Using Meinberg NTP server on my laptop and pointing my MicroTik Hap Lite to it for time.
My laptop has correct time and date.
My Hap Lite is always wrong.
Laptop time :
C:\Windows\system32>time
The current time is: 18:52:34.74
Enter the new time:
C:\Windows\system32>date
The current date is: Sun 02/13/2022
Enter the new date: (mm-dd-yy)
C:\Windows\system32>
Hap Lite time:
Wifi address
10.117.87.134 / 8 |
LAN address
10.170.188.49 / 29 |
WAN address
192.168.0.19 / 24 |
default gateway
192.168.0.1 |
SSID
AREDN-10-v3 |
Channel
-2 |
Bandwidth
10 MHz |
|
Signal/Noise/Ratio
N/A Charts |
firmware version
3.22.1.0 |
system time
Sun Jan 16 2022
17:53:35 PST |
uptime
load average
28 min
0.13, 0.03, 0.00 |
free space
flash = 8828 KB
/tmp = 29924 KB
memory = 34904 KB |
OLSR Entries
Total = 0
Nodes = 0 |
|
Do you have time zone and DST set correctly?
I had not at first noticed the date error. Obviously more than just time zone...
It is correct on my Laptop and when I use FT-8 I am only .004 or less different than the other station.
Is the Time Zone and DST set correct on My NTP server? Need to check that.
Sorry, I am not familiar with Meinberg, but could it be configured as a 'client' only and not a 'NTP server'?
73, Chuck
Thanks for catching it.
Andre
On the AREDN network I am using a NTP server running on a raspberry pi,
which uses my home LAMP server as a reference, which uses ubuntu.pool.net.
gelmce@nc8q-mesh-server:~ $ ntpq -p
remote refid st t when poll reach delay offset jitter
==============================================================================
*www 50.205.57.38 2 u 327 1024 377 4.522 -0.463 3.628
gelmce@nc8q-mesh-server:~ $ ssh www
gelmce@www's password:
Last login: Sun Feb 13 10:43:09 2022 from 192.168.8.99
gelmce@www:~$ ntpq -p
remote refid st t when poll reach delay offset jitter
==============================================================================
0.ubuntu.pool.n .POOL. 16 p - 64 0 0.000 0.000 0.000
1.ubuntu.pool.n .POOL. 16 p - 64 0 0.000 0.000 0.000
2.ubuntu.pool.n .POOL. 16 p - 64 0 0.000 0.000 0.000
3.ubuntu.pool.n .POOL. 16 p - 64 0 0.000 0.000 0.000
ntp.ubuntu.com .POOL. 16 p - 64 0 0.000 0.000 0.000
+157.245.125.229 209.51.161.238 2 u 140 1024 377 33.310 0.336 1.556
*50-205-57-38-st .GPS. 1 u 663 1024 377 35.849 1.003 0.995
+t1.time.bf1.yah 98.139.133.62 2 u 361 1024 377 38.975 0.571 1.656
+ntp2.lonet.org 129.134.29.123 2 u 535 1024 377 32.248 0.009 2.397
+chilipepper.can 17.253.34.251 2 u 832 1024 377 96.039 1.069 1.685
gelmce@www:~$
Do any other devices keep time using your 'Meinberg' NTP service?
73, Chuck
I've been musing about mesh time, and just decided not to care too much. Someone does have a ntp server up on my local mesh about 4 hops away, I was surprised ntpd could lock onto it; ping time was about 40 msec, jitter was over 30. As long as the jitter is relatively consistent, ntpd can deal with this. Don't expect sub-millisecond accuracy with those stats.
Can the AREDN OS actually run ntpd? All I see on mine is "/usr/sbin/ntpclient -s -l -h <server IP>" which seems to be a hung process started at boot.
Maybe ntpclient is hung because my own ntp server is usually never reachable? Or the call to adjtimex() never returns? I don't have any clients (only Fedora and Raspbian) that have the ntpclient available in a package so I can test this.
So -- maybe check that the ntp server is reachable and locked to a time source before the node is booted?
Also the last time I dealt with this problem, Linux syncs the hardware RTC to UTC, Windows syncs it to local. Make sure your computers are doing it correctly.
I have thought about buying a real time clock or cheapo GPS board for the Pi I keep attached to the local mesh, you can configure chrony or ntpd to use them as a reference. AREDN doesn't need time accurate to the millisecond.
I have also noticed that my TP Link CPE 510's seem to have a (typically wildly) inaccurate RTC built in. After a power cycle, they come back with a time off by a day or two rather than zero. I would not expect the AREDN software to go through much effort to sync a RTC on every device, given that most nodes are powered down by yanking the power cable.
I posted the initial concept and a couple follow-up posts on this (below) back in 2018 and it's still working well on our little remote mesh island. I am about to look at using the GPS USB unit approach for primary time and keep the old internet approach for a backup when the GPS hockey-puck gets eaten by a local fish eagle (or whatever):
https://www.arednmesh.org/content/how-make-time-your-mesh
HTH,
- Don / AA7AU