Hello! On Tue, Aug 01, 2006 at 07:02:47PM +0200, Kurt Roeckx wrote: > On Tue, Aug 01, 2006 at 11:18:08AM +0200, Thomas Schwinge wrote: > > Might this be a problem with the kernel not fulfilling the adjtimex() > > calls? > > I see 2 strange things here: > [...] > So it's not really making any sense to me why offset is 0 > in all but the last cases. I'm not sure which value > strace shows, the one we're trying to set, or the one the > kernel returns, but I assume both should be the same in > this case. > > So this smells like a kernel bug to me.
I don't know if you noticed in the original report (I should have been making it more prominent there): this is in the dom0 of a Xen system, i.e. the Linux 2.6 kernel running under the Xen-3.0 hypervisor. Perhaps that's relevant? > The other weird thing: > > Jul 31 17:15:01 taylor ntpd[5444]: reply from 192.83.249.31: negative delay > > -0.001068 > > Jul 31 17:15:07 taylor ntpd[5444]: reply from 192.36.143.151: negative > > delay -0.000159 > > Jul 31 17:15:09 taylor ntpd[5444]: reply from 64.52.111.11: negative delay > > -0.000107 > > Jul 31 17:53:28 taylor ntpd[5444]: reply from 64.5.1.130: negative delay > > -0.000125 > > All your 4 peers seem to have "negative delay". This > basicly means that you got a reply from the peer before > you've even send the request. Interesting, isn't it? ;-) So far I didn't bother with that, as I wasn't able to naturally interpret that output of openntpd and at that time didn't look it up in the code. Perhaps we should / have to forward this to some Xen people? Regards, Thomas -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
