> Le 29 août 2015 à 04:10, ※踏梦狼※ <[email protected]> a écrit :
From your previous message you show ntpq data from your servers:
[root@S10-2-3-2 ~]# ntpq -p 10.2.3.30
remote refid st t when poll reach delay offset jitter
==============================================================================
+dns1.synet.edu <http://dns1.synet.edu/>. 202.118.1.46 2 u 95 1024 377
44.908 38.345 62.575
-gus.buptnet.edu <http://gus.buptnet.edu/> 202.112.10.60 2 u 744 1024 377
53.044 -4.059 34.174
*dns2.synet.edu <http://dns2.synet.edu/>. 202.118.1.46 2 u 757 1024 377
47.791 15.684 43.683
+news.neu.edu.cn <http://news.neu.edu.cn/> 202.118.1.46 2 u 70 1024 377
49.002 40.916 63.273
LOCAL(0) .LOCL. 10 l - 64 0 0.000 0.000 0.000
[root@S10-2-3-2 ~]# ntpq -p 10.2.4.30
remote refid st t when poll reach delay offset jitter
==============================================================================
*dns1.synet.edu <http://dns1.synet.edu/>. 202.118.1.46 2 u 316 1024 377
42.514 41.338 24.914
+gus.buptnet.edu <http://gus.buptnet.edu/> 202.112.10.60 2 u 874 1024 377
53.074 10.053 32.569
+dns2.synet.edu <http://dns2.synet.edu/>. 202.118.1.46 2 u 411 1024 375
42.782 43.630 23.393
LOCAL(0) .LOCL. 10 l 8d 64 0 0.000 0.000 0.000
For me this would not be considered acceptable. I would NEVER have ntp servers
in virtual machines. Especially if they are LAN primary servers. The figures
for delay and jitter are abnormally high. Maybe due to that.
There is also some « visibility » issue with the pool server dense.synet.edu..
from 10.2.4.30, though it is ok from the other.
Your maxpoll default of 10(1024s) for the pool servers is too high. Try
dropping it to 6 (64s) or 7(128s).
>
> yesterday , I delete drift file /var/lib/ntp/drift , restart ntpd services,
> it's appear freq_not_set in ntp.log 。but today /var/lib/ntp/drift file
> automatic appear，it's value -4.835，it's value seems always change。i want to
> ask what's meaning the values ?
>
> [root@S10-2-1-2 ~]# cat /var/log/messages
> Aug 28 18:53:00 S10-2-1-2 ntpd[24438]: Listening on routing socket on fd #27
> for interface updates
> Aug 28 18:53:00 S10-2-1-2 ntpd[24438]: format error frequency file
> /var/lib/ntp/drift
>
> [root@S10-2-1-2 ~]# cat /var/lib/ntp/drift
> -4.835
This is not a bad value. It indicates a 4.835 ppm ( parts per million ) slow
local clock. The file is updated periodically by ntpd and when you shut it
down.
>
> [root@S10-2-1-2 ~]# cat /var/log/ntp/ntp.log
> 28 Aug 18:53:00 ntpd[24438]: 0.0.0.0 c016 06 restart
> 28 Aug 18:53:00 ntpd[24438]: 0.0.0.0 c012 02 freq_set kernel 0.000 PPM
> 28 Aug 18:53:00 ntpd[24438]: 0.0.0.0 c011 01 freq_not_set
> 28 Aug 18:53:01 ntpd[24438]: 0.0.0.0 c614 04 freq_mode
> 28 Aug 19:10:33 ntpd[24438]: 0.0.0.0 0612 02 freq_set kernel 27.779 PPM
> 28 Aug 19:10:33 ntpd[24438]: 0.0.0.0 0615 05 clock_sync
> 29 Aug 01:58:38 ntpd[24438]: 0.0.0.0 0613 03 spike_detect +0.157552 s
> 29 Aug 02:15:53 ntpd[24438]: 0.0.0.0 0615 05 clock_sync
> you say possible reason is routing 。 A part of client machines time is normal
> in the same LAN conditions。the time unnormal client ntp machine restart ntpd
> services，it's time change ok。the ntp server is virtual machine，i don't know
> what's the wrong with my ntp client or ntp server ? But the time is unnormal
> client machine has same issue when use ntp -p,like this:
>
> [root@S10-2-3-2 ~]# ntpq -pn
> remote refid st t when poll reach delay offset jitter
> ==============================================================================
> 10.2.3.30 202.112.29.82 3 u 67 64 1 0.465 2.052 0.000
> 10.2.4.30 202.112.29.82 3 u 1 64 3 0.510 8.930 2.254
This client is not connecting reliably to either of the servers. The when
value of 10.2.3.30 exceeds the poll interval. That is to say transmitted
packets did not get a response. There are numbers of reasons why that may not
be the case which have nothing to to with ntp. The reach value is not optimal.
