Docker repeatedly crashing due to faulty daemon.json

Technical Support
, , ,
1

hey guys,

when using the apolloX toolchain, i have seen that when connecting to the board, the daemon.json file under /etc/docker/daemon.json is written to with the ip adress of the dev machine

for some reason, my toolchain managed to write the following to the file:

{
   "insecure-registries" : [":5002"]
}

this lead to the docker instance on my board crash immediately on startup
the device is now in a bootloop i cant get it out of
however, when im lucky, i can ssh onto the board as soon as it starts up and have like 5 seconds until docker starts, crashes and the docker watchdog reboots the board

thank you for your help!

SoM: Colibri iMX 6ULL emmc
TorizonOS: 6.1.46-6.5.0+git.8e6a2ddd4fe6 #1-TorizonCore SMP PREEMPT

2

Hi, I’m the Toradex AI Assistant. I’m in early beta, and I would appreciate your feedback.
The Toradex Support team will follow up with you shortly!

Did you like my answer?

:+1: :-1:

Hello,

It seems like the issue you’re encountering is due to an improperly formatted entry in the daemon.json file for Docker, specifically the "insecure-registries" field. The value you’ve provided lacks an IP address or hostname before the port, which is likely causing Docker to fail upon startup.

To resolve this issue, you will need to correct the daemon.json file with the proper IP address and port of your insecure registry, or if it was not intended to use an insecure registry, remove or correct this entry. Here’s how you can do it within the limited time window you have after booting:

  1. Prepare the Correct Configuration: First, determine the correct configuration for your daemon.json. If you’re using an insecure registry, it should include the IP address or hostname, for example:

    {
   "insecure-registries" : ["192.168.1.100:5002"]
}

    If you do not need this configuration, you can either remove the line or ensure the file is correctly formatted for your Docker setup.

  2. Quick SSH and Edit: As soon as you SSH into the board, quickly navigate to the /etc/docker/ directory. You can use a command-line text editor like nano or vi to edit the daemon.json file. Given your tight time frame, you might find it easier to prepare the command beforehand and paste it as soon as you log in:

    sudo nano /etc/docker/daemon.json

    Then, correct the file as per your requirements.

  3. Restart Docker: If you manage to correct the file before Docker causes the system to reboot, you should try to manually restart the Docker service to see if the issue is resolved. This can be done with:

    sudo systemctl restart docker

    If you’re quick enough, this might prevent the boot loop from continuing.

  4. Alternative Approach: If the window of opportunity is too short for editing the file, consider creating a script on your development machine that logs into the board and corrects the file automatically. This script would use SSH to access the board and could be executed in a loop from your development machine until it successfully corrects the daemon.json file.

Remember, the key to resolving this issue is to ensure that the daemon.json file is correctly formatted and contains the appropriate entries for your Docker setup. Incorrect entries or syntax errors in this

Please, also check if these links can help you solve your problem:

Reliability on Torizon | Toradex Developer Center.

3

i have already managed to fix the problem by repairing the file with this command
echo '{"insecure-registries" : ["192.168.xx.xx:5002"]}' | sudo tee /etc/docker/daemon.json > /dev/null
but i am still trying to figure out why this happened

4

Greetings @xlukem,

For some reason our IDE extension wasn’t able to determine your system’s IP address. This was seen in the past, but we thought we improved the situation since then.

We may need more details regarding your dev machine. Is this a windows or linux based machine?

Does your machine have multiple active network interfaces at the same time?

Any other pertinent details?

EDIT: I checked with our team internally, and it seems this issue should have been fixed already. What version of our IDE extension do you have installed? Also, what is the git branch of /home/<user>/.apollox on your system that is running VSCode?

Best Regards,
Jeremias