I am working on an RPi4-4GB that will be used in the pump shack of a small community water system. The Pi has a USB Labjack data collection device to read pressure sensors, digital flow meters, float switches, etc. to predict and/or debug an issue that could cause a water outage.
Once installed, it will not be connected to any network. The user interface is a pair of outside led's that flash green for good, red for error, and, of course, the system display in the shack.
Data is presented to the user by a python3 app collecting the data and presenting real time on a simple tkinter graphic window. The python also feeds the data to an Influxdb so that the user can open a browser to have Grafana display time history graphs.
We have a severe problem with our power going out randomly multiple times per year. Thus, the system which had been working for several years, stopped due to a problem with Influxdb continually crashing after a short time in a Docker container.
I decided to use the latest RPI OS, dispense with Docker, and break out my battery backup code from the main python to an independent running task. This battery backup code is supposed to call for a normal shutdown when AC power goes down and then shutdown the Pi until AC is restored for an auto reboot. I think this was not happening leading to the docker influxdb problem.
I am testing the system at home with a hardware simulator for Labjack inputs and have the system working on the following OS:
lcl@watermon:~ $ uname -a
Linux watermon 6.1.0-rpi7-rpi-v8 #1 SMP PREEMPT Debian 1:6.1.63-1+rpt1 (2023-11-24) aarch64 GNU/Linux
lcl@watermon:~ $ cat /etc/os-release
PRETTY_NAME="Debian GNU/Linux 12 (bookworm)"
NAME="Debian GNU/Linux"
VERSION_ID="12"
VERSION="12 (bookworm)"
VERSION_CODENAME=bookworm
ID=debian
HOME_URL="https://www.debian.org/"
SUPPORT_URL="https://www.debian.org/support"
BUG_REPORT_URL="https://bugs.debian.org/"
"raspi-config reported it is at latest version per"
Version: 20240104
My latest code is working when it boots but after a random period of time, usually several hours or more, the display freezes. My simulator and python task are still working because the Labjack led signal is still blinking. I can ssh into the system but could not determine what was causing the display to stop. Of course, the mouse and keyboard are not responding and a ctl-alt-delete does not change the display.
Finally, I found a hint of what might be the issue. I noted the time at which the screen stopped on the frozen display. I googled and found a way to use journactl. My notes:
2024/01/14, 05:58:27 PM
Display still freezing, shows 2024-01-14 03:08:06
journalctl --since "2024-01-14 03:08:00" --until "2024-01-14 03:08:30"
...
Jan 14 03:08:04 watermon influxd-systemd-start.sh[1319]: [httpd] ::1 - root [14/Jan/2024:03:08:04 -0800] "POST /write?db=LCL_Monitor_test&precision=ms HTTP/1.1 " 204 0 "-" "py>
Jan 14 03:08:05 watermon influxd-systemd-start.sh[1319]: [httpd] ::1 - root [14/Jan/2024:03:08:05 -0800] "POST /write?db=LCL_Monitor_test&precision=ms HTTP/1.1 " 204 0 "-" "py>
Jan 14 03:08:06 watermon influxd-systemd-start.sh[1319]: [httpd] ::1 - root [14/Jan/2024:03:08:06 -0800] "POST /write?db=LCL_Monitor_test&precision=ms HTTP/1.1 " 204 0 "-" "py>
Jan 14 03:08:06 watermon kernel: vc4-drm gpu: [drm] *ERROR* Failed to allocate DLIST entry. Requested size=17. ret=-28
Jan 14 03:08:06 watermon kernel: vc4-drm gpu: [drm] *ERROR* Failed to allocate DLIST entry. Requested size=17. ret=-28
Jan 14 03:08:06 watermon kernel: vc4-drm gpu: [drm] *ERROR* Failed to allocate DLIST entry. Requested size=17. ret=-28
Jan 14 03:08:06 watermon kernel: vc4-drm gpu: [drm] *ERROR* Failed to allocate DLIST entry. Requested size=17. ret=-28
Jan 14 03:08:06 watermon kernel: vc4-drm gpu: [drm] *ERROR* Failed to allocate DLIST entry. Requested size=17. ret=-28
Jan 14 03:08:06 watermon kernel: vc4-drm gpu: [drm] *ERROR* Failed to allocate DLIST entry. Requested size=17. ret=-28
Jan 14 03:08:06 watermon kernel: vc4-drm gpu: [drm] *ERROR* Failed to allocate DLIST entry. Requested size=17. ret=-28
Jan 14 03:08:06 watermon kernel: vc4-drm gpu: [drm] *ERROR* Failed to allocate DLIST entry. Requested size=17. ret=-28
...
Googling the error text led me to:
Appears to be known bug per following:
viewtopic.php?t=357826
https://github.com/raspberrypi/linux/issues/5674
https://github.com/raspberrypi/linux/pull/5684
My problem is that I don't know how to read those bug comments in order to find the latest version that may have a fix.
Yesterday, I ran "sudo apt dist-upgrade" but the error still appears after a few hours of testing.
1. Can someone tell me if this bug is supposed to be fixed and what version I need?
2. Would it be better to just start from scratch with an install of Bullseye (64bit?/desktop) and abandon the Bookworm Version?
It takes me quite a bit of time to setup everything from a fresh install so I'd rather not. But it looks like I am on the bleeding edge of this Wayland stuff. Maybe Bookworm is not stable enough for a system that will never be updated again?
