This bug I observed in all Armbian variants Ubuntu Jammy, Noble & Debian Bookworm
Bug changes but mostly it is like this
Error response from daemon: failed to create task for container: failed to create shim task: OCI runtime create failed: runc did not terminate successfully: exit status 134: unknown
Sometimes it changes to something more gibbersish but in the end CasaOS fails to initiate container at proxy binding step. Container do install 100%. The output of sudo docker ps -a display the container state as created not started This bug occurs after you have create 4 to 5 containers. This user experienced exactly same issue
The problem still exists today. I have now reinstalled the OS and Docker several times and tested the setup with different containers. After all that time I assume that there is actually a problem with the resources, because I can deploy between 5 and 6 containers at a time, depending on the images. One more run command and I get the aforementioned error.
My problem is that I have no idea which resource might be the limiting one. Using htop I can see that all CPUs are idling at about 0% to 1%. When I request a website from the docker, one of the four CPUs goes up to 25%, the others are still around 1%. Also, the memory is constantly between 900 and 1000 MB, but there are 8 GB available
Some user insisted its a memory leak / resources problem Then I tried to edit /etc/security/limits.conf as explained in RHEL tutorial but didn't succeed
If the output of sysctl kernel.threads-max is lower than 2000 increase it by editing sudo nano /etc/sysctl.conf
kernel.threads-max = 2055 # at EOF
I got this value from Armbian noble image running on same CPU RK3566 on OPi3b v2.1. Reboot afterwards
Finally I used some old debian packages. It must be noted in official Docker documentation these legacy packages are meant to be removed but in my case I have to reinstall these! Issued the following command & it skipped some packages but did install some other legacy packages. After this my existing CasaOS containers were completely removed. I installed new ones with no error
for pkg in docker.io docker-doc docker-compose docker-compose-v2 podman-docker containerd runc golang-github-containernetworking-plugin-dnsname; do sudo apt-get reinstall $pkg; done
Note golang-github-containernetworking-plugin-dnsname is important for internal DNS name resolution especially if you wanr to run paperless-ngx stack issue
After a reboot hold these packages to avoid update
sudo apt-mark hold docker-* python3-compose* python3-docker* runc crun podman* containerd containernetworking-*
To view previous held packages apt-mark showhold
To un-hold held packages sudo apt-mark unhold $(apt-mark showhold)
I recently updated Casa-OS after about x6 months of stable system run. Again this OCI-runtime / crun errors are happening. Considering same podman version as well as same Ubuntu Noble release it seems others also have this issue
https://askubuntu.com/questions/1517387/container-is-stuck-in-state-stopping-cannot-clear-it
I have installed a great docker management CLI utility ctop by running sudo ctop -a it showed me all containers managed by podman but it was incomplete list. This helped me but I didnt want to loose by containers & respective data
So I issued in the terminal (but didn't press yes) sudo podman system reset & it showed x2 old containers that were invisible in Casa-OS webui, ctop or podman ps -a! I removed those x2 entries one by one via sudo podman rm <container_id> & reboot system. Everything stable for now.
The other issue I discovered relates to systemd & startup of containers at boot. This is dicsussed here:
https://github.com/defencedog/radxazero3E/blob/main/Systemd_Service_Docker_health_checker.md