ntpMerlin ntpMerlin - NTP Daemon for AsusWRT Merlin

[JGrana]

Not sure, my network is all ipv4

 

[Vexira]

Not sure, my network is all ipv4

Same here but I have devices that look for IPv6 time servers.

Also wanted to ask how does it compare to ntpd in general accuracy?

 

[JGrana]

Same here but I have devices that look for IPv6 time servers.

Also wanted to ask how does it compare to ntpd in general accuracy?

Its been working very well for me - I also have a chrony timeserver using a GPS module on an RPi.
On my AX88U, my graphs typically show very low jitter and offset.

It is also now the default NTP server/client on Red Hat and SUSE Linux - and Facebook has converted their entire network to use chrony.

 

[Vexira]

Its been working very well for me - I also have a chrony timeserver using a GPS module on an RPi.
On my AX88U, my graphs typically show very low jitter and offset.

It is also now the default NTP server/client on Red Hat and SUSE Linux - and Facebook has converted their entire network to use chrony.

Sound awesome I can't wait till windows and all the other Linux distributions make it default as well as the router firmware.

 

[nephilim]

I need some guidance, please. I have an Ai-Mesh with the master node running 384.19 and (amongst a few scripts) ntpMerlin 2.5.1. An N66 connects as repeater (running John's LTS fork and ntpMerlin, too). One of the clients in my network is a DIY fine-dust sensor based on a NodeMCU ESP8266. This devices connects via the N66 to the network and tries to get time updates from pool.ntp.org. On the N66 I have switched off the ntp traffic redirect. What puzzles me is that the NodeMCU cannot reach the pool.ntp.org servers when ntp traffic redirect is activated on the master node. I need to switch it off - otherwise the sensor will not receive updates. I see the port forwarding rule when traffic redirect is on. Any ideas?

 

[Wade Coxon]

I took this for a test drive earlier in the week, and things seemed to be working well. All graphs populating and good time stability.
However, looking at the router logs, I am seeing frequent 7z crash messages (one every hour or two). Uninstalled the script and no more crashes.

Anyone have an idea what was happening there? I presume the script is periodically archiving data, and something is going wrong?

 

[Jack Yaz]

I took this for a test drive earlier in the week, and things seemed to be working well. All graphs populating and good time stability.
However, looking at the router logs, I am seeing frequent 7z crash messages (one every hour or two). Uninstalled the script and no more crashes.

Anyone have an idea what was happening there? I presume the script is periodically archiving data, and something is going wrong?

Can you share the actual errors? Kinda hard to troubleshoot without...

 

[Wade Coxon]

Yep, sorry, good point. I thought I would just ask in case it was a common issue.
Dozens like this that are apparently occurring at scheduled (round) times of day.

Aug 27 21:00:00 kernel: pgd = ffffffc00a11c000
Aug 27 21:00:00 kernel: [00000008] *pgd=0000000014334003, *pud=0000000014334003, *pmd=0000000000000000
Aug 27 21:00:00 kernel: CPU: 0 PID: 11950 Comm: 7z Tainted: P           O    4.1.27 #2
Aug 27 21:00:00 kernel: Hardware name: Broadcom-v8A (DT)
Aug 27 21:00:00 kernel: task: ffffffc0191d94c0 ti: ffffffc00b968000 task.ti: ffffffc00b968000
Aug 27 21:00:00 kernel: PC is at 0x7f8180b8b0
Aug 27 21:00:00 kernel: LR is at 0x7f8180b898
Aug 27 21:00:00 kernel: pc : [<0000007f8180b8b0>] lr : [<0000007f8180b898>] pstate: 60000000
Aug 27 21:00:00 kernel: sp : 0000007f7d1dc8d0
Aug 27 21:00:00 kernel: x29: 0000007f7d1dc8d0 x28: 0000007f81a98000
Aug 27 21:00:00 kernel: x27: 000000000e781c10 x26: 0000007f81a98000
Aug 27 21:00:00 kernel: x25: 0000007f81a94000 x24: 000000000e7814d8
Aug 27 21:00:01 kernel: x23: 0000000000000000 x22: 000000000e781638
Aug 27 21:00:01 kernel: x21: 000000000e778f20 x20: 000000000e781c10
Aug 27 21:00:01 kernel: x19: 000000000e781590 x18: 000000005f473dd0
Aug 27 21:00:01 kernel: x17: 0000007f81972750 x16: 0000000000491308
Aug 27 21:00:01 kernel: x15: 0000007f81a661a8 x14: 0000000000000002
Aug 27 21:00:01 kernel: x13: 0000000000000150 x12: 0000007f7d1dd1d0
Aug 27 21:00:01 kernel: x11: 00000000003d0f00 x10: 0000007f81a71b80
Aug 27 21:00:01 kernel: x9 : 000000000000270f x8 : 00000000000000de
Aug 27 21:00:01 kernel: x7 : 0000000000000069 x6 : 0000000000000000
Aug 27 21:00:01 kernel: x5 : 0000000000000000 x4 : 0000000000000000
Aug 27 21:00:01 kernel: x3 : 000000000000000c x2 : 0000007f7d1dd8c0
Aug 27 21:00:01 kernel: x1 : 0000000000000001 x0 : 0000007f818e4318