3. Other ideas? I can provide more detail on the display setup, the usb-128gb ssd, etc., but I think I have been too wordy on this post already.
Once installed, it will not be connected to any network. The user interface is a pair of outside led's that flash green for good, red for error, and, of course, the system display in the shack.
Data is presented to the user by a python3 app collecting the data and presenting real time on a simple tkinter graphic window. The python also feeds the data to an Influxdb so that the user can open a browser to have Grafana display time history graphs.
We have a severe problem with our power going out randomly multiple times per year. Thus, the system which had been working for several years, stopped due to a problem with Influxdb continually crashing after a short time in a Docker container.
I decided to use the latest RPI OS, dispense with Docker, and break out my battery backup code from the main python to an independent running task. This battery backup code is supposed to call for a normal shutdown when AC power goes down and then shutdown the Pi until AC is restored for an auto reboot. I think this was not happening leading to the docker influxdb problem.
I am testing the system at home with a hardware simulator for Labjack inputs and have the system working on the following OS:
lcl@watermon:~ $ uname -a
Linux watermon 6.1.0-rpi7-rpi-v8 #1 SMP PREEMPT Debian 1:6.1.63-1+rpt1 (2023-11-24) aarch64 GNU/Linux
lcl@watermon:~ $ cat /etc/os-release
PRETTY_NAME="Debian GNU/Linux 12 (bookworm)"
NAME="Debian GNU/Linux"
VERSION_ID="12"
VERSION="12 (bookworm)"
VERSION_CODENAME=bookworm
ID=debian
HOME_URL="https://www.debian.org/"
SUPPORT_URL="https://www.debian.org/support"
BUG_REPORT_URL="https://bugs.debian.org/"
"raspi-config reported it is at latest version per"
Version: 20240104
My latest code is working when it boots but after a random period of time, usually several hours or more, the display freezes. My simulator and python task are still working because the Labjack led signal is still blinking. I can ssh into the system but could not determine what was causing the display to stop. Of course, the mouse and keyboard are not responding and a ctl-alt-delete does not change the display.
Finally, I found a hint of what might be the issue. I noted the time at which the screen stopped on the frozen display. I googled and found a way to use journactl. My notes:
2024/01/14, 05:58:27 PM
Display still freezing, shows 2024-01-14 03:08:06
journalctl --since "2024-01-14 03:08:00" --until "2024-01-14 03:08:30"
...
Jan 14 03:08:04 watermon influxd-systemd-start.sh[1319]: [httpd] ::1 - root [14/Jan/2024:03:08:04 -0800] "POST /write?db=LCL_Monitor_test&precision=ms HTTP/1.1 " 204 0 "-" "py>
Jan 14 03:08:05 watermon influxd-systemd-start.sh[1319]: [httpd] ::1 - root [14/Jan/2024:03:08:05 -0800] "POST /write?db=LCL_Monitor_test&precision=ms HTTP/1.1 " 204 0 "-" "py>
Jan 14 03:08:06 watermon influxd-systemd-start.sh[1319]: [httpd] ::1 - root [14/Jan/2024:03:08:06 -0800] "POST /write?db=LCL_Monitor_test&precision=ms HTTP/1.1 " 204 0 "-" "py>
Jan 14 03:08:06 watermon kernel: vc4-drm gpu: [drm] *ERROR* Failed to allocate DLIST entry. Requested size=17. ret=-28
Jan 14 03:08:06 watermon kernel: vc4-drm gpu: [drm] *ERROR* Failed to allocate DLIST entry. Requested size=17. ret=-28
Jan 14 03:08:06 watermon kernel: vc4-drm gpu: [drm] *ERROR* Failed to allocate DLIST entry. Requested size=17. ret=-28
Jan 14 03:08:06 watermon kernel: vc4-drm gpu: [drm] *ERROR* Failed to allocate DLIST entry. Requested size=17. ret=-28
Jan 14 03:08:06 watermon kernel: vc4-drm gpu: [drm] *ERROR* Failed to allocate DLIST entry. Requested size=17. ret=-28
Jan 14 03:08:06 watermon kernel: vc4-drm gpu: [drm] *ERROR* Failed to allocate DLIST entry. Requested size=17. ret=-28
Jan 14 03:08:06 watermon kernel: vc4-drm gpu: [drm] *ERROR* Failed to allocate DLIST entry. Requested size=17. ret=-28
Jan 14 03:08:06 watermon kernel: vc4-drm gpu: [drm] *ERROR* Failed to allocate DLIST entry. Requested size=17. ret=-28
...
Googling the error text led me to:
Appears to be known bug per following:
viewtopic.php?t=357826
https://github.com/raspberrypi/linux/issues/5674
https://github.com/raspberrypi/linux/pull/5684
My problem is that I don't know how to read those bug comments in order to find the latest version that may have a fix.
Yesterday, I ran "sudo apt dist-upgrade" but the error still appears after a few hours of testing.
1. Can someone tell me if this bug is supposed to be fixed and what version I need?
2. Would it be better to just start from scratch with an install of Bullseye (64bit?/desktop) and abandon the Bookworm Version?
It takes me quite a bit of time to setup everything from a fresh install so I'd rather not. But it looks like I am on the bleeding edge of this Wayland stuff. Maybe Bookworm is not stable enough for a system that will never be updated again?
3. Other ideas? I can provide more detail on the display setup, the usb-128gb ssd, etc., but I think I have been too wordy on this post already.
Statistics: Posted by bobstanl — Mon Jan 15, 2024 5:13 am