Linux server time getting out of sync frequently.

[John Horne]

On Fri, 2008-06-20 at 12:50 +0200, Mertens, Bram wrote:
> 
> -----Original Message-----
> > From: redhat-list-bounces at redhat.com 
> > [mailto:redhat-list-bounces at redhat.com] On Behalf Of John Horne
> > Sent: woensdag 28 mei 2008 0:04
> > To: General Red Hat Linux discussion list
> > Subject: Re: Linux server time getting out of sync frequently.
> > 
> > On Tue, 2008-05-27 at 14:50 -0700, Josh Miller wrote:
> > > John Horne wrote:
> > > >> as i recall, there are a few parameters you can set 
> > > within the vmware app for the guest os, in order to sync the timing, and to 
> > stop the skew from occuring...
> > > >>
> > > >> i think you might also have to modify the kernel startup 
> > attributes.
> > > >>
> > > > The only solution I have found to work is to stop NTP on 
> > the guest and
> > > > simply run ntpdate (getting the time from other reliable 
> > server) every
> > > > hour or so via cron. The only 'solution' I have not tried 
> > is rebuilding
> > > > the kernel. Suggestions like use the PIT time source on the kernel
> > > > startup line may well improve the timekeeping, but it 
> > still loses time.
> > > 
> > > Hi, coming into this late, but I have been very successful with the 
> > > following solution:
> > > 
> > > 1. make sure all ESX hosts are syncing time via NTP with a 
> > reliable source
> > > 2. disable NTPD in all guests
> > > 3. set each guest to sync time via VMware tools by setting 
> > > tools.SyncTime=TRUE
> > > 4. in each guest, on the kernel line in grub.conf, set 
> > clock=pit and reboot
> > > 
> > Tried it, but didn't work. The guests still lost time 
> > (although the host
> > was still accurate).
> 
> I don't know if this is still an issue for you.  I haven't seen any
> report you got this solved so I thought I'd add our experience.
> 
> A while ago we noticed severely degraded performance on some of our
> systems (running WebSphere deployment managers).  And also noticed they
> were having problems syncing time.  The ntpd daemon died just like in
> your situation.
> 
> The systems would report 100% CPU usage while on the host system CPU
> utilization was close to nihil.  Although CPU utilization inside a VM
> can not be reliably measured this decrepancy was huge.  Our VM
> maintainer tried all sorts of tricks - moving the VM back and forth,
> trying various options, kernel options and so on to no avail.  Then as
> the box was originally built on an Intel host but was currently running
> on an AMD host the VM was moved back to Intel.  At this point both the
> host and the VM reported very high CPU usage.  After moving back to AMD
> again the host reported that the VM was hardly suing any CPU untill we
> removed the Vmware tools and reinstalled them.
> 
> So right now everything is back like it was configured before but both
> the performance and time syncing issues have been resolved.  The only
> difference I see is that the vmware tools have now been configure while
> the VM was running on the AMD host.
> 
> So perhaps all you need to do is reconfigure the Vmware tools?  Perhaps
> upgrade to the latest version as well?
> 
In our case the hardware hasn't changed, so reconfiguring the tools
would make no difference.


John.

-- 
---------------------------------------------------------------
John Horne, University of Plymouth, UK  Tel: +44 (0)1752 587287
E-mail: John.Horne at plymouth.ac.uk       Fax: +44 (0)1752 587001