debian: runc create failed: error mounting cgroup to rootfs: something slice: ENOENT #22165
Description
Two instances, both seen only in last week's new VM build:
# podman [options] run --pod testpod quay.io/libpod/alpine:latest cat /etc/hosts
time="2024-03-25T17:20:16Z" level=error msg="Starting some container dependencies"
time="2024-03-25T17:20:16Z" level=error msg="\"runc: runc create failed: unable to start container process: error during container init: error mounting \\\"cgroup\\\" to rootfs at \\\"/sys/fs/cgroup\\\": stat /sys/fs/cgroup/devices/machine.slice/machine-libpod_pod_0b2bfa75b9e139af1542a302869e7de36c11ec40691616fd980cfc2a04813699.slice/libpod-1ae333654906d875ed36bea81bf0be74da47dc7e867c7ecc22c44b407240db64.scope: no such file or directory: OCI runtime attempted to invoke a command that was not found\""
Error: starting some containers: internal libpod error
Seems to be related to pods. Root only (so far).
Relevant VM diffs:
- kernel 6.6.15 -> 6.7.9
- runc 1.1.12+ds1-1 -> -2
- systemd 255.3 -> .4
Probably a don't-care, given our intention to abandon runc.
- debian-13 : int podman debian-13 root host sqlite
- 03-25 13:22 in Podman run networking podman run with pod does not add extra 127 entry to /etc/hosts
- 03-19 16:14 in Podman kube play with multiple networks
EDIT: forgot to mention, there's another current flake having to do with cgroup-slice, #20044, but I don't think it's related. Still, maybe.