ntpMerlin ntpMerlin - NTP Daemon for AsusWRT Merlin

[EmeraldDeer]

Done. @EmeraldDeer the raw stats can be found in /tmp/ntp-rrdstats.$$ where $$ is substituted for a process ID

Downloaded updated v1.2.1 and watching the sys_jitter goose eggs go by

 

[JGrana]

Downloaded updated v1.2.1 and watching the sys_jitter goose eggs go by

Can you folks run ntpq -p? jitter is the last column.

 

[Butterfly Bones]

Can you folks run ntpq -p? jitter is the last column.

Using only us.pool.ntp.org.

     remote           refid      st t when poll reach   delay   offset  jitter
============================================================================
==*linode1.ernest- 198.72.72.10     3 u   63  128  377   91.190   -0.713   1.666

 

[RMerlin]

Personally I did not even know the Asus firmware was running an ntp server.

It doesn't, only a client.

 

[XIII]

It doesn't, only a client.

Oops... Stupid mistake from my side

How long should it take for stuff to settle? Isn't this taking longer than usual?

 

[JGrana]

Using only us.pool.ntp.org.

     remote           refid      st t when poll reach   delay   offset  jitter
============================================================================
==*linode1.ernest- 198.72.72.10     3 u   63  128  377   91.190   -0.713   1.666

Hmmm, shows relatively normal jitter. Not sure why it doesn't show up on your graph. One idea (a stretch), add another server to ntp.conf. Maybe:

server 0.pool.ntp.org iburst

Then restart ntpd. Give it a few hours and see...

 

[EmeraldDeer]

Hmmm, shows relatively normal jitter. Not sure why it doesn't show up on your graph. One idea (a stretch), add another server to ntp.conf. Maybe:

server 0.pool.ntp.org iburst

Then restart ntpd. Give it a few hours and see...

The graph is not using "ntpq -p" for data, it is using "ntpq -4 -c rv".

Not a stretch, it has been established that sys_jitter is zero for one server but not for more than one server.

Here is more one server data:

# ntpq -4 -c rv | egrep "offset|jitter"
mintc=3, offset=0.019588, frequency=1.186, sys_jitter=0.000000,
clk_jitter=0.071, clk_wander=0.051, tai=37, leapsec=201701010000,

# grep jitter /tmp/ntp-rrdstats*
/tmp/ntp-rrdstats.1313: sys_jitter=0.000000
/tmp/ntp-rrdstats.1313:clk_jitter=0.183
/tmp/ntp-rrdstats.1828: sys_jitter=0.000000
/tmp/ntp-rrdstats.1828:clk_jitter=0.149
/tmp/ntp-rrdstats.2164: sys_jitter=0.000000
/tmp/ntp-rrdstats.2164:clk_jitter=0.121
/tmp/ntp-rrdstats.2487: sys_jitter=0.000000
/tmp/ntp-rrdstats.2487:clk_jitter=0.088
/tmp/ntp-rrdstats.2994: sys_jitter=0.000000
/tmp/ntp-rrdstats.2994:clk_jitter=0.075
/tmp/ntp-rrdstats.3311: sys_jitter=0.000000
/tmp/ntp-rrdstats.3311:clk_jitter=0.065
/tmp/ntp-rrdstats.3642: sys_jitter=0.000000
/tmp/ntp-rrdstats.3642:clk_jitter=0.060
/tmp/ntp-rrdstats.4190: sys_jitter=0.000000
/tmp/ntp-rrdstats.4190:clk_jitter=0.188
/tmp/ntp-rrdstats.4520: sys_jitter=0.000000
/tmp/ntp-rrdstats.4520:clk_jitter=0.152
/tmp/ntp-rrdstats.4857: sys_jitter=0.000000
/tmp/ntp-rrdstats.4857:clk_jitter=0.119
/tmp/ntp-rrdstats.5348: sys_jitter=0.000000
/tmp/ntp-rrdstats.5348:clk_jitter=0.098
/tmp/ntp-rrdstats.5666: sys_jitter=0.000000
/tmp/ntp-rrdstats.5666:clk_jitter=0.085
/tmp/ntp-rrdstats.5988: sys_jitter=0.000000
/tmp/ntp-rrdstats.5988:clk_jitter=0.101
/tmp/ntp-rrdstats.6306: sys_jitter=0.000000
/tmp/ntp-rrdstats.6306:clk_jitter=0.183
/tmp/ntp-rrdstats.6630: sys_jitter=0.000000
/tmp/ntp-rrdstats.6630:clk_jitter=0.135
/tmp/ntp-rrdstats.7116: sys_jitter=0.000000
/tmp/ntp-rrdstats.7116:clk_jitter=0.097
/tmp/ntp-rrdstats.7520: sys_jitter=0.000000
/tmp/ntp-rrdstats.7520:clk_jitter=0.080
/tmp/ntp-rrdstats.7887: sys_jitter=0.000000
/tmp/ntp-rrdstats.7887:clk_jitter=0.073
/tmp/ntp-rrdstats.826: sys_jitter=0.000000
/tmp/ntp-rrdstats.826:clk_jitter=0.232
/tmp/ntp-rrdstats.958: sys_jitter=0.000000
/tmp/ntp-rrdstats.958:clk_jitter=0.218

