[ntp:questions] Regular spike_detect on syslog
michael.cook at sfr.fr
Sat Sep 15 20:55:04 UTC 2018
> Le 14 sept. 2018 à 10:50, Sean Austin Critica <sean.critica at gmail.com> a écrit :
> I'm new to configuring NTP, and am having a bit of trouble figuring out why
> there are regular spike_detect messages on my syslog.
> I already checked the connection to the NTP server and it is stable with no
> packet drops.
> Additionally, that same NTP server is used by another machine on the same
> network and it is synchronizing fine. So it seems isolated to this
> particular machine.
> I find one particular sequence curious. From 17:30 to 18:53, there was a
> positive spike, followed by a negative spike, followed by a positive spike
> interleaved with clock_sync.
> Can you give me some hints regarding what may be causing this?
> kernel 0.059 PPM
> Jun 5 14:50:21 info localhost ntpd: 0.0.0.0 c615 05 clock_sync
> Jun 5 15:03:36 info localhost ntpd: 0.0.0.0 0618 08 no_sys_peer
I think that it is probably your servers clock. NTP is unable to decide if a potential peer is stable enough.
> Jun 5 15:07:55 info localhost ntpd: 0.0.0.0 0613 03 spike_detect
> +0.269228 s
> Jun 5 15:16:37 info localhost ntpd: 0.0.0.0 061c 0c clock_step
> +0.303049 s
+64 PPM if my sums are OK
> Jun 5 15:16:37 info localhost ntpd: 0.0.0.0 0614 04 freq_mode
> Jun 5 15:16:38 info localhost ntpd: 0.0.0.0 c618 08 no_sys_peer
> Jun 5 15:31:55 info localhost ntpd: 0.0.0.0 c612 02 freq_set
> kernel -186.040 PPM
This is anomalous . Not yet outside NTPs limit of 500ppm but modern hardware is usually much better.
Maybe your server is frequency shifting . Check your BIOS settings .
Are you using VMWARE or something like it. If it is not configured correctly virtual systems can get duff clock states.
< snip rest >
> Sean Critica
> questions mailing list
> questions at lists.ntp.org
People have only as much liberty as they have the intelligence to want and the courage to take.
More information about the questions