Two Pods are running: a website and the API behind it.

Both live in a namespaceNamespaceA named partition of the cluster. Most objects live in exactly one.: a named partition of the cluster. Every Pod belongs to exactly one. kubectl get pods on its own only looks in the namespace called default, so to see Pods anywhere else you add -n <namespace>.

Check on our Pods:

kubectl get pods -n shallows-1wc58hj

One Pod is Running. The other cycles between Error and CrashLoopBackOff, with a RESTARTS count that keeps climbing. That is a crash loop: the container starts, exits, and Kubernetes restarts it, waiting a little longer each time.

Read before you touch

Crash loops explain themselves, if you ask the right thing:

kubectl logs market-api -n shallows-1wc58hj

The API is complaining about something it needs and does not have. The second objective wants the name of that thing exactly as the log spells it.

describe adds the other half of the picture: the exit code, the Last State, and an Events timeline of every back-off:

kubectl describe pod market-api -n shallows-1wc58hj

Replace, don't repair

The Pod is running the wrong image. stackmonsters/crasher is a placeholder someone forgot to swap out; the real API image is stackmonsters/monster-market-api, which already knows how to find its database.

A Pod's container image can be swapped in place, but it is cleaner, and more common in practice, to delete the broken Pod and create the right one:

kubectl delete pod market-api -n shallows-1wc58hj
kubectl run market-api --image=stackmonsters/monster-market-api -n shallows-1wc58hj

Then confirm:

kubectl get pods -n shallows-1wc58hj

Both Pods Running, RESTARTS at zero.