Thanks, you for the work put into this, and for updating it. I really don’t know if this is a bug or just me not knowing docker, basicstation or development in general.
I’ve just tried to update an older gateway based on a Raspberry Pi 3B with an Pi Supply Lora GW HAT – RAK833 (https://www.thethingsnetwork.org/docs/gateways/pisupply-hat/). The GW has been running https://github.com/mpous/basicstation for a long time with GW_RESET_PIN set to 22, MODEL equal to SX1301.
I’ve made a fresh Pi OS 64-bit lite (Bookworm) installation on the Pi3 and followed the docker procedure with setting I’ve used before.
Using the following docker-compose
services:
basicstation:
image: xoseperez/basicstation:latest
container_name: basicstation
restart: unless-stopped
privileged: true
network_mode: host
environment:
MODEL: "RAK833"
INTERFACE: "SPI"
DEVICE: "/dev/spidev0.0"
RESET_GPIO: 22
GATEWAY_EUI: "b8xxxxxxxxxxxx65"
HAS_GPS: 1
GPS_DEV: "/dev/ttyS0"
TTS_REGION: "eu1"
TLS_SNI: true
TC_KEY: "NNSXS.*"
The output was
basicstation | Concentrator enabled through GPIO20 (using sysfs)
basicstation | Warning: /sys/class/gpio/gpio20 does not exist
basicstation | Concentrator reset through GPIO22 (using sysfs)
basicstation | Warning: /sys/class/gpio/gpio22 does not exist
and
basicstation | 2025-09-29 19:06:06.738 [S00:INFO] SX130x LBT not enabled
basicstation | 2025-09-29 19:06:06.738 [S00:INFO] Station device: /dev/spidev0.0 (PPS capture enabled)
basicstation | 2025-09-29 19:06:06.738 [S00:INFO] [lgw_spi_open:101] Setting SPI speed to 2000000
basicstation | 2025-09-29 19:06:06.738 [S00:INFO] [lgw_spi_open:135] Note: SPI port opened and configured ok
basicstation | 2025-09-29 19:06:06.738 [S00:ERRO] [lgw_start:742] FAIL TO CONNECT BOARD
Due to this I added USE_LIBGPIOD: 1 and GPIO_CHIP: "gpiochip0" to my environment.
This made reset.sh to time out.
basicstation | Concentrator reset through gpiochip0:22 (using libgpiod)
basicstation | 2025-09-29 20:33:47.475 [S00:ERRO] Process /app/config/reset.sh (pid=35) did not terminate within 200ms - killing it (SIGTERM)
basicstation | 2025-09-29 20:33:47.475 [S00:CRIT] Slave radio start up failed with status 0x04
basicstation | 2025-09-29 20:33:47.455 [RAL:ERRO] Slave (0) - EOF
basicstation | 2025-09-29 20:33:47.455 [any:CRIT] Slave pid=34 idx=0: Fatal exit
I finally got to work by fall back to legacy sysfs by adding dtoverlay=gpio-no-irq to /boot/firmware/config.txt and changed USE_LIBGPIOD to 0.
Could it be something off with reset.sh.gpiod?
According to https://forums.balena.io/t/balenaos-6-5-17-rev1-and-rak883/374073/6 the time out issue with reset.sh is not new to latest update.
Thanks, you for the work put into this, and for updating it. I really don’t know if this is a bug or just me not knowing docker, basicstation or development in general.
I’ve just tried to update an older gateway based on a Raspberry Pi 3B with an Pi Supply Lora GW HAT – RAK833 (https://www.thethingsnetwork.org/docs/gateways/pisupply-hat/). The GW has been running https://github.com/mpous/basicstation for a long time with GW_RESET_PIN set to 22, MODEL equal to SX1301.
I’ve made a fresh Pi OS 64-bit lite (Bookworm) installation on the Pi3 and followed the docker procedure with setting I’ve used before.
Using the following docker-compose
The output was
and
Due to this I added USE_LIBGPIOD: 1 and GPIO_CHIP: "gpiochip0" to my environment.
This made reset.sh to time out.
I finally got to work by fall back to legacy sysfs by adding dtoverlay=gpio-no-irq to /boot/firmware/config.txt and changed USE_LIBGPIOD to 0.
Could it be something off with reset.sh.gpiod?
According to https://forums.balena.io/t/balenaos-6-5-17-rev1-and-rak883/374073/6 the time out issue with reset.sh is not new to latest update.