all servers stuck in INIT after reboot

[Andrew Schulman]

Since I updated to 18.04.1 (ntp version 1:4.2.8p10+dfsg-5ubuntu7.1), ntpd at boot leaves all remote servers in the INIT state:

$ ntpq -c peers
Thu Nov 1 06:33:19 EDT 2018
     remote refid st t when poll reach delay offset jitter
==============================================================================
 otc2.psu.edu .INIT. 16 u - 64 0 0.000 0.000 0.000
 ntp-2.vt.edu .INIT. 16 u - 64 0 0.000 0.000 0.000
 dns3.cit.cornel .INIT. 16 u - 64 0 0.000 0.000 0.000

This happens after every reboot. I have to restart the ntp service, then it runs normally again.

ProblemType: Bug
DistroRelease: Ubuntu 18.04
Package: ntp 1:4.2.8p10+dfsg-5ubuntu7.1
ProcVersionSignature: Ubuntu 4.15.0-38.41-generic 4.15.18
Uname: Linux 4.15.0-38-generic x86_64
NonfreeKernelModules: nvidia_modeset nvidia
ApportVersion: 2.20.9-0ubuntu7.4
Architecture: amd64
CurrentDesktop: ubuntu:GNOME
Date: Thu Nov 1 18:23:58 2018
InstallationDate: Installed on 2015-05-01 (1280 days ago)
InstallationMedia: Ubuntu 14.04.2 LTS "Trusty Tahr" - Release amd64 (20150218.1)
NtpStatus:
 remote refid st t when poll reach delay offset jitter
 ==============================================================================
 +otc2.psu.edu 146.186.222.4 3 u 344 137m 375 31.612 0.222 1.043
 *ntp-2.vt.edu 198.82.247.164 2 u 89m 137m 377 14.930 1.267 2.263
 +dns3.cit.cornel 146.186.222.14 2 u 2032 137m 377 23.562 2.050 1.578
SourcePackage: ntp
UpgradeStatus: Upgraded to bionic on 2018-09-27 (35 days ago)
modified.conffile..etc.default.ntp: [modified]
modified.conffile..etc.ntp.conf: [modified]
mtime.conffile..etc.default.ntp: 2016-04-29T06:09:59.875852
mtime.conffile..etc.ntp.conf: 2016-07-19T06:06:23.592683

[Andrew Schulman (z-ubuntuone-andrex)]

OK, looking in /var/log/syslog, I see these entries from this morning's boot:

Nov 1 06:27:47 helium ntpd[1212]: ntpd 4.2.8p10@1.3728-o (1): Starting
Nov 1 06:27:47 helium ntpd[1212]: Command line: /usr/sbin/ntpd -p /var/run/ntpd.pid -4g -u 125:135
Nov 1 06:27:47 helium ntpd[1243]: proto: precision = 0.042 usec (-24)
Nov 1 06:27:47 helium ntpd[1243]: restrict: ignoring line 11, address/host 'louie.udel.edu' unusable.
Nov 1 06:27:47 helium ntpd[1243]: restrict: ignoring line 12, address/host 'clock.psu.edu' unusable.
Nov 1 06:27:47 helium ntpd[1243]: restrict: ignoring line 13, address/host 'ntp-2.vt.edu' unusable.
Nov 1 06:27:47 helium ntpd[1243]: restrict: ignoring line 14, address/host 'ntp0.cornell.edu' unusable.
Nov 1 06:27:47 helium ntpd[1243]: Listen and drop on 0 v4wildcard 0.0.0.0:123
Nov 1 06:27:47 helium ntpd[1243]: Listen normally on 1 lo 127.0.0.1:123
Nov 1 06:27:47 helium ntpd[1243]: Listening on routing socket on fd #18 for interface updates
Nov 1 06:27:52 helium ntpd[1243]: Listen normally on 2 eth0 192.168.1.2:123
Nov 1 06:27:52 helium ntpd[1243]: new interface(s) found: waking up resolver
Nov 1 06:33:40 helium ntpd[1243]: ntpd exiting on signal 15 (Terminated)
Nov 1 06:33:40 helium ntpd[1243]: 128.118.25.3 local addr 192.168.1.2 -> <null>
Nov 1 06:33:40 helium ntpd[1243]: 198.82.247.71 local addr 192.168.1.2 -> <null>
Nov 1 06:33:40 helium ntpd[1243]: 132.236.56.250 local addr 192.168.1.2 -> <null>
Nov 1 06:33:40 helium ntpd[3473]: ntpd 4.2.8p10@1.3728-o (1): Starting
Nov 1 06:33:40 helium ntpd[3473]: Command line: /usr/sbin/ntpd -p /var/run/ntpd.pid -4g -u 125:135
Nov 1 06:33:40 helium ntpd[3476]: proto: precision = 0.058 usec (-24)
Nov 1 06:33:40 helium ntpd[3476]: Listen and drop on 0 v4wildcard 0.0.0.0:123
Nov 1 06:33:40 helium ntpd[3476]: Listen normally on 1 lo 127.0.0.1:123
Nov 1 06:33:40 helium ntpd[3476]: Listen normally on 2 eth0 192.168.1.2:123
Nov 1 06:33:40 helium ntpd[3476]: Listening on routing socket on fd #19 for interface updates

So ntpd is at first ignoring all of the 'restrict' commands because the hostnames are 'unusable'. Why would that be? Is ntpd trying to start before the external network interface is up?

[Christian Ehrhardt  (paelzer)]

Hi Andrew,
Sorry I can't help you from the logs I see, but I wanted to let you know that NTP had accrued way too many issues over the time.
Therefore from Ubuntu 18.04 and onwards the recommended and primarily supported ntp components are:
- systemd-timesyncd (if you need NTP client only, default installed)
- chrony (for more advanced NTP client logic)
- chrony (NTP serving)

[Andrew Schulman (andrex)]

Thank you Christian. I tried chrony in the past, but found it less featureful. I may try one of the other ntp packages. Meanwhile, I guess let's leave this open in case there's a solution.