driver=podman, container-engine=docker, restart stopped minikube OCI runtime error #7996
Labels
co/podman-driver
podman driver issues
co/runtime/docker
Issues specific to a docker runtime
kind/bug
Categorizes issue or PR as related to a bug.
priority/important-longterm
Important over the long term, but may not be staffed and/or may need multiple releases to complete.
Comments
elegos
changed the title
|
Possibly related to containers/podman#4481 and migration issues from cgroups v2 EDIT: Nope, seems like only the error was the same. Easy to reproduce this (stop and start)
|
afbjorklund
added
kind/bug
|
But it does seem like So it might be something related after all, though more like systemd vs cgroupfs than v1 vs v2 ? Seems to be more like an accident, but anyway. The usual start was:
This start now is:
So this failure is somehow more acceptable to it. |
|
This bug seems reproducible with |
|
The code says: // to run nested container from privileged container in podman https://bugzilla.redhat.com/show_bug.cgi?id=1687713
if ociBin == Podman {
args = append(args, "--cgroup-manager", "cgroupfs")
}So if we are going to use that workaround, we need it for "podman start" as well... Because on fedora, the default is systemd. But it seems it doesn't work for DIND:
On a side note, the error reporting code is broken. There is an |
afbjorklund
added
priority/important-soon
The bug is with the driver runtime, so independent on the inner runtime. |
|
Hello @afbjorklund I've tried your tree, and I've encountered the same issue: |
|
That is weird, where did the "--cgroup-manager cgroupfs" go ? as in: |
It is not in your output: 🤦 StartHost failed, but will try again: driver start: start: sudo podman start minikube: exit status 125 |
|
I've added |
|
In any case I've tried manually restarting the minikube container, but with no better results (though it fails with a more talking error): |
|
I think you also want to increase verbosity:
But the error looked different this time ? |
|
I know nothing but what the console tells me :( And before you ask: |
|
Looks happy enough now This is just an rpm packaging issue: |
|
Unfortunately it did nothing :( |
|
New fact: now it seems it's not "hard" failing, but still there's something to do with systemd cgroup, this is the minikube's container log (from when I try to wake it up): It seems that, even if the cgroup-manager is set to cgroupfs, minikube's image is still trying to use systemd? |
|
Another fact: the previous logs are with |
|
See #8033 (comment) and #8033 (comment) for why systemd isn't starting when running with podman |
afbjorklund
added
priority/important-longterm
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Minikube is able to start once. When stopped, podman won't be able to turn it on again due to an
OCI runtime error. This seems to happen only on minikube (i.e. no problems stopping and starting other services).Steps to reproduce the issue:
minikube start --driver=podmanminikube stopminikube startFull output of failed command:
minikube startwith--alsologtostderroption:sudo podman start minikubeFull output of
minikube startcommand used, if not already included:Optional: Full output of
minikube logscommand:Thank you for your awesome work :)
The text was updated successfully, but these errors were encountered: