[THIN] Re: OT: Windows 2k web server can't keep time

  • From: "TSguy92 Lan" <tsguy92@xxxxxxxxx>
  • To: thin@xxxxxxxxxxxxx
  • Date: Tue, 9 Jan 2007 15:07:19 -0800

Thanks for the responses on this gang, a couple of further notes.

- We do have all of our ESX host systems syncing via NTP. As mentioned
previously I've set the flag to sync off the ESX host within VMtools which
didn't change the behavior we're seeing.
- we've moved this virtual system between multiple ESX host systems. Only
this single VM appears affected. This issue appears to occur regardless of
either ESX, or virtual host load.

I've dug through the VMware forums as well on this, but I haven't posted
about this there yet, but I'll give it a go in a few here.

As an interim workaround we have an update script running as a task from
another server, forcing our errant server to sync it's time each minute.
It's duct tape over the wound really . .but that's working with windows
sometimes.

I think Bob's comment on "something may be polling the clock" has some
merit, as we have PGP process tied to the webservices on this system which
are very time dependant. This might be a good lead to send our webdevs
chasing after as I look to open up a case with vmware.

thanks for the insights Thinners :)

Cheers,

Lan


On 1/9/07, TSguy92 Lan <tsguy92@xxxxxxxxx> wrote:

This is just so odd and I haven't come across anything useful in
websearches so I thought I'd bounce it off the mighty brains of this list :)

We have a windows 2000 sp4 server (currently running as a virtual machine
on ESX 3.01), that has started mis-counting the time alloted for Seconds.
I'm really not sure how else to phrase this, we can pull up the clock on the
server from the systray, and can visually watch the seconds tick away at
extended intervals. (IE - instead of 1 second every second, we'll see 1
second tick, then can have a pause of up to 3-5 seconds before the next
second ticks by.) This delay gradually builds up until the server is off
from the domain time by several minutes . . obviously this is an issue.

On a reboot of this machine, the problem is no longer evident, in fact we
rebooted it in the evening to avoid production down time, all was good when
it came back up, but this morning (shortly after start of business) our time
issues showed up again.

 As a work around we created a script to sync the server every minute and
set it to run as a local system task. But of course since the system counts
it's minutes rather slowly, the script doesn't run as often as it should, so
I'll be re-keying this to run from a remote system.

The setting within Vmware tools to "sync time between the virtual machine
and the console operating system" is disabled, however enabling it has not
made any notable changes. We've gone ahead and migrated this VM to a couple
different physical ESX hosts as well without any apparent differences
either.

My web-devs really want to point the finger at VMware, I'm inclined so far
to believe this is more of a windows issue, but I can't think of what it
could be. Any MS KB's that I've found related to system time have been
unhelpful so far. Only the fact that this issue appears to re-occur once our
business day starts up leans me twoards suspecting the underlying coding for
the webservices hosted on this system...we're working on getting time
allocated to disable webrelated services on this box to see if we can
isolate this issue to them.

If anyone has any thoughts for tracking this one down, I'd greatly
appreciate it. As is I'm loosing too much hair on this one ;)

TIA

Lan



Other related posts: