System clock jumping backwards on



3 years ago
3 years ago


(Reporter: kaie, Assigned: ryanc)





3 years ago
Hi Dustin,

you had helped in the past to setup this machine:
which is a VM on parallels.

We have repatedly experienced test suite failures, because the system clock jumped backwards.

Apparently, there are two different daemons attempting to sync the time, see /var/log/system.log

I see entries from "com.parallels.TimeSync" which apprently decides not to sync the time at all, and occassionally I see ntpd, which decides to move the time backwards.

Is there a conflict between daemons? Is there a problem with the configuration to sync from the host? Do you have any ideas how this could be fixed?

Thanks in advance!


3 years ago
Component: Community IT → MOC: Service Requests
QA Contact: mzeier → lypulong

Comment 1

3 years ago
(fwiw, I don't think this belongs under MOC, but it definetely didn't belong under Community IT. Thanks Tanner!)
This does belong in MOC, actually :)

I don't know if we have anyone on staff who knows much about Parallels.  Are there Parallels KB articles on doing time sync for Leopard?  Perhaps TimeSync is seeing that ntpd is enabled, and that's why it's not doing any sync?  In general, ntp doesn't work well on VMs due to cycle stealing.  Have you tried disabling the ntpd service?

Comment 3

3 years ago
Thanks for the suggestion, let's try that.

I've executed 
  sudo launchctl unload -w /System/Library/LaunchDaemons/org.ntp.ntpd.plist

to disable ntpd.

Comment 4

3 years ago
We don't usually work with these -- the host is accessible via SSH, but requires a password.

Please specify where to locate credentials if you'd like us to have a look.
They're in the community password file.

Comment 6

3 years ago

Disabled time sync from hypervisor as mentioned here,

Enabled NTP again. Time looks to be in sync. Reopen if you're still having issues.
Assignee: nobody → rchilds
Last Resolved: 3 years ago
Resolution: --- → FIXED
You need to log in before you can comment on or make changes to this bug.