Cleaning up all containers, tasks and their processes when KillMode=process

[joedborg]

Hi all,

I'm on the fence regarding whether this is a bug or not. https://github.com/containerd/containerd/blob/d8527867057daa93c17423ec7cd3438696b1c899/docs/ops.md and https://github.com/containerd/containerd/blob/d8527867057daa93c17423ec7cd3438696b1c899/containerd.service both suggest that KillMode=process, which is great to ensure tasks continue to run if containerd goes down for whatever reason. However, this poses an issue when I want to purge containerd from a system as it will leave tasks running.

I've tried running

for task in $(ctr task ls | sed -n '1!p' | awk '{print $1}')
do
    ctr task pause $task
    ctr task kill -s SIGKILL $task
    ctr task delete --force $task
done

for container in $(ctr containers ls | sed -n '1!p' | awk '{print $1}')
do
    ctr container delete --force $container
done

but still end up with dangling processes that were spawned by containerd.

Is there any way to safely do this? Maybe via systemctl?