better error output when docker isn't/can't log in

[squillace]

What would you like to be added:
The ability get better error output in the non-default case when creating a new cluster and not specifying the default kind create cluster.

Why is this needed:
I stumbled around for the better part of a week trying to understand why this didn't work, below. only when I ran it as the default command did it appear to output the error -- it's picking up my docker login and I'd changed the password but had not re-logged in. After logging in properly, it is creating properly.

╭─ralph at surfacebook in ~/work/go/bin using
╰─○ kind create cluster --image=kindest/node:v1.14.0 --name kindly
Creating cluster "kindly" ...
 ✓ Ensuring node image (kindest/node:v1.14.0) 🖼
 ✓ Preparing nodes 📦
 ✓ Creating kubeadm config 📜
 ✗ Starting control-plane 🕹️
Error: failed to create cluster: failed to init node with kubeadm: exit status 1
╭─ralph at surfacebook in ~/work/go/bin using
╰─○ kind create cluster
Creating cluster "kind" ...
 ✓ Ensuring node image (kindest/node:v1.14.2) 🖼
ERRO[17:42:25] Unable to find image 'kindest/node:v1.14.2@sha256:33539d830a6cf20e3e0a75d0c46a4e94730d78c7375435e6b49833d81448c319' locally
ERRO[17:42:25] docker: Error response from daemon: Get https://registry-1.docker.io/v2/kindest/node/manifests/sha256:33539d830a6cf20e3e0a75d0c46a4e94730d78c7375435e6b49833d81448c319: unauthorized: incorrect username or password.
ERRO[17:42:25] See 'docker run --help'.
 ✗ Preparing nodes 📦
ERRO[17:42:25] docker run error: exit status 125
Error: failed to create cluster: docker run error: exit status 125

[BenTheElder]

cc @amwat exactly the kind of self-diagnosis I was talking about earlier ^^

xref: #39
/priority important-longterm

[squillace]

@BenTheElder this was why it was failing on my machine at KubeCon. :-)

[fejta-bot]

Issues go stale after 90d of inactivity.
Mark the issue as fresh with /remove-lifecycle stale.
Stale issues rot after an additional 30d of inactivity and eventually close.

If this issue is safe to close now please do so with /close.

Send feedback to sig-testing, kubernetes/test-infra and/or fejta.
/lifecycle stale

[philips]

Hit this issue on a debian machine where I didn't have access to the docker socket.

[BenTheElder]

/remove-lifecycle stale

Yep, error handling / reporting / logging is something I'm back at reshaping right now, it's subpar :(

We do at least have a note about the docker socket in the known issues guide now..

[BenTheElder]

/assign
/lifecycle active

[aojea]

I think that @BenTheElder fixed this with the new logging mechanism
Currently, the error that kind shows is meaningful

./kind create cluster
ERROR: failed to list clusters: command "docker ps -q -a --no-trunc --filter label=io.k8s.sigs.kind.cluster=kind --format '{{.Names}}'" failed with error: exec: "docker": executable file not found in $PATH

/close

[k8s-ci-robot]

@aojea: Closing this issue.

In response to this:

I think that @BenTheElder fixed this with the new logging mechanism
Currently, the error that kind shows is meaningful

./kind create cluster
ERROR: failed to list clusters: command "docker ps -q -a --no-trunc --filter label=io.k8s.sigs.kind.cluster=kind --format '{{.Names}}'" failed with error: exec: "docker": executable file not found in $PATH

/close

Instructions for interacting with me using PR comments are available here. If you have questions or suggestions related to my behavior, please file an issue against the kubernetes/test-infra repository.

[estensen]

IMO, as a beginner, this is not beginner-friendly:

$ kind create cluster
ERROR: failed to list clusters: command "docker ps -q -a --no-trunc --filter label=io.x-k8s.kind.cluster=kind --format '{{.Names}}'" failed with error: exit status 1

There should be a hint that Docker is not running

[aojea]

IMO, as a beginner, this is not beginner-friendly:

$ kind create cluster
ERROR: failed to list clusters: command "docker ps -q -a --no-trunc --filter label=io.x-k8s.kind.cluster=kind --format '{{.Names}}'" failed with error: exit status 1

There should be a hint that Docker is not running

the problem is that it may be caused by another reasons, not just that Docker is not running, so it can be wrong to always return an error saying "Docker is not running"

[BenTheElder]

What aojea said.

Sorry, we know this isn't friendly, there's just a LOT to do and initially we were targeting kubernetes (literally kubernetes itself) developers. There's a lot to do. I have a sketch for more sophisticated help in the command but we've got some deep rework that needs doing to fix e.g. host reboot support first.

[BenTheElder]

I recommend running again with -v 1 on the latest release when you encounter errors until that rework is published.

[BenTheElder]

Especially now that we auto-detect between multiple backends (podman, docker), it would be helpful to write some code specifically to probe for "is the host capable of hosting a cluster"

We have some of this now related to e.g. cgroupsv2 and rootless where we will throw more descriptive errors, but