 

[EmeraldDeer]

Oops... Stupid mistake from my side

How long should it take for stuff to settle? Isn't this taking longer than usual?
View attachment 16845 View attachment 16846

If your clock gets knocked off every two hours, then the culprit is the ntp client.
killall ntp

 

[XIII]

If your clock gets knocked off every two hours, then the culprit is the ntp client.
killall ntp

I'm not sure that's running:

ps | grep ntp
...   ntpd -c /jffs/configs/ntp.conf -g

No "ntp". And isn't "killall ntp" executed by /opt/etc/init.d/S77ntpd?

PRECMD="killall ntp"

 

[XIII]

Oops: looks like I removed the iburst directive in the config file when editing the servers...

 

[EmeraldDeer]

I'm not sure that's running:

ps | grep ntp
...   ntpd -c /jffs/configs/ntp.conf -g

No "ntp". And isn't "killall ntp" executed by /opt/etc/init.d/S77ntpd?

PRECMD="killall ntp"

S77ntpd does stop the ntp client. But I have noticed when making router configuration changes that it can be started.

So I take it that your anomaly at about 18:20 UTC was that you restarted the ntp server without the iburst.

 

[XIII]

So I take it that your anomaly at about 18:20 UTC was that you restarted the ntp server without the iburst.

Oh, should have indeed explained that. 18:20 was actually a reboot of the router (for an unrelated issue with a device).

Will check again tomorrow.

 

[doczenith1]

I just installed ntpMerlin, rebooted the router and upon selecting "1. Generate updated ntpMerlin graphs now" I'm getting this.

Choose an option: 1

ntpq: error while loading shared libraries: libcrypto.so.1.1: wrong ELF class: E LFCLASS32
Press enter to continue...

Moments before I installed ntpMerlin I did Diversion>pixelsrv-tls>manage Entware and update installed packages.

I see the mention to Run opkg update;opkg upgrade in a search of this thread but I am leery to do this as I remember reading a post not to long ago where someone ran that and it screwed up their Diversion/pixelserv-tls install. Perhaps this is a different scenario?

 

[L&LD]

I just installed ntpMerlin, rebooted the router and upon selecting "1. Generate updated ntpMerlin graphs now" I'm getting this.

Choose an option: 1

ntpq: error while loading shared libraries: libcrypto.so.1.1: wrong ELF class: E LFCLASS32
Press enter to continue...

Moments before I installed ntpMerlin I did Diversion>pixelsrv-tls>manage Entware and update installed packages.

I see the mention to Run opkg update;opkg upgrade in a search of this thread but I am leery to do this as I remember reading a post not to long ago where someone ran that and it screwed up their Diversion/pixelserv-tls install. Perhaps this is a different scenario?

You need to update them after installing ntpMerlin.

https://www.snbforums.com/threads/ntpmerlin-installer-for-kvic-ntp-daemon.55756/page-2#post-474769

 

[doczenith1]

I just did option 6. update and upgrade installed packages and when choosing option 1 I now get this error. Is there a reboot needed?

Choose an option: 1

ntpq: read: Connection refused
Press enter to continue...

 

[L&LD]

I just did option 6. update and upgrade installed packages and when choosing option 1 I now get this error. Is there a reboot needed?

Choose an option: 1

ntpq: read: Connection refused
Press enter to continue...

It couldn't hurt. But maybe it just needs some time for everything to settle?

 

[Jack Yaz]

I just installed ntpMerlin, rebooted the router and upon selecting "1. Generate updated ntpMerlin graphs now" I'm getting this.

Choose an option: 1

ntpq: error while loading shared libraries: libcrypto.so.1.1: wrong ELF class: E LFCLASS32
Press enter to continue...

Moments before I installed ntpMerlin I did Diversion>pixelsrv-tls>manage Entware and update installed packages.

I see the mention to Run opkg update;opkg upgrade in a search of this thread but I am leery to do this as I remember reading a post not to long ago where someone ran that and it screwed up their Diversion/pixelserv-tls install. Perhaps this is a different scenario?

I'm not convinced entware is in a good state after the last repo update

 

[JGrana]

I'm not convinced entware is in a good state after the last repo update

I agree. After doing an opkg update && opkg upgrade I later tried running adb (Android Debugger). It complained about a missing lib. I use adb every few days...
I did another update/upgrade. Never saw adb getting updated.
I fixed it buy uninstalling adb, then re-install.
Ran fine after that...

 

[RMerlin]

I'm not convinced entware is in a good state after the last repo update

Have to upgrade already installed packages, to move to the newer OpenSSL.

opkg update
opkg upgrade

 

[Jack Yaz]

Have to upgrade already installed packages, to move to the newer OpenSSL.

opkg update
opkg upgrade

There's been some reports that didn't help, and some people have had to remove and reinstall for the dependencies to update. I think this is a known issue - opkg isn't smart enough to always update dependencies