Error when trying to run minikube with podman machine

[afbjorklund]

Something is different between podman and podman-remote (podman machine):

Error: error preparing container cc26ed492ddbd1f0f4dac495e4c64a3bc29c246d00c246c2df58ca87ea6e66ba for attach: error configuring network namespace for container cc26ed492ddbd1f0f4dac495e4c64a3bc29c246d00c246c2df58ca87ea6e66ba: error adding pod minikube-preload-sidecar_minikube-preload-sidecar to CNI network "podman": unexpected end of JSON input

[afbjorklund]

Same thing happens with "hello-world" or what have you, not minikube problem

Fedora CoreOS 34.20210919.2.0

podman-plugins-3.3.1-1.fc34.x86_64

podman version 3.3.1

[afbjorklund]

The bug is related to the new "podman-machine" CNI plugin, in the network config

      {
         "type": "portmap",
         "capabilities": {
            "portMappings": true
         }
      },
      {
         "type": "dnsname",
         "domainName": "dns.podman",
         "capabilities": {
            "aliases": true
         }
      },
      {
         "type": "podman-machine",
         "capabilities": {
            "portMappings": true
         }
      }

Fixed in podman-cni v0.2.0 (podman 3.4.0) containers/podman-machine-cni@3c5837b

[afbjorklund]

Editing /etc/cni/net.d/minikube.conflist to remove "podman-machine" makes it limp along

But then gvproxy fails to open the published ports from the host.

libmachine: Error dialing TCP: dial tcp 127.0.0.1:38891: connect: connection refused

Downgrading the Fedora CoreOS image (and thus podman version) should also work...

---

Even though the minikube container is running inside the VM:

[root@localhost ~]# podman ps
CONTAINER ID  IMAGE                                COMMAND     CREATED        STATUS            PORTS                                                                                                                                 NAMES
cd832881d511  gcr.io/k8s-minikube/kicbase:v0.0.27              5 minutes ago  Up 5 minutes ago  127.0.0.1:38891->22/tcp, 127.0.0.1:44603->2376/tcp, 127.0.0.1:38489->5000/tcp, 127.0.0.1:34769->8443/tcp, 127.0.0.1:45769->32443/tcp  minikube

[afbjorklund]

Note that downgrading only works for 10 minutes, then it auto-upgrades itself and reboots.

[afbjorklund]

There is a workaround, to get the new version from "devel": see containers/podman#11837 (comment)

[jorhett]

Seems you can just recreate the machine. This worked for me https://stackoverflow.com/a/69626738/3078154