Anything under 377 is not good. Was this command executed just after a boot?
That might explain the low reach, but not the excessive poll. Neither server
has been selected as good enough to be used as a reference. Likewise, if
neither server has a stable connection and data, how can NTP chose between
them? Was this the complete output? If it is you are committing a cardinal
error in having only 2 servers.
A Man with 2 clocks cannot know what the time is. An ancient Chinese
proverb ;-) .
You MUST have at least 3, better 5.
If the above data was taken just when ntp restarted, monitor over time to see
how the reach/delay/jitter evolves.
If you see reach not 377 you have connection issues. That is not an NTP issue,
NTP is just the unfortunate victim. It could be network, machine load , process
scheduling bugs…
I would recommend that you look at getting your primary NTP servers off a VM
infrastructure. If this is a large business environment, recommend the
installation of dedicated primary time servers using reference clocks, GPS for
example.
>
>
>
> ------------------ 原始邮件 ------------------
> 发件人: "Mike Cook";<[email protected]>;
> 发送时间: 2015年8月28日(星期五) 晚上7:36
> 收件人: "※踏梦狼※"<[email protected]>;
> 抄送: "questions"<[email protected]>;
> 主题: Re: [ntp:questions] what's the matter with my ntp
>
>
> > Le 28 août 2015 à 12:36, ※踏梦狼※ <[email protected]> a écrit :
> >
> > hello,
> > you say 128ms max offset ,is it refer to the time difference of the
> > client ntp system time and ntp peer time? if the time difference bigger
> > 128ms, ntp don't sync clock ?
> >
> No, the system clock will be synced, but may be « stepped » rather than «
> slewed ».
>
> > thanks
> >
> >
> > ------------------ 原始邮件 ------------------
> > 发件人: "Mike Cook";<[email protected]>;
> > 发送时间: 2015年8月28日(星期五) 下午5:59
> > 收件人: "※踏梦狼※"<[email protected]>;
> > 抄送: "questions"<[email protected]>;
> > 主题: Re: [ntp:questions] what's the matter with my ntp
> >
> >
> > > Le 28 août 2015 à 11:43, ※踏梦狼※ <[email protected]> a écrit :
> > >
> > > The ntp client have stand alone and virtual machines, Those machines run
> > > serveral days，the time differ 30 seconds compare the normal time。 you
> > > look the below log:
> > >
> > > > 20 Aug 20:05:15 ntpd[14554]: 0.0.0.0 0615 05 clock_sync
> > > > 20 Aug 20:05:16 ntpd[14554]: 0.0.0.0 c618 08 no_sys_peer
> > > > 20 Aug 20:42:53 ntpd[14554]: 0.0.0.0 c613 03 spike_detect +0.949681 s
> > > > 20 Aug 21:00:09 ntpd[14554]: 0.0.0.0 c618 08 no_sys_peer
> > > > 21 Aug 14:24:50 ntpd[14554]: ntpd exiting on signal 15
> > >
> > > could you tell me what's the meaning of no_sys_peer、spike_detect and
> > > clock_step ? Are those normal ?
> >
> > What’s normal? The messages are in response to what the client ntpd
> > detects when comparing the local clock with the set of configured servers
> > (peers).
> > IIRC:
> > no_sys_peer Indicates that there is no server that satisfies ntpd’s
> > stability criteria.
> > spike_detect This is informational in that a difference between the local
> > clock and the valid servers was temporarily greater than the 128ms max
> > offset.
> > clock_step Normally modifies the local system clock frequency to adjust for
> > differences between it and the valid servers. If the offset is greater than
> > 128ms, then unless
> > configured otherwise, will step the system clock.
> > Mike
> >
> > >
> > > my clock seeting like this:
> > > [root@S10-2-1-3 ~]# cat /etc/sysconfig/clock
> > > ZONE="Asia/Shanghai"
> > >
> > > How i solved this problem ?
> > >
> > >
> > > ------------------ 原始邮件 ------------------
> > > 发件人: "Mike Cook";<[email protected]>;
> > > 发送时间: 2015年8月28日(星期五) 下午5:25
> > > 收件人: "※踏梦狼※"<[email protected]>;
> > > 抄送: "questions"<[email protected]>;
> > > 主题: Re: [ntp:questions] what's the matter with my ntp
> > >
> > >
> > > > Le 28 août 2015 à 04:52, ※踏梦狼※ <[email protected]> a écrit :
> > > >
> > > > 20 Aug 12:42:52 ntpd[14554]: 0.0.0.0 c618 08 no_sys_peer
> > > > 20 Aug 12:58:17 ntpd[14554]: 0.0.0.0 c612 02 freq_set kernel -586.374
> > > > PPM *********
> > > > 20 Aug 12:58:17 ntpd[14554]: 0.0.0.0 c61c 0c clock_step -0.522594 s
> > > > 20 Aug 12:58:17 ntpd[14554]: 0.0.0.0 c615 05 clock_sync
> > > > 20 Aug 12:58:18 ntpd[14554]: 0.0.0.0 c618 08 no_sys_peer
> > > > 20 Aug 13:11:51 ntpd[14554]: 0.0.0.0 0613 03 spike_detect +0.247108 s
> > > > 20 Aug 13:28:12 ntpd[14554]: 0.0.0.0 061c 0c clock_step +0.174737 s
> > > > 20 Aug 13:28:12 ntpd[14554]: 0.0.0.0 0614 04 freq_mode
> > > > 20 Aug 13:28:13 ntpd[14554]: 0.0.0.0 c618 08 no_sys_peer
> > > > 20 Aug 13:43:52 ntpd[14554]: 0.0.0.0 c612 02 freq_set kernel -461.842
> > > > PPM
> > > > 20 Aug 13:43:52 ntpd[14554]: 0.0.0.0 c615 05 clock_sync
> > > > 20 Aug 19:48:08 ntpd[14554]: 0.0.0.0 0613 03 spike_detect +1.010504 s
> > > > *********
> > > > 20 Aug 20:05:14 ntpd[14554]: 0.0.0.0 061c 0c clock_step +1.491656 s
> > > > 20 Aug 20:05:15 ntpd[14554]: 0.0.0.0 0615 05 clock_sync
> > > > 20 Aug 20:05:16 ntpd[14554]: 0.0.0.0 c618 08 no_sys_peer
> > > > 20 Aug 20:42:53 ntpd[14554]: 0.0.0.0 c613 03 spike_detect +0.949681 s
> > > > 20 Aug 21:00:09 ntpd[14554]: 0.0.0.0 c618 08 no_sys_peer
> > > > 21 Aug 14:24:50 ntpd[14554]: ntpd exiting on signal 15
> > > > 28 Aug 10:25:48 ntpd[5937]: 0.0.0.0 c016 06 restart
> > > > 28 Aug 10:25:48 ntpd[5937]: 0.0.0.0 c012 02 freq_set kernel -471.344 PPM
> > > > 28 Aug 10:29:04 ntpd[5937]: 0.0.0.0 c61c 0c clock_step +0.233855 s
> > > > 28 Aug 10:29:04 ntpd[5937]: 0.0.0.0 c614 04 freq_mode
> > > > 28 Aug 10:29:05 ntpd[5937]: 0.0.0.0 c618 08 no_sys_peer
> > > > 28 Aug 10:44:43 ntpd[5937]: 0.0.0.0 c612 02 freq_set kernel -25.526 PPM
> > > > 28 Aug 10:44:43 ntpd[5937]: 0.0.0.0 c61c 0c clock_step +0.418623 s
> > > > 28 Aug 10:44:44 ntpd[5937]: 0.0.0.0 c618 08 no_sys_peer
> > > >
> > >
> > > The above indicates that this clients local clock (cpu) is running too
> > > slow, exceeding the ntpd’s max drift factor of 500ppm and also the max
> > > offset factor of 128ms.
> > > So if this is the only system affected it looks like the clock is sick.
> > > Monitor it’s drift outside NTP .
> > > Is this the only client seeing the issue, or are others? Are the clients
> > > stand alone or virtual machines? If virtual, maybe you are missing a
> > > system config parameter.
> > >
> > > Hope that helps.
> > >
> > > "Ceux qui sont prêts à abandonner une liberté essentielle pour obtenir
> > > une petite et provisoire sécurité, ne méritent ni liberté ni sécurité."
> > > Benjimin Franklin
> >
> > "Ceux qui sont prêts à abandonner une liberté essentielle pour obtenir une
> > petite et provisoire sécurité, ne méritent ni liberté ni sécurité."
> > Benjimin Franklin
>
> "Ceux qui sont prêts à abandonner une liberté essentielle pour obtenir une
> petite et provisoire sécurité, ne méritent ni liberté ni sécurité."
> Benjimin Franklin
"Ceux qui sont prêts à abandonner une liberté essentielle pour obtenir une
petite et provisoire sécurité, ne méritent ni liberté ni sécurité."
Benjimin Franklin
_______________________________________________
questions mailing list
[email protected]
http://lists.ntp.org/listinfo/questions
