About one month ago my Cerbo GX (fw 3.66) was prepared for the initial installation on a temporary DHCP network. Time was set automatically, timezone was set, everything was good! Then I moved it to the permanent location and entered a static IP address for the Victron network.
One month later… the local time on my Cerbo is off by about 90 seconds. Running tcpdump on my network router for a while I see zero NTP packets coming from the Cerbo, so it’s not even trying to do anything with NTP. Checking that over root/SSH console:
root@einstein:~# connmanctl clock
Time = 1762769358
TimeUpdates = auto
Timezone = (null)/Universal
TimezoneUpdates = auto
Timeservers = [ ]
… confirms my suspicion: there is no NTP server set. Well, where do I configure NTP now?
I really would like to avoid using DHCP on this small separate Victron-network. The Cerbo settings screens don’t show any field to enter this:
Please tell me I’m missing something or that there’s a hidden entry somewhere to configure NTP without DHCP and without modifying root filesystem.
N.B.:
Although I cannot set NTP in the EVCS, it does sync the time on that one just fine with a random NTP server on the internet in case of a static IP network and also reports a successful timesync with “source: SNTP”. Ideally I would like to set a manual NTP server there, but at least it’s syncing with some server. The problem is just the Cerbo GX, on the same network/LAN, which does not do anything, it seems.
No, I’m not using WiFi, it’s disabled. There’s only one IP connection to the Cerbo which is this Ethernet/LAN network designated for Victron equipment (Cerbo GX, EVCS, 2x energy meter).
Timeservers = [ ] remains empty, BUT it started to send NTP packets to the first timeserver specified, my NTP server replied and the time was set. FINALLY my Cerbo has a correct time!
I hope you understand I am very much disappointed in this configuration experience. I hope this will improve in the future. Please, Victron, this makes me facepalm right now.
Oh god, now this popped up, only for setting NTP servers, REALLY?!
I have the same problem. A reboot fixes the time once. This has been costing me real money with charging drifting into the expensive periods.
Timeservers = [ ] remains empty, BUT it started to send NTP packets to the first timeserver specified, my NTP server replied and the time was set. FINALLY my Cerbo has a correct time!
I use a plain Linux machine as my router where all traffic passes. There I run tcpdump on the Victron’s interface and I can see everything crossing the boundaries of the LAN. (If I wanted to, I could go deeper I can even run a port-mirror on the ethernet switch.)
The actual tcpdump command that I run on my router:
$ tcpdump -i victron -n -v 'host 10.50.6.10 and udp port 123'
(where victron is my interface name for the Victron VLAN and 10.50.6.10 is the Cerbo IP.
It only attempts to synchronize a few times per day here, so the output is pretty quiet.
I would have loved to see a proper ntpd running on the Cerbo rather than a periodic short NTP call.
Ah, thanks. I shall check that out on mine. I’ve been tearing my hair out as this has happened before. DNS works, DHCP set up for NTP. I’ll see if I can see NTP calls on my unify.
opkg wasn’t working. An install of the latest firmware fixed that.
Run the following commands from the terminal after ssh in.
opkg update
opkg install ntp
opting install ntpq
run this. minpoll 12 maxpoll 12 changes polling to once per hour. That has stopped the continuous polling.
cat > /etc/ntp.conf <<EOF
# Use standard pool
pool pool.ntp.org iburst minpoll 12 maxpoll 12
pool 0.europe.pool.ntp.org iburst minpoll 12 maxpoll 12
pool 1.europe.pool.ntp.org iburst minpoll 12 maxpoll 12
# Local clock fallback, but very low preference
server 127.127.1.0
fudge 127.127.1.0 stratum 14
driftfile /var/lib/ntp/ntp.drift
EOF
killall ntpd
ntpd -gq
ntpq -p
ntpq -p
This should show something similar to
remote refid st t when poll reach delay offset jitter
==============================================================================
pool.ntp.org .POOL. 16 p - 64 0 0.000 +0.000 0.001
0.europe.pool.n .POOL. 16 p - 64 0 0.000 +0.000 0.001
1.europe.pool.n .POOL. 16 p - 64 0 0.000 +0.000 0.001
LOCAL(0) .LOCL. 14 l 1 64 1 0.000 +0.000 0.001
178-79-150-226. 82.68.30.35 3 u 1 64 1 45.550 -372.55 0.001
Before running these, all I could see was
LOCAL(0) .LOCL. 14 l 1 64 1 0.000 +0.000 0.001
I’ll have to wait a while to see if it drifts now.
