Hi--
On Apr 7, 2014, at 12:12 PM, Harlan Stenn <[email protected]> wrote:
> Why would you not want to iburst all of the servers?
iburst helps to populate the reachability field quickly.
That reduces the interval it takes for an NTP server to move from:
associd=0 status=c012 leap_alarm, sync_unspec, 1 event, freq_set =>
associd=0 status=0615 leap_none, sync_ntp, 1 event, clock_sync
...which is the transition needed before clients of the server will trust it
for time.
However, the transition doesn't happen any faster if you only use one server
with iburst,
or with all of the server lines. The data I've seen doesn't show any
significant
differences to the local clock variables like frequency, sys_jitter, clk_jitter.
(My offset is bouncing around because ntpd is being started up using -g.)
Regards,
--
-Chuck
PS: For more detailed comparison; let's consider three examples. #1; start
with:
$ cat /etc/ntp.conf
server 17.254.0.27 # aka time1.apple.com
server 17.254.0.28 # aka time2.apple.com
server 17.254.0.31 # aka time3.apple.com
server 17.151.16.20 # aka time4.apple.com
Restart ntpd. Run:
repeat 100; do; sleep 10; ntpq -npcrl; done
It takes ~200 seconds before the transition happens:
remote refid st t when poll reach delay offset jitter
==============================================================================
*17.254.0.27 17.150.142.121 2 u 6 64 17 3.903 14.965 0.082
17.254.0.28 17.150.142.121 2 u 136 64 14 55.557 40.997 2.545
17.254.0.31 17.150.142.121 2 u 67 64 6 41.580 33.254 4.612
17.151.16.20 17.107.131.11 2 u 67 64 7 3.183 14.882 0.109
associd=0 status=0615 leap_none, sync_ntp, 1 event, clock_sync,
version="ntpd [email protected] Fri May 28 01:20:53 UTC 2010 (1)",
processor="x86_64", system="Darwin/12.5.0", leap=00, stratum=3,
precision=-20, rootdelay=4.911, rootdisp=980.970, refid=17.254.0.27,
reftime=d6ed797f.fcf6fd16 Mon, Apr 7 2014 12:22:39.988,
clock=d6ed7985.0fb84e27 Mon, Apr 7 2014 12:22:45.061, peer=39832, tc=6,
mintc=3, offset=14.965, frequency=-11.064, sys_jitter=0.082,
clk_jitter=5.291, clk_wander=0.000
======
#2; start with:
$ cat /etc/ntp.conf
server 17.254.0.27 iburst
server 17.254.0.28
server 17.254.0.31
server 17.151.16.20
Restart ntpd. Run:
repeat 100; do; sleep 10; ntpq -npcrl; done
...except there's no need, since ntpd claims to be synced immediately:
remote refid st t when poll reach delay offset jitter
==============================================================================
*17.254.0.27 17.150.142.121 2 u 6 64 1 3.301 -33.495 0.077
17.254.0.28 .INIT. 16 u - 64 0 0.000 0.000 0.000
17.254.0.31 .INIT. 16 u - 64 0 0.000 0.000 0.000
17.151.16.20 17.107.131.11 2 u 8 64 1 3.198 -33.747 0.001
associd=0 status=0615 leap_none, sync_ntp, 1 event, clock_sync,
version="ntpd [email protected] Fri May 28 01:20:53 UTC 2010 (1)",
processor="x86_64", system="Darwin/12.5.0", leap=00, stratum=3,
precision=-20, rootdelay=4.308, rootdisp=251.489, refid=17.254.0.27,
reftime=d6ed7a30.28ff9651 Mon, Apr 7 2014 12:25:36.160,
clock=d6ed7a36.32b6d90d Mon, Apr 7 2014 12:25:42.198, peer=31035, tc=6,
mintc=3, offset=-33.495, frequency=-11.068, sys_jitter=0.077,
clk_jitter=11.076, clk_wander=0.001
======
#3; start with:
$ cat /etc/ntp.conf
server 17.254.0.27 iburst
server 17.254.0.28 iburst
server 17.254.0.31 iburst
server 17.151.16.20 iburst
Restart ntpd. Run:
repeat 100; do; sleep 10; ntpq -npcrl; done
...except there's no need (again), since ntpd becomes synced immediately:
remote refid st t when poll reach delay offset jitter
==============================================================================
*17.254.0.27 17.150.142.121 2 u 8 64 1 3.246 -28.288 0.597
17.254.0.28 .INIT. 16 u - 64 0 0.000 0.000 0.000
17.254.0.31 .INIT. 16 u - 64 0 0.000 0.000 0.000
17.151.16.20 17.107.131.11 2 u 5 64 1 2.787 -28.584 0.070
associd=0 status=0615 leap_none, sync_ntp, 1 event, clock_sync,
version="ntpd [email protected] Fri May 28 01:20:53 UTC 2010 (1)",
processor="x86_64", system="Darwin/12.5.0", leap=00, stratum=3,
precision=-20, rootdelay=3.798, rootdisp=985.368, refid=17.254.0.27,
reftime=d6ed7adc.97957b98 Mon, Apr 7 2014 12:28:28.592,
clock=d6ed7ae6.03b8c8dc Mon, Apr 7 2014 12:28:38.014, peer=51228, tc=6,
mintc=3, offset=-28.318, frequency=-11.064, sys_jitter=0.787,
clk_jitter=10.012, clk_wander=0.000
[ Perhaps I should retry with ntpd [email protected]. And track the variables
over a longer period of time, perhaps several hours.... ]
_______________________________________________
questions mailing list
[email protected]
http://lists.ntp.org/listinfo/questions
