this container without privileged? Perhaps there is a better way to do NFS testing?

I tested glusterfs and it worked okay, however really just don’t want to have to have 3 servers minimum to use it. I want all the source volumes to “live” on one server for now.

What alternatives are there that will let me create a simple NFS server which I can connect ConvoyNFS too?