StarLink as source for NTP
-
Looking at some pcaps I can see no difference at all between queries from ntpd when initialises and those from ntpdate.
That's against an ntpd server. Check a pcap against the starlink disk when using ntpdate. See if it's doing something odd. I'd be very surprised if it is though....
-
@stephenw10 said in StarLink as source for NTP:
No it's not in our repo. Yet.
Well... Hopefully, "Yet" is a good thing.
I'd like to get this working, but don't need it "now".It looks like all the Linux distros are dropping or have dropped ntpdate, So I guess it's an inevitable shift to Chrony or systemd-timesyncd. I'm assuming there is no fundamental difference between nptdate for Linux and FreeBSD?
Rick
-
Seems odd that chrony would work, but not the latest version of ntp, I show 25.11 using the latest ntp that I am aware of 4.2.8p18
Would be curious to see the pcap of send from chrony vs one from ntp.. normal ntp can for sure query chrony.. I run chrony myself on my pi ntp server. but most of my clients are just running normal ntp client.
-
@dennypage said in StarLink as source for NTP:
I use Chrony everywhere but pfSense.
Same, as it is far more accurate at tracking the references time.
As a result I do not use pfsence as a time server.
However I would love to see pfsense move to chrony as pfense is well placed to provide time to all devices on my network. -
Do you have POWERD enabled?

I have it disabled and GPS+PPS disciplined NTP server is accurate on a Protectli FW4C
-
@johnpoz said in StarLink as source for NTP:
Seems odd that chrony would work, but not the latest version of ntp,
Yup. Even more odd that it responds to an ntp query from ntpdate but not queries from ntpd. hey are essentially the same as far as I can see.
-
NTPDATE and CHRONY both use random source ports, whereas NTPD uses source port 123
Also, does Starlink support iburst? If not, it might stop responding or send KOD. PCAPs would let you know.
-
Hmm, a clue perhaps! Some state issue in the starlink router then maybe.
Though that isn't the case in pfSense(FreeBSD?). If you run ntpdate it will fail if ntpd is running because it tries to use the same source port. But only if you try to actually update the system clock. If you run with query only or debug it uses a non-privileged port and succeeds:
[25.11-RELEASE][admin@5100.stevew.lan]/root: ntpdate 172.21.16.1 5 Jan 14:15:46 ntpdate[49624]: the NTP socket is in use, exiting [25.11-RELEASE][admin@5100.stevew.lan]/root: ntpdate -q 172.21.16.1 server 172.21.16.1, stratum 3, offset -0.004204, delay 0.02602 5 Jan 14:15:51 ntpdate[51456]: adjust time server 172.21.16.1 offset -0.004204 sec@Mission-Ghost Can you test that after stopping ntpd? That would prove it if it's a port issue.
-
Even if it objects to iburst it should still respond to the first query. As understand it a pcap shows no replies at all. Also I think Chrony also uses iburst by default. But that could potentially be an issue I agree.
-
@elvisimprsntr I don’t use powerd.
-
@ stephenw10
In my stable baseline boot environment I’d been running before experimenting with this problem:
NTPd off:
ntpdate 192.168.100.1
Jan 08:49:13 ntpdate[28279]: adjust time server 192.168.100.1 offset -0.000721 secntpdate -q 192.168.100.1
server 192.168.100.1, stratum 1, offset -0.000321, delay 0.02838
5 Jan 08:54:09 ntpdate[41177]: adjust time server 192.168.100.1 offset -0.000321 secntpd on:
ntpdate 192.168.100.1
5 Jan 08:51:22 ntpdate[54373]: adjust time server 192.168.100.1 offset +0.000147 secntpdate -q 192.168.100.1
server 192.168.100.1, stratum 1, offset +0.000475, delay 0.03049
5 Jan 08:52:13 ntpdate[15425]: adjust time server 192.168.100.1 offset +0.000475 secNow just booted in my experimental boot environment I’d been using to work on this problem:
Ntpd on:
ntpdate 192.168.100.1
5 Jan 09:03:26 ntpdate[39431]: the NTP socket is in use, exitingntpdate -q 192.168.100.1
5 Jan 09:05:17 ntpdate[23895]: no server suitable for synchronization foundNtpd off:
ntpdate 192.168.100.1
5 Jan 09:07:38 ntpdate[70359]: sendto(192.168.100.1): Permission denied
5 Jan 09:07:40 ntpdate[70359]: sendto(192.168.100.1): Permission denied
5 Jan 09:07:42 ntpdate[70359]: sendto(192.168.100.1): Permission denied
5 Jan 09:07:44 ntpdate[70359]: sendto(192.168.100.1): Permission denied
5 Jan 09:07:46 ntpdate[70359]: no server suitable for synchronization foundntpdate -q 192.168.100.1
5 Jan 09:06:45 ntpdate[36712]: no server suitable for synchronization foundIt appears my experiments made things worse. I hope this gives you something useful though.
