Troubleshooting

Enabling debug logs

Pass --alsologtostderr to minikube commands to see detailed log output output. To increase the log verbosity, you can use:

• -v=1: verbose messages
• -v=2: really verbose messages
• -v=8: more log messages than you can possibly handle.

Example:

minikube start --alsologtostderr --v=2 will start minikube and output all the important debug logs to stderr.

Post-mortem minikube debug logs

minikube stores post-mortem logs in the temporary directory of your system. One log file is created per subcommand and any subsequent invocations of the subcommand with the same args will append to the same file. If the log file has exceeded 1MB in size, a new log file is created. On macOS or Linux, it’s easy to get a list of recent logs:

find $TMPDIR -mtime -1 -type f -name "*minikube*" -ls  2>/dev/null

For instance after running minikube start, the above comamnd will show:

-rw-r--r-- 1 user grp 718 Aug 18 12:40 /var/folders/n1/qxvd9kc/T//minikube_start_dc950831e1a232e0318a6d6ca82aaf4f4a8a048b_0.log

These are plain text log files: you may rename them to “.log” and then drag/drop them into a GitHub issue for further analysis by the minikube team. You can quickly inspect the final lines of any of these logs via:

tail -n 10 <filename>

for example, this shows:

I0818 12:40:17.027317   63501 out.go:197] Setting ErrFile to fd 2...
I0818 12:40:17.027321   63501 out.go:231] isatty.IsTerminal(2) = true
I0818 12:40:17.027423   63501 root.go:272] Updating PATH: /Users/tstromberg/.minikube/bin
I0818 12:40:17.027715   63501 mustload.go:64] Loading cluster: minikube

Gathering VM logs

To debug issues where Kubernetes failed to deploy, it is very useful to collect the Kubernetes pod and kernel logs:

minikube logs

Viewing Pod Status

To view the deployment state of all Kubernetes pods, use:

kubectl get po -A

Example output:

NAMESPACE     NAME                        READY   STATUS    RESTARTS   AGE
kube-system   coredns-5c98db65d4-299md    1/1     Running   0          11m
kube-system   coredns-5c98db65d4-qlpkd    1/1     Running   0          11m
kube-system   etcd-minikube               1/1     Running   0          10m
kube-system   gvisor                      1/1     Running   0          11m
...
kube-system   storage-provisioner         1/1     Running   0          11m

To view more detailed information about a pod, use:

kubectl describe pod <name> -n <namespace>

Debugging hung start-up

minikube will wait ~8 minutes before giving up on a Kubernetes deployment. If you want to see startup fails more immediately, consider using:

minikube logs --problems

This will attempt to surface known errors, such as invalid configuration flags. If nothing interesting shows up, try minikube logs.