\# GL-X2000 Spitz Plus – Intermittent Internet connection loss / QMI timeouts (Quectel EG120K-EA)

Router Cellular, USB modems
1

GL-X2000 Spitz Plus – Intermittent Internet connection loss / QMI timeouts (Quectel EG120K-EA)

Hello,

I am experiencing intermittent Internet connection losses on my GL-X2000 Spitz Plus. I have carried out several investigations and would appreciate help from the GL.iNet team.

  1. Hardware and software

Component| Details
Router| GL.iNet GL-X2000 Spitz Plus
Modem| Quectel EG120K-EA
Modem firmware| EG120KEAAAR01A11M2G
USB ID| "2c7c:030b"
Router firmware| OpenWrt 19.07-SNAPSHOT
Kernel| 5.4.213
Build date| June 20, 2025
Modem driver| "qmi_wwan_q V1.2.7"
Connection mode| QMI / QMAP mode 1
QMI device| "/dev/cdc-wdm0"
Data interface| "wwan0_1"
Protocol| "qcm"
Carrier| Orange France
APN| "orange"
IP mode| IPv4 only

  1. Main symptoms

Several times a day, the router loses its Internet connection.

  • Wi-Fi remains available.
  • Devices stay connected to the router.
  • Internet access stops working.
  • The modem sometimes recovers automatically after several minutes.

On November 5, 2026, I personally observed an outage starting at approximately 22:17, with Internet access returning around 22:22.

Restarting only the modem connection from the router's web interface usually restores Internet access immediately.

  1. Investigations already performed

USB and hardware

  • Checked the USB device and its interfaces.
  • Verified that the modem remains detected during the investigated outages.
  • Analysed USB and XHCI logs.
  • Checked USB power management.
  • USB autosuspend is disabled ("power/control=on").
  • No physical USB disconnection or modem re-enumeration was observed in the "dmesg" logs corresponding to the November 5 outage.

The SanDisk SSD previously connected to the router has been unplugged and is not involved in the current tests.

Network and modem configuration

  • Tested different IPv6 settings.
  • Currently using IPv4 only.
  • Disabled automatic SIM switching and APN polling.
  • Disabled hardware acceleration during troubleshooting.
  • Checked the modem interface and its associated DHCP interface.
  • Inspected the GL.iNet modem monitoring scripts.

The modem uses:

  • "modem_2_1" for the QMI connection.
  • "modem_2_1_4" for DHCP.
  • "wwan0_1" as the data interface.

GL.iNet script investigation

I inspected:

"/lib/functions/modem.sh"

In particular:

  • "check_ip()"
  • "modem_net_monitor()"

The "check_ip()" function can terminate the DHCP client when the IP address reported by the modem differs from the interface IP and connectivity checks fail.

The "modem_net_monitor()" function can bring the modem interface down and then back up when no IPv4 address is detected.

However, I have not established that either function caused the observed outage.

I also inspected:

"/lib/netifd/proto/qcm.sh"

The QCM process is launched using:

qcm -v -f /tmp/qcm.log -4 -n 1

The script creates a dynamic DHCP interface and uses "proto_set_keep 1".

There is no explicit QCM restart loop in this shell script.

  1. Most important finding: QMI timeouts

I enabled verbose QCM logging in "/tmp/qcm.log".

During normal operation, QMI requests work correctly:

  • "QMIWDS_GET_PKT_SRVC_STATUS_REQ"
  • "QMIWDS_GET_RUNTIME_SETTINGS_RESP"

The modem reports "CONNECTED" and correctly returns its IPv4 address, gateway and DNS servers.

However, several incidents show the same sequence.

November 5, 2026 – 22:17

At approximately 22:17:31:

requestQueryDataCall message timeout
requestQueryDataCall err = 110

QCM immediately attempts to restore the data connection.

At approximately 22:18:

requestSetupDataCall err = 0
requestGetIPAddress err = 0
QmiWwanThread exit
qmi_main exit

The previous data connection is successfully stopped and QMI clients are released.

QCM then detects the modem again:

  • USB ID "2c7c:030b"
  • "/dev/cdc-wdm0" detected
  • Driver "qmi_wwan_q"
  • QMAP mode 1
  • Network interface "wwan0_1"

The modem subsequently reports:

PS: Attached, DataCap: LTE

A new data connection is established and the modem returns an IPv4 address.

Internet access returned at approximately 22:22.

Other occurrences

The same QMI timeout and internal QCM restart sequence was also recorded at approximately:

  • November 5, 21:30
  • November 5, 22:07

I was not present to confirm an actual Internet outage during these two events, so I cannot say whether they caused connection losses.

  1. Radio signal quality

At one point, the modem reported:

Parameter| Value
LTE band| B3 (1800 MHz)
Bandwidth| 20 MHz
RSRP| -110 dBm
RSRQ| -13 dB
SINR| 7 dB

The signal is not particularly strong, but the modem remains registered on the LTE network during the investigated incident.

I cannot determine whether radio conditions contribute to the QMI timeouts.

  1. What I would like GL.iNet to investigate

The main question is why QMI control requests time out with "err=110", even though:

  • The modem remains physically connected via USB.
  • The modem remains registered on the Orange LTE network.
  • The QMI interface is still detected.
  • QMI polling works normally between incidents.

Could this be related to:

  1. A QMI communication issue between the EG120K-EA modem and the "qmi_wwan_q" driver?
  2. A QCM / QConnectManager issue?
  3. A modem firmware issue?
  4. A GL.iNet monitoring mechanism or "netifd" terminating or restarting the QCM process?
  5. A known issue with this modem firmware or the GL-X2000 firmware version?

I would also like to know whether a newer router firmware, modem driver, QCM version or modem firmware is available to address this type of issue.

I can provide the detailed QCM logs and additional diagnostic outputs if required.

Thank you for your assistance.