I have a Debian 10 system that uses chronyd
to keep its clock in sync. The config is pretty simple:
pool 2.debian.pool.ntp.org offline iburst
bindaddress ::1
bindaddress 127.0.0.1
bindcmdaddress 127.0.0.1
allow 127
deny
keyfile /etc/chrony/chrony.keys
driftfile /var/lib/chrony/chrony.drift
logdir /var/log/chrony
log tracking measurements statistics
maxupdateskew 100.0
directive.
hwclockfile /etc/adjtime
rtcsync
makestep 1 3
It's happily synced:
# chronyc sources
210 Number of sources = 4
MS Name/IP address Stratum Poll Reach LastRx Last sample
===============================================================================
^- time.panq.nl 2 6 0 83h -1247us[-1191us] +/- 26ms
^* time.cloudflare.com 3 6 0 83h +1343ns[ +58us] +/- 2669us
^- metronoom.dmz.cs.uu.nl 2 6 0 83h -63us[ -63us] +/- 25ms
^- . 3 6 0 83h +2171us[+2171us] +/- 64ms
However, the "root dispersion" keeps rising steadily. From What is NTP dispersion and how do I control it? it seems this is a measure of the maximum error in the upstream server's clock. It's rising quite slowly, the process has been up for about 70 hours or so and it stands at 22.5s. I do know from experience that this will keep rising until chronyd
restarts.
# chronyc tracking
Reference ID : E1FE1EBE (time.cloudflare.com)
Stratum : 4
Ref time (UTC) : Sun Jan 26 23:19:16 2020
System time : 0.000000005 seconds fast of NTP time
Last offset : +0.000056495 seconds
RMS offset : 0.000056495 seconds
Frequency : 79.909 ppm slow
Residual freq : +17.510 ppm
Skew : 56.420 ppm
Root delay : 0.004632703 seconds
Root dispersion : 22.573289871 seconds
Update interval : 1.6 seconds
Leap status : Normal
This seems unusual to me. I have a load of other systems that sync time with a Stratum 1 server where the root dispersion is low and constant. I don't think I'm doing anything odd on the config, and the idea that the "maximum error in the upstream clock" rises steadily smells a bit off.
Is this normal?