 

[Vexira]

Its been working very well for me - I also have a chrony timeserver using a GPS module on an RPi.
On my AX88U, my graphs typically show very low jitter and offset.

It is also now the default NTP server/client on Red Hat and SUSE Linux - and Facebook has converted their entire network to use chrony.

Do you have any instructions on how to set it up with the current NTP Merlin?

 

[JGrana]

The chrony package is in the latest Entware. The conf file (chrony.conf) is very similar to ntp.conf.
I am oot at the moment, I can give more details later this week.

 

[Wade Coxon]

Can you share the actual errors? Kinda hard to troubleshoot without...

I thought that this problem may have also been a result of me not formatting the jffs partition following my 384.19 update.
I hadn't formatted since I wasn't experiencing any issues, and my partition seemed to be intact.
Anyway, I formatted and reinstalled ntpMerlin, and no change, the 7z crashes started up again.

I have turned ntpMerlin off in the meantime. It does appear to work and the graphs are generated just fine, but I am not keen on seeing all those crashes in my logs.

 

[DonCatoli]

I get the following error when I attempt to install...am I missing something?

Checking ntpd... dead.
S77ntpd: Waiting for NTP to sync before starting...
Starting ntpd... failed.
ntpq: error while loading shared libraries: libcrypto.so.1.1: cannot open shared object file: No such file or directory

 

[Wade Coxon]

I get the following error when I attempt to install...am I missing something?

I found one little bug with the install procedure on my second installation attempt. This may or may not have affected you.

I had already enabled the standard NTP server in the web interface prior to installing netMerlin.
As part of the install process, the ntpMerlin install script will disable the pre-installed NTP server.
Unfortnately, it looks like disabling the ntp server, has the undesirable side effect of restarting the router's ssh server, which boots you out of the script.

I found that I needed to uninstall ntpMerlin, then re-install it, and the second install went well (since it no longer needed to disable the stock ntp server this time).

 

[DonCatoli]

I found one little bug with the install procedure on my second installation attempt. This may or may not have affected you.

I had already enabled the standard NTP server in the web interface prior to installing netMerlin.
As part of the install process, the ntpMerlin install script will disable the pre-installed NTP server.
Unfortnately, it looks like disabling the ntp server, has the undesirable side effect of restarting the router's ssh server, which boots you out of the script.

I found that I needed to uninstall ntpMerlin, then re-install it, and the second install went well (since it no longer needed to disable the stock ntp server this time).

My NTP server was off as I never enabled it.

 

[EmeraldDeer]

The chrony package is in the latest Entware. The conf file (chrony.conf) is very similar to ntp.conf.
I am oot at the moment, I can give more details later this week.

Just thinking aloud here...what if:
Entware build supported interleaved Ethernet hardware timestamping
Router ethernet supported receiving hardware timestamping
Raspberry Pi ethernet supported sending hardware timestamping

Then perhaps the router clock could reach microsecond level offsets from GPS as opposed to millisecond.

 

[Jack Yaz]

Following an upgrade to ntpd from Entware, you'll need to use option uf in ntpmerlin to correct the S77ntpd file

 

[aslak]

I'm getting these errors upon reinstalling ntpmerlin, the same show up if I run the update graphs script. uf command doesn't resolve it for me.

 

[Linux_Chemist]

The chrony package is in the latest Entware. The conf file (chrony.conf) is very similar to ntp.conf.
I am oot at the moment, I can give more details later this week.

I would love to hear more about Chrony - how much setup would it require beyond writing a conf? I assume it just uses port 123 on the firewall but would require a lot of the plumbing ntpmerlin has?

 

[Vexira]

I would love to hear more about Chrony - how much setup would it require beyond writing a conf? I assume it just uses port 123 on the firewall but would require a lot of the plumbing ntpmerlin has?

im still waiting also im really curious about how much difference it has compared to ntpd

 

[Jack Yaz]

I'm getting these errors upon reinstalling ntpmerlin, the same show up if I run the update graphs script. uf command doesn't resolve it for me.

View attachment 26208

opkg update && opkg remove --autoremove sqlite3-cli && opkg install sqlite3-cli