Reminder to self: Today I am disabling Hyper-V time sync integration services on my DC VMs and (running) Linux VMs, so basically almost all of them. I’ll need to see if anything starts drifting, but I think all the Linux machines have ntp or equivalient configured by default, so I’ll just need to be sure the DCs are set up to sync via NTP.

A random unordered list of commands that may or may not have been used:

w32tm /query /peers
w32tm /resync
netdom /query fsmo
Move-ADDirectoryServerOperationMasterRole -Identity lennydc -OperationMasterRole RIDMaster,PDCEmulator, InfrastructureMaster, SchemaMaster, DomainNamingMaster
w32tm /config /manualpeerlist:",,," /syncfromflags:manual /reliable:yes /update
w32tm /config /syncfromflags:domhier /update

I also moved all the FSMO roles off the old “primary” DC and onto LennyDC.

Other thoughts

  • I’ll need to change my non-AD DNS zones as they are currently homed on the old primary AD server.
  • Why is my laptop saying it’s synced but has the wrong time?
  • Check later if the picky java apps on Linux quit complaining about clock skips and reverses.
    • They did it once more at 2016-12-18 17:52, but that’s when I finally got NTP working on that box by enabling systemd-resolved, so I’ll need to keep monitoring
  • Windows can natively be an NTP server; do I want to enable it and point my Linux VMs at it?


Lessons Learned

  • One one of my systemd-based hosts, systemd-resolved was disabled which prevented name resolution from happening which was the root cause of time not syncing

      sudo timedatectl
      sudo systemctl status systemd-timesyncd ntpd
      sudo vi /etc/systemd/network/
      sudo systemctl enable systemd-resolved
      sudo systemctl restart systemd-resolved
      sudo systemctl restart systemd-networkd
      cat /etc/resolv.conf
  • Noticed that one of my coreos boxes keeps churning IPV6 adapters resulting in NTP continually reconnecting