ATC700 stops sending records after receiving several queued GPRS commands

Hello, I encountered a problem with my Teltonika ATC700 tracker: it hangs after receiving several commands over Codec 12 and recovers only after a reboot.

Setup

Device: ATC700 (Hw:ATC700 Mod:21), FW 3.20.0-rc.1+7.exp.autonomous (the latest ATC700 build in the firmware changelog), modem FW EG915UEUABR03A01M08_01.215.01.215

Server: Traccar 6.16 over TCP, receiving records in Codec 8 Extended, sending commands in Codec 12

Problem

When the device connects and receives several queued Codec 12 commands at once, it answers only the first 2, reports a record count mismatch and goes to sleep with the socket open. From then on it saves records as configured (every 5 minutes in my case) but never sends them over the LTE, not even after the server closes the TCP connection. Only a restart helps. After the restart it sends all stored records, so no data is lost, but tracking is down until someone restarts the device by hand.

Steps to reproduce

I found it while reading all parameters with getparam:

  1. Device in Online Sleep (parameters 102:3, 1000:60, 10000:300).
  2. While it sleeps, queue 14 getparam commands on the server (each with about 15 parameter IDs, ~100 B).
  3. The device wakes up and connects. Traccar answers the IMEI with 01 and sends all queued commands right away (~1.7 kB).
  4. ATC700 answers only the first two and then hangs (without sending TCP FIN, so the TCP connection is kept open until the server timeout).

It happened 3 times out of 3. The device stayed silent for up to 2.4 hours until I restarted it. The same commands sent one at a time while the device is awake work fine, and 3 queued commands were also answered correctly.

Logs

Server log (Traccar, times are UTC = device log − 2 h)

I decoded the corresponding lines from the Traccar logs (and redacted the IMEI):

19:59:13 < 000f<IMEI>                login
19:59:13 > 01                        IMEI ack
19:59:13 > 00000000000000670c0105…   Codec 12 command, 115 B: getparam 12150;10050;10150;…
19:59:13 > 00000000000000610c0105…   Codec 12 command, 109 B: getparam 106;10993;1002;1000;…
19:59:13 > 000000000000005e0c0105…   Codec 12 command, 106 B: getparam 260002;13087;2025;…
           … 10 more commands, 104–148 B each …
19:59:13 > 00000000000000480c0105…   Codec 12 command,  84 B: getparam 1008001;1008004;…
19:59:14 < 00000000000000910c0106…   Codec 12 response, 157 B: Param ID:12150 Value:1;10050:30;…
19:59:14 < 00000000000000528e0100…   Codec 8E, 1 record, 94 B
         + 000000000000009e0c0106…   Codec 12 response, 170 B: Param ID:106 Value:1;10993:0;…
19:59:14 > 00000001                  record ack
20:29:14   timed out, disconnected   (no data from the device for 30 min)

All 14 commands (1 742 B) were sent within the same millisecond, right after the IMEI ack. The server acknowledged the record correctly with 00000001, but the device never read it. The other 12 commands were never answered.

Device log (from the Dump, times are local CEST)

I got the Dump through TCT. The device reads the IMEI ack together with the queued commands, processes 2 of them, and then reads 4 bytes of command text as the record acknowledgement (0x30 is the ASCII 0):

21:59:13 [M:ATCMD] [1] >> +QIRD: 1024
21:59:13 [M:SOCKET] Received data: ... 0x01 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x67 0x0C 0x01 0x05 ...
21:59:13 [M:GPRS_CMD] Unpacked data: getparam 12150;10050;10150;...
21:59:13 [M:ATCMD] [2] >> +QIRD: 718
21:59:13 [M:ATCMD] [1] << AT+QISEND=0,94
21:59:13 [M:GPRS_CMD] Unpacked data: getparam 106;10993;1002;1000;107;...
21:59:14 [M:RECORDS__TCP] Record count mismatch! Transmitted: 0x01, Confirmed: 0x30
21:59:14 [M:RECORDS__SENDER] Send failed! Err: 6
21:59:14 [M:RECORDS__MNG] Sending finished with code: 7, Number of records sent: 0
21:59:14 [M:ATCMD] [0] >> +QIURC: "recv",0
21:59:15 [M:SLEEP_MNG__SEND_IN_SLEEP] Send in sleep ended

The last +QIURC: "recv" is the server’s real acknowledgement (00000001), which is never read. After that the log shows only Record Saved! every 5 minutes. When the server closed the connection at 22:29:14, no +QIURC: "closed" was logged and the device didn’t try to reconnect.

Dumps covering all 3 hangs, the configuration and the server-side log are available. I can send them privately.

Expected

  • The device should split incoming data into packets, process all received commands and read the real record acknowledgement.
  • After a failed send it should retry or reconnect instead of staying silent until a restart.

Good Day,

Thank you for providing a detailed explanation of the issue.

To investigate further, we will need to analyze the device logs and determine why the device displays the “Sending finished” message without transmitting any records.

Kindly contact your sales manager, who will assist you with creating a Partner Portal ticket so that the issue can be investigated in detail.

Thanks, and best regards,
Imran S.