-
In my stable baseline boot environment I’d been running before experimenting with this problem:
NTPd off:
ntpdate 192.168.100.1
Jan 08:49:13 ntpdate[28279]: adjust time server 192.168.100.1 offset -0.000721 secntpdate -q 192.168.100.1
server 192.168.100.1, stratum 1, offset -0.000321, delay 0.02838
5 Jan 08:54:09 ntpdate[41177]: adjust time server 192.168.100.1 offset -0.000321 secntpd on:
ntpdate 192.168.100.1
5 Jan 08:51:22 ntpdate[54373]: adjust time server 192.168.100.1 offset +0.000147 secntpdate -q 192.168.100.1
server 192.168.100.1, stratum 1, offset +0.000475, delay 0.03049
5 Jan 08:52:13 ntpdate[15425]: adjust time server 192.168.100.1 offset +0.000475 secNow just booted in my experimental boot environment I’d been using to work on this problem:
Ntpd on:
ntpdate 192.168.100.1
5 Jan 09:03:26 ntpdate[39431]: the NTP socket is in use, exitingntpdate -q 192.168.100.1
5 Jan 09:05:17 ntpdate[23895]: no server suitable for synchronization foundNtpd off:
ntpdate 192.168.100.1
5 Jan 09:07:38 ntpdate[70359]: sendto(192.168.100.1): Permission denied
5 Jan 09:07:40 ntpdate[70359]: sendto(192.168.100.1): Permission denied
5 Jan 09:07:42 ntpdate[70359]: sendto(192.168.100.1): Permission denied
5 Jan 09:07:44 ntpdate[70359]: sendto(192.168.100.1): Permission denied
5 Jan 09:07:46 ntpdate[70359]: no server suitable for synchronization foundntpdate -q 192.168.100.1
5 Jan 09:06:45 ntpdate[36712]: no server suitable for synchronization foundIt appears my experiments made things worse. I hope this gives you something useful though.
-
Experimental boot environment:
ntpq -np 192.168.100
ntpd running;
192.168.100.1: timed out, nothing received
***Request timed outntpd stopped:
192.168.100.1: timed out, nothing received
***Request timed outStable baseline boot environment:
ntpd running:
5 Jan 09:27:31 ntpd[87125]: ntpd 4.2.8p18@1.4062-o Wed Aug 20 12:32:34 UTC 2025 (1): Starting
5 Jan 09:27:31 ntpd[87125]: Command line: ntpd -np 192.168.100.1
5 Jan 09:27:31 ntpd[87125]: ----------------------------------------------------
5 Jan 09:27:31 ntpd[87125]: ntp-4 is maintained by Network Time Foundation,
5 Jan 09:27:31 ntpd[87125]: Inc. (NTF), a non-profit 501(c)(3) public-benefit
5 Jan 09:27:31 ntpd[87125]: corporation. Support and training for ntp-4 are
5 Jan 09:27:31 ntpd[87125]: available at https://www.nwtime.org/support
5 Jan 09:27:31 ntpd[87125]: ----------------------------------------------------
5 Jan 09:27:31 ntpd[87125]: proto: precision = 0.083 usec (-23)
5 Jan 09:27:31 ntpd[87125]: getconfig: Couldn't open </etc/ntp.conf>: No such file or directory
5 Jan 09:27:31 ntpd[87125]: unable to bind to wildcard address :: - another process may be running - EXITINGntpd stopped:
It appears to start ntpd but with no servers.
-
If a client continues to bombard a NTP sever, the server will first send KOD packets, i.e. reply with invalid time, stratum 0, and request the client reduce its poll rate. At a certain point it will stop responding altogether. The only way to stop getting responses blocked is to stop bombarding the NTP server for several minutes.
The server sends a KoD packet with: Stratum = 0 Reference ID = "RATE"Packet captures is the only way to diagnose what is going on.
Diagnostics -> Packet Capture

-
@Mission-Ghost said in StarLink as source for NTP:
Ntpd off:
ntpdate 192.168.100.1
5 Jan 09:07:38 ntpdate[70359]: sendto(192.168.100.1): Permission deniedAh well that seems like a clue.
Do you see any blocked in the firewall log?
If not try running a pcap whilst testing that. Is the starlink router sending back the denied packet?
Are you running that as root/admin? It could just be a local user permissions issue that differs from the stable BE.
-
@luckman212 I am not an attorney, however, I have spent hundreds of hours working with several qualified IP and open source license attorneys and their external consultants. With this experience, the one thing I can say definitively about the GNU license is that there are a wide range of legal opinions as to what is allowed and safe for corporate entities with regard to the license. While there is consensus that option A is fully allowed and safe for a corporate entity, and there is consensus that option Z is not allowed. In between A & Z, attorneys have widely ranging opinions as to what is safe depending on their own background, their own risk tolerance, and the risk tolerance of their clients.
In short, it's risk vs. reward, and each corporate entity, Netgate included, has to make their own decision based on their own circumstance.
-
Yeah it's definitely a gray araa. I did some cursory searches, and found a few notable examples
All seem to ship Chrony with some of their products (use ⌘/⌃+F), so there's at least a precendent for it. I raised an issue on their GitLab, hope that isn't too annoying.
-
@stephenw10 perhaps, but that's an outlier.
In any case, I'm running the stable baseline environment now. I'll try to reproduce and report back, with a packet capture.
I didn't have logging enabled. More to come.
-
ntpdate 192.168.100.1 5 Jan 13:47:56 ntpdate[51331]: adjust time server 192.168.100.1 offset +0.002516 secPacket capture (with redactions as x):
13:47:50.437956 IP 100.117.x.x.123 > 192.168.100.1.123: UDP, length 48 13:47:50.438573 IP 192.168.100.1.123 > 100.117.x.x.123: UDP, length 48 13:47:52.443053 IP 100.117.x.x.123 > 192.168.100.1.123: UDP, length 48 13:47:52.444011 IP 192.168.100.1.123 > 100.117.x.x.123: UDP, length 48 13:47:54.432155 IP 100.117.x.x.123 > 192.168.100.1.123: UDP, length 48 13:47:54.435073 IP 192.168.100.1.123 > 100.117.x.x.123: UDP, length 48 13:47:56.444794 IP 100.117.x.x.123 > 192.168.100.1.123: UDP, length 48 13:47:56.445424 IP 192.168.100.1.123 > 100.117.x.x.123: UDP, length 48Added 192.168.100.1 back into NTP service as a server:

Redacted log records immediately after adding to NTP up to now at 13:55 (y.y = backup T-Mobile; x.x = Starlink):
17 Matched NTP Log Entries. (Maximum 250) Jan 5 13:52:38 ntpd 52576 192.168.100.1 8011 81 mobilize assoc 12535 Jan 5 09:37:26 ntpd 87856 Command line: ntpd -np 192.168.100.1 Jan 5 09:36:16 ntpd 64486 Command line: ntpd -np 192.168.100.1 Jan 5 09:29:06 ntpd 52330 Command line: ntpd -np 192.168.100.1 Jan 5 09:27:31 ntpd 87125 Command line: ntpd -np 192.168.100.1 Jan 5 09:27:23 ntpd 84605 Command line: ntpd -np 192.168.100.1 Jan 5 08:30:25 ntpd 67419 Command line: ntpd -q 192.168.100.1 Jan 4 15:18:17 ntpd 37124 192.168.100.1 local addr 100.117.x.x -> 192.168.100.2 Jan 4 15:11:49 ntpd 37124 192.168.100.1 local addr 192.168.y.y -> 100.117.x.x Jan 4 15:11:19 ntpd 37124 192.168.100.1 local addr 100.117.x.x -> 192.168.y.y Jan 4 09:28:29 ntpd 77534 192.168.100.1 8012 82 demobilize assoc 45084 Jan 4 08:51:39 ntpd 77534 192.168.100.1 8011 81 mobilize assoc 45084 Jan 3 23:52:54 ntpd 43025 192.168.100.1 local addr <null> -> 100.117.x.x Jan 3 23:52:54 ntpd 43025 192.168.100.1 local addr 192.168.100.2 -> <null> Jan 3 23:40:12 ntpd 57466 192.168.100.1 local addr 100.117.x.x -> 192.168.100.2 Jan 3 22:38:48 ntpd 80572 192.168.100.1 local addr <null> -> 10.10.10.1 Jan 3 22:38:48 ntpd 80572 Soliciting pool server 192.168.100.1Attempt to execute ntpd -np 192.168.100.1 manually from Diagnostics>Command:
ntpd -np 192.168.100.1 5 Jan 13:57:32 ntpd[49544]: ntpd 4.2.8p18@1.4062-o Wed Aug 20 12:32:34 UTC 2025 (1): Starting 5 Jan 13:57:32 ntpd[49544]: Command line: ntpd -np 192.168.100.1 5 Jan 13:57:32 ntpd[49544]: ---------------------------------------------------- 5 Jan 13:57:32 ntpd[49544]: ntp-4 is maintained by Network Time Foundation, 5 Jan 13:57:32 ntpd[49544]: Inc. (NTF), a non-profit 501(c)(3) public-benefit 5 Jan 13:57:32 ntpd[49544]: corporation. Support and training for ntp-4 are 5 Jan 13:57:32 ntpd[49544]: available at https://www.nwtime.org/support 5 Jan 13:57:32 ntpd[49544]: ---------------------------------------------------- 5 Jan 13:57:32 ntpd[49544]: proto: precision = 0.084 usec (-23) 5 Jan 13:57:32 ntpd[49544]: getconfig: Couldn't open </etc/ntp.conf>: No such file or directory 5 Jan 13:57:32 ntpd[49544]: unable to bind to wildcard address :: - another process may be running - EXITINGRestart NTP; Packet capture:
13:59:14.554093 IP 192.168.100.1.123 > 100.117.x.x.123: UDP, length 48 14:00:19.544090 IP 100.117.x.x.123 > 192.168.100.1.123: UDP, length 48 14:00:19.544640 IP 192.168.100.1.123 > 100.117.x.x.123: UDP, length 48 14:01:24.534167 IP 100.117.x.x.123 > 192.168.100.1.123: UDP, length 48 14:01:24.535605 IP 192.168.100.1.123 > 100.117.x.x.123: UDP, length 48 14:02:15.803032 IP 100.117.x.x.123 > 192.168.100.1.123: UDP, length 48 14:02:15.803483 IP 192.168.100.1.123 > 100.117.x.x.123: UDP, length 48 14:03:20.807447 IP 100.117.x.x.123 > 192.168.100.1.123: UDP, length 48 14:03:20.808365 IP 192.168.100.1.123 > 100.117.x.x.123: UDP, length 48Note: 100.117.x.x.123 is the Starlink WAN address.
