[ntp:questions] Re: Watching the leap.

Ulrich Windl Ulrich.Windl at RZ.Uni-Regensburg.DE
Fri Jan 20 13:51:25 UTC 2006


Simon Lyall <simon at darkmere.gen.nz> writes:

> Well I'm not sure my machines all worked correctly today. Running ntpd 
> 4.1 and 4.2 on Debian/Ubuntu boxes that are stratum 3 and 4 I didn't see 
> any leap second indicators from most of my peers. And the time went all 
> boringly:
> 
> Sun Jan  1 12:59:59 NZDT 2006
> Sun Jan  1 12:59:59 NZDT 2006
> Sun Jan  1 12:59:59 NZDT 2006
> Sun Jan  1 12:59:59 NZDT 2006

FYI: In syslog my box with Novell/SuSE SLES9 SP3 (SuSE Enterprise Server 9 Service Pack 3) reported at 00:59:59 (UTC+1hour):
"kernel: Clock: inserting leap second 23:59:60 UTC"

NTP logged a "STA_INS" ("kernel time sync enabled 0011") at 00:32:37 (23:32:37
UTC), and at 00:40:58 the STA_INS was reset again.

Note that STA_INS might have been a bit late if you nmeed to propagate it in
time (if you poll every 20 minutes).

That was all Linux. On HP-UX 11.23 (shipping with old xntpd 3.5-something) the
message at 00:00:01 UTC said: "xntpd[...]: leap second occurred, stepped time
back 1 second"

> Sun Jan  1 13:00:00 NZDT 2006
> Sun Jan  1 13:00:00 NZDT 2006
> Sun Jan  1 13:00:01 NZDT 2006
> 
> Watching my peers now I can see more and more of them with offsets of 
> around 1 second so I guess in a few minutes my machines will decide they 
> are right and advance by themselves.
> 
> But not as tidy as I would have hoped. I think it is going to take 1024 
> seconds per stratum for many machines to get updated.

"updated" == "armed/alarmed"
Correct.

Regards,
Ulrich




More information about the questions mailing list