Can you try the iburst option to see if it changes it?
http://www.cisco.com/en/US/docs/ios/netmgmt/command/reference/nm_10.html#wp1102197
I don't see how going back to 24T would change it as it has v4 there
too...unless there was something later that made it even worse.
Rodney
On 6/30/10 5:25 PM, Peter Rathlev wrote:
On Wed, 2010-06-30 at 14:12 -0400, Rodney Dunn wrote:
I'm not a NTP expert but I know there were some issues when it went to
v4 regarding sync times, etc...
One of them was:
CSCte91471 NTP v4 takes several hours to sync when multiple servers
are configured
From looking at the source code that fix doesn't appear to be in
15.0(1)M2.
You may want to try it with the next rebuild when it comes out.
That sounds just like it. I can see the 15.1(1)T also has the same bug,
so I guess there's no point in trying that one.
I've had an expect script run "show ntp status" every five minutes on
the two devices the last 5-6 hours, and if I extrapolate linearly
(R^2=0.76 though) the one running 15.0(1)M2 will be synchronized in
around 11 hours, from an offset of ~15 ms now.
The bug description says "... after few hours.", but this one will have
taken nearly 36 hours if the linear extrapolation is correct.
Tomorrow I will try to downgrade to 12.4(24)T3 to make sure it isn't
hardware related. The one running 12.4(24)T3 now has the past 5 hours
never exceeded 1 usec offset according to "show ntp status".
_______________________________________________
cisco-nsp mailing list [email protected]
https://puck.nether.net/mailman/listinfo/cisco-nsp
archive at http://puck.nether.net/pipermail/cisco-nsp/