Packet capture view option High continuing with NTP service including Starlink 192.168.100.1:
14:06:38.815239 IP (tos 0xb8, ttl 64, id 55851, offset 0, flags [none], proto UDP (17), length 76) 100.117.x.x.123 > 192.168.100.1.123: [bad udp cksum 0xc679 -> 0x120d!] NTPv4, Client, length 48 Leap indicator: (0), Stratum 2 (secondary reference), poll 6 (64s), precision -23 Root Delay: 0.060485, Root dispersion: 0.193786, Reference-ID: 0x2a14cae6 Reference Timestamp: 3976635747.879349730 (2026-01-05T21:02:27Z) Originator Timestamp: 3976635931.803164958 (2026-01-05T21:05:31Z) Receive Timestamp: 3976635931.804594510 (2026-01-05T21:05:31Z) Transmit Timestamp: 3976635998.815215859 (2026-01-05T21:06:38Z) Originator - Receive Timestamp: +0.001429551 Originator - Transmit Timestamp: +67.012050900 14:06:38.816555 IP (tos 0x48, ttl 64, id 7514, offset 0, flags [DF], proto UDP (17), length 76) 192.168.100.1.123 > 100.117.x.x.123: [udp sum ok] NTPv4, Server, length 48 Leap indicator: (0), Stratum 1 (primary reference), poll 0 (1s), precision -9 Root Delay: 0.000000, Root dispersion: 0.000000, Reference-ID: ^@^@^@^@ Reference Timestamp: 3976635998.815293311 (2026-01-05T21:06:38Z) Originator Timestamp: 3976635998.815215859 (2026-01-05T21:06:38Z) Receive Timestamp: 3976635998.814606666 (2026-01-05T21:06:38Z) Transmit Timestamp: 3976635998.815293311 (2026-01-05T21:06:38Z) Originator - Receive Timestamp: -0.000609192 Originator - Transmit Timestamp: +0.000077452 14:07:45.828008 IP (tos 0xb8, ttl 64, id 65173, offset 0, flags [none], proto UDP (17), length 76) 100.117.x.x.123 > 192.168.100.1.123: [bad udp cksum 0xc679 -> 0x7b6d!] NTPv4, Client, length 48 Leap indicator: (0), Stratum 2 (secondary reference), poll 6 (64s), precision -23 Root Delay: 0.060485, Root dispersion: 0.194793, Reference-ID: 0x2a14cae6 Reference Timestamp: 3976635747.879349730 (2026-01-05T21:02:27Z) Originator Timestamp: 3976635998.815293311 (2026-01-05T21:06:38Z) Receive Timestamp: 3976635998.816618714 (2026-01-05T21:06:38Z) Transmit Timestamp: 3976636065.827952147 (2026-01-05T21:07:45Z) Originator - Receive Timestamp: +0.001325402 Originator - Transmit Timestamp: +67.012658835 14:07:45.828606 IP (tos 0x48, ttl 64, id 4388, offset 0, flags [DF], proto UDP (17), length 76) 192.168.100.1.123 > 100.117.x.x.123: [udp sum ok] NTPv4, Server, length 48 Leap indicator: (0), Stratum 1 (primary reference), poll 0 (1s), precision -9 Root Delay: 0.000000, Root dispersion: 0.000000, Reference-ID: ^@^@^@^@ Reference Timestamp: 3976636065.827302932 (2026-01-05T21:07:45Z) Originator Timestamp: 3976636065.827952147 (2026-01-05T21:07:45Z) Receive Timestamp: 3976636065.827219962 (2026-01-05T21:07:45Z) Transmit Timestamp: 3976636065.827302932 (2026-01-05T21:07:45Z) Originator - Receive Timestamp: -0.000732184 Originator - Transmit Timestamp: -0.000649215So am I seeing Starlink is sending back the expected NTP records? Then, it seems NTP is not doing anything with them (reach is still 0 now at 14:09).
-
@Mission-Ghost Too short of a period for meaningful analysis. Reachability of 3 only indicates 2 interactions.
For diagnostic purposes, I would recommend shortening the maximum poll interval to 3 or 4. Also, wait until at least one source shows reachability of 377 before making any determinations.
Hello! It looks like you're interested in this conversation, but you don't have an account yet.
Getting fed up of having to scroll through the same posts each visit? When you register for an account, you'll always come back to exactly where you were before, and choose to be notified of new replies (either via email, or push notification). You'll also be able to save bookmarks and upvote posts to show your appreciation to other community members.
With your input, this post could be even better 💗
Register LoginPrivacy Policy · Cookie Policy