Building a Stratum 1 NTP Server Using Odroid C2

[bingo600]

@jcpingu

Have a look here
https://serverfault.com/questions/1048870/ntpd-fails-to-sync-time-error-0x41-clock-unsynchronized

Especially this command : ntpq -c sysinfo

Normally as @stephenw10 mentioned , ntpd won't just accept a new clocksource , before investigating if it's "ok".

I usually expect approx 5 minutes (after daemon start) for the daemon to "trust/select" a clocksource.

[jcpingu]

@bingo600
Here's the output of that command. I will check out the link. Thanks...

[jcpingu]

@stephenw10
This is my ntp.conf file. Based on bingo600 comments above. Is there anything else that needs to happen here?

server 127.127.20.0 mode 81 minpoll 4 maxpoll 4 prefer
fudge 127.127.20.0 flag1 1 refid GPS

security

restrict default kod limited nomodify nopeer
restrict -6 default kod limited nomodify nopeer

Local users may interrogate the NTP server fully.

restrict 127.0.0.1
restrict -6 ::1

stats

driftfile /var/lib/ntp/ntp.drift

Enable statistics logging.

statsdir /var/log/ntpstats/
statistics loopstats peerstats clockstats
filegen loopstats file loopstats type day enable
filegen peerstats file peerstats type day enable
filegen clockstats file clockstats type day enable

leap file

leapfile /var/lib/ntp/leap-seconds.list

[stephenw10]

No, I think it's working as expected.
That error will always appear when you restart ntpd for any reason. I would only vbe concerned if it is still logging new errors after running for some time.

Steve

[bingo600]

@jcpingu

Try to put in a few ntp pool servers for reference.

Add this to your ntp.conf

# Internet time servers for sanity
server 0.pool.ntp.org iburst prefer 
server 1.pool.ntp.org iburst 
server 2.pool.ntp.org iburst 
server 3.pool.ntp.org iburst

From here:
http://doc.ntp.org/4.2.6/drivers/driver20.html

It seems like the $GNRMC is not a supported sentence , and that either
$GPZDA or $GPZDG must be sent for the NTP system to be fully autonomous. As in being able to set date/time fully via NTP.

What GPS model do you have ?
Is it from "China" ... Where most Ublox'es are "fakes"

This is not an "all bad" mesage , as the PPS still will keep your clock extremely accurate (sub seconds) . You just need a bit of help setting the seconds.

But it's strange that the nmea driver is so picky.
Maybe gpsd would help here

/Bingo

[bingo600]

@bingo600

The NTP nmea refdriver doc is "buggy"

I just had a quick glance at the source
https://github.com/ntp-project/ntp/blob/master-no-authorname/ntpd/refclock_nmea.c

And it's comparing the last 3 (4) characters in the sentence so any RMC works

And it seems like it's parsing/setting the date from RMC sentenses too.
The only thing ZDA is providing "extra" is YYYY vs YY from RMC.

It should be possible to make a working NTP system with just RMC sentences , if NTP has been updated to assume that YY (21) means 2021.
Which i expect the unfold_century function does.

/Bingo

[jcpingu]

@bingo600 said in Building a Stratum 1 NTP Server Using Odroid C2:

ntp pool servers

Does it matter where the ntp pool servers are inserted in the ntp.conf file?

[bingo600]

@jcpingu

This guy puts it after the drift file

# Drift file to remember clock rate across restarts
driftfile /var/lib/ntp/ntp.drift
# Servers
pool uk.pool.ntp.org iburst

And have a look here
https://lintut.com/how-to-verify-ntp-setup-is-working-properly-in-linux-or-unix-server/

Verifying a stratum1 server might not be "super easy"

Use
ntpstat

My box says (No GPS installed) :

ntpstat
synchronised to NTP server (80.71.132.xxx) at stratum 2 
   time correct to within 48 ms
   polling server every 1024 s

And
ntpq -p

ntpq -p
     remote           refid      st t when poll reach   delay   offset  jitter
==============================================================================
-n1.taur.xx      .PPS0.           1 u  753 1024  377    7.164   -0.397   0.383
*80.71.132.10x.i .GPS.            1 u  985 1024  377    1.862   -0.114   0.111
+mmo2.ntp.xx  .PPS.            1 u  141 1024  377    2.827   -0.571   0.479
 5.103.139.163.s .GPS.            1 u   8d 1024    0    3.267    0.348   0.000
-82.180.61.122   193.204.114.233  2 u  735 1024  377   40.205    0.981   0.419
+grape.inet.xx.n 68.166.61.255    2 u  315 1024  377    1.207   -0.107   0.089

The peer with the * in front (here 80.71.132.10x.i ) is the one that is used (synced to) right now.

o = pps peer
* = sys peer
# = too distant
+ = selected
x = false ticker
- = discarded

Show the output of your ntpq -p

[jcpingu]

@bingo600
here's ntpq -p output

[jcpingu]

@bingo600
Here's systemctl status ntp output, still showing clock unsync error:

[stephenw10]

That error is when you started the ntp daemon. It has no data at that point, it will never sync. It's expected.

That output from ntpq looks good. You just need to wait for more peers with greater reach values.

Steve

[jcpingu]

@stephenw10
My question now is do I really need these? or was my original ntp.conf was just fine without these...

Internet time servers for sanity

server 0.pool.ntp.org iburst prefer
server 1.pool.ntp.org iburst
server 2.pool.ntp.org iburst
server 3.pool.ntp.org iburst

[stephenw10]

I would keep some external servers defined. More sources in agreement is going to make ntp happier, faster

[jcpingu]

@stephenw10
Thank you