The error message I always face is as below: sudo systemctl restart docker sudo usermod -aG docker $USER newgrp docker Job for docker.service failed because the control process exited with error code. See āsystemctl status docker.serviceā and ājournalctl -xeu docker.serviceā for details.
When I run systemctl status docker.service, the following is the message: Ć docker.service - Docker Application Container Engine
I gave it one final try by reflashing my SD card to the 6.2 image again.
This round what I did was after the initial setup pages after the system bootup, I did not allow the system to upgrade Ubuntu and the rest of the software when it prompts.
From there, I was able to install the dusty jet containers, ollama and even the open-web UI as per the Jetson ai labs page without any issue and docker was running fine.
However! After I find that everything is running fine, I let the system to upgrade itself and all of the software and voila, the above problem happens again with the daemon and docker unable to run anymore (same error as my previous post). Does it means I forever cannot upgrade my Jetsonās software?
Is this even normal in the 1st place for the Jetson series?
Will Appreciate any help as I am really at my witās end after trying and searching so much for help.
Iāve been playing with this thing for a week.
Iām both impressed and frustrated in equal parts, lol. Been a bit of a learning curve.
To answer your second question; ubuntu will keep those at that level until you apt unhold them. Obviously, at some point they will be significantly out of date.
At some point, I assume this will be fixed. The solution is a jetson kernel-level issue, so Iām assuming this will be a Jetpack 6.3+ level fix.
Unless the SDK manager forces a specific version, I think this hold will remain in effect, even if you do an apt upgrade
Thank you!
Yeah I share the same sentiments as you, was initially impressed by such a little thing but powerfulā¦ but the docker issue (which I spend almost a week every since purchase) frustated me.
As a pure Windows guy, I think this will be a very steep learning but fun curve for me, I hope :)
I would suggest using an SSD, and making backups of it using a USB caddy, and partition it sensibly, so user files, models, etc are on a different partition. That way when you break it in new and interesting ways, you can backtrack.
docker is also hungry for space, so youāll find you run out pretty quickly.
Ask me how many times Iāve reflashed this thing in the last weekā¦ lol.
haha seeing how many reflashes which I have done for the past week as I dunno the solution, I cannot imagine how many reflashes you went through? 6 to 7 times?
I am following this guide to migrate from SD card to SSD:
Iāve updated to Jetpack 6.2 and get the same issues after having run nicely before.
Iāve reflashed everything and get the same errors. Iāve tried the official docker install script: same errors. It must be caused by docker v28. Earlier versions worked for me, but i cannot get GPU support running in docker any longer. Iāll have to reflash to an earlier Jetpack.
I canāt seem to find a solution, so Iāll just downgrade and try it again in a few weeks.
Iām starting docker with nvidia runtime but it does not result in GPU usage. jtop shows 100% CPU, GPU remains at 0%. This has not happened with my latest setup.
As discussed above, this issue is caused by the latest docker (released last week) requires some extra kernel configuration.
Our internal team is working on this issue actively.
For now, please downgrade the docker to 27.5.1 or build a custom kernel manually.
Please find below for more info: