[linux] NTP a posun o 8-9hodin

[Martin Kyrc]

ahojte,
na jednom z dvojice linux (debian) serverov riesim problem s tym, ze sa 
mi akosi nechce zosynchronizovat cas voci ntp serveru. servre su takmer 
identicke instalacie (jeden v lan, druhy v dmz ako virtualna masina). 
pouzivaju rovnakych ntp daemonov a rovnake ntp zdroje.

v dmz-ke mam stale posun o nejakych asi 8-9hodin a nie a nie ho 
dosynchronizovat. momentalne pouzivam chronyd (pred tym ntpd).

# cat /etc/timezone
Europe/Bratislava

# date
Wed Nov 28 05:17:04 CET 2007
(skutocny cas je nieco okolo 13:25 CET)

chronyc> activity
200 OK
2 sources online
0 sources offline
0 sources doing burst (return to online)
0 sources doing burst (return to offline)

chronyc> sources
210 Number of sources = 2
MS Name/IP address           Stratum Poll LastRx Last sample
============================================================================
^* 10.2.1.1                      2    6      4  +9958us[+2733us] +/-   82ms
^~ ntp.bts.sk                    1    6      7  +4944ms[ +787ms] +/- 5562us

chronyc> tracking
Reference ID    : 10.2.1.1 (10.2.1.1)
Stratum         : 3
Ref time (UTC)  : Wed Nov 28 12:42:40 2007
System time     : 30374.361576 seconds slow of NTP time
Frequency       : 0.000 ppm fast
Residual freq   : -32768.000 ppm
Skew            : -32768.000 ppm
Root delay      : 0.138733 seconds
Root dispersion : 84.716980 seconds

a v logu toto:
Nov 28 05:16:20 virtual-dmz chronyd[17677]: Selected source 194.160.23.2
Nov 28 05:16:20 virtual-dmz chronyd[17677]: System clock wrong by 
30375.488301 seconds, adjustment started
Nov 28 05:16:20 virtual-dmz chronyd[17677]: System clock wrong by 
-2.528316 seconds, adjustment started
Nov 28 05:16:20 virtual-dmz chronyd[17677]: Can't synchronise: no 
selectable sources
Nov 28 05:16:23 virtual-dmz chronyd[17677]: Elapsed<0! n=7 i=0 
latest=[Wed 11/28/07 12:42:36.347441] doing=[Wed 11/28/07 
12:40:40.781108] elapsed=-29793.165641
Nov 28 05:16:23 virtual-dmz chronyd[17677]: Elapsed<0! n=7 i=1 
latest=[Wed 11/28/07 12:42:36.347441] doing=[Wed 11/28/07 
12:40:49.476659] elapsed=-29852.479670
Nov 28 05:16:23 virtual-dmz chronyd[17677]: Selected source 10.2.1.1
Nov 28 05:16:23 virtual-dmz chronyd[17677]: System clock wrong by 
2.654090 seconds, adjustment started
Nov 28 05:16:23 virtual-dmz chronyd[17677]: System clock wrong by 
1.717480 seconds, adjustment started
Nov 28 05:17:26 virtual-dmz chronyd[17677]: Can't synchronise: no majority

co presne znamena 'Can't synchronise: no majority'? nejak som sa nevedel 
dopatrat k potrebnej informacii.

ak vsak pouzijem ntpdate, tak sa cas nastavi spravne:

# ntpdate 10.2.1.1
28 Nov 13:46:52 ntpdate[17692]: step time server 10.2.1.1 offset 
30528.947125 sec
# date
Wed Nov 28 13:46:53 CET 2007

no po par hodinach sa o5 rozladi.

dik za nakopnutie, asi nieco prehliadam.

mk