George Magklaras wrote:
Ryan Golhar wrote:
9.2.2.2 says xen just passed the time-related calls to Dom0, but that
can't be the case because the times are different.
That is because if you do run ntpd inside the VM, this can create the
problem.
noselect? inside the VM?
Yes on the guest system, and run ntpd normally on the Dom0.
I'm running ntpd normally on the host, and ntpd inside the VM using the
noselect parameter on the server line, such as:
server siriusc.umdnj.edu noselect
server 0.rhel.pool.ntp.org noselect
server 1.rhel.pool.ntp.org noselect
server 2.rhel.pool.ntp.org noselect
After about a half hour after restarting the ntpd server inside the
guest, I'm noticing clock skew of about 30 seconds. I'm sure this will
continue to grow.
Anyone have any other ideas?
I'm surprised there isn't a solution to this. This is a huge problem,
especially for an enterprise system that relies on the date/time being
correct.
Ryan
We are running on a Sun SunFire X4150.
What I stated was on a Dell 2950 PowerEdge, but I shall get hold of a
SunFire and repeat the test, see if I get the same issue. It does state
that there might be some kernel parameters that need tweaking, but, I am
curious to find out what's going on, as we will be using SunFire kit
with Xen soon.
GM
George Magklaras wrote:
This page might offer some help:
http://support.ntp.org/bin/view/Support/KnownOsIssues#Section_9.2.2.
To recap, we use the 'noselect' keyword to avoid the running ntpd on
the client to attempt to sync with the defined servers and ntpq does
not report any disturbing drifts on the client...and we do run ntpd
normally (no 'noselect') on Dom0.
What happens (happened if you have already tried) if you try and do
this? And what's your hardware platform?
GM
--
redhat-list mailing list
unsubscribe mailto:redhat-list-request@xxxxxxxxxx?subject=unsubscribe
https://www.redhat.com/mailman/listinfo/redhat-list
