[ntp:questions] Re: autokey, NIST leap-seconds, and TAI on Win32
martin.burnicki at meinberg.de
Fri Nov 18 11:33:56 UTC 2005
Serge Bets wrote:
> Hello Martin,
> On Wednesday, November 16, 2005 at 9:47:02 AM +0100, Martin Burnicki
>> Serge Bets wrote:
>>> leapsec=200506280000, tai=33". One month less, and one second more.
>> No, a bug which has been fixed between the version you're using on
>> Linux, and the one you're using under Windows.
> Thanks! I had looked at bug tracker before posting, but also seemingly
> before the report of bug 525. If it's this one.
>>> with autokey on Win32, I had very frequently service hangs, NTP
>>> becoming unresponsive, unstoppable, and unkillable.
>> We've also observed this and are currently looking for the reason why
>> this happens under Windows.
> I see in bug 527 that the problem is more important than just sysstats.
> Anyway I'll follow the events there to see if I can help.
> While at it, I never succeeded initiating a "peer server iburst autokey"
> association. I created a fresh trusted cert and id files on client,
> "ntp-keygen -T -I -G -V5 -q one_pw_4_both -p one_pw_4_both", then copied
> the full (not -e) IFF parms file from client to server, and the full IFF
> from server to client. Restart service, and got some negociation failing
> with lots of (Eventlog to Syslog) logs like:
> | Nov 14 12:49:50 client NTP: N/A: sendto(192.168.7.10) (fd=216): Une
> | opération d'entrée/sortie avec chevauchement est en cours d'exécution.
> Or repeated many times:
> | Nov 14 12:49:57 client NTP: N/A: sendto(192.168.7.10) (fd=216): No error
> And the association gets "reachable" only first poll, and never again,
> while "ntpdc -c listpeers server" never shows the sym_passive incoming.
> I'm wondering if I'm doing "peer autokey" correctly?
I'm not sure, either. I've only yet tested a Windows client using autokey
against a Linux server, which worked OK.
More information about the questions