Timestamp discrepancy

I’ve noticed an odd discrepancy with the reported time of received packets, screen shot of ChirpStack’s device -> raw data tab attached. Note the rxInfo->time is …13:38:36…
BTW I have “send

gateway meta-data” turned on, my gateway has GPS and the system time is correct. Host is an RPi with RAK2287 concentrator, otherwise working perfectly.
Weird thing is that the packet was actually received at 14:34:09 as per ChirpStack’s own window caption 2:34:09 PM (note we’re in DST here so would expect to see 1Hr difference)
The data arrival time-stamp is thus 4.5 minutes into the future.
Can anyone explain?
Thanks a lot.

all time stamps in ChirpStack is UTC

Is it consistently 4.5 minutes, or does it keep changing? I have an issue where my rx timestamps get progressively further into the future, but up to a maximum of just one minute, and then the clock corrects itself, only for the problem to repeat itself. This is also on a RPI, but not with a RAK2287. I believe the rxTimestamp is coming from the packet forwarder. I don’t have the PPS feed connected from the GPS module, which I think may be related to this somehow.

1 Like

Hi James, thanks for responding. The 4.5 minutes seemed stable over a few days - but I do have GPS PPS connected. Note also that the RAK2287 uses the new Semtech 1302 chip-set. I just tried a shut-down & power-cycle on the RPi host and concentrator: at first the time discrepancy was +02:06s, but then magically disappeared (to -1s) after a few minutes. Perhaps there is some kind of initialisation race between the 1302+pkt-fwder and the SNTP on the RPi?
BTW I know time-stamps are in UTC - of course.

RxInfo time you see in the logsussually be one of these:

  1. Time of system on which chirpstack-gateway-bridge is running if no (GPS) time is provided by “packet-forwarder” (and fake_rx_time=true)

  2. GPS (or other) time, provided by “packet-forwarder”

Martin, thanks. I don’t have fake_rx_time set to true and I do have working GPS with PPS routed to the concentrator and serial data routed to the Pi’s UART pins - yet the rxInfo time was persistently 4.5 minutes fast (for several days until a reboot). Looks like a bug to me :slight_smile:

I’ve just solved my issue by connecting the PPS feed, so I suspect the two problems are unrelated.

I reckon connecting your PPS will have stopped your time-drift issue. My time-shift went away after a reboot but I think there’s an underlying issue that only shows itself under some very specific condition at boot-up.
Thanks for your input, glad you’ve solved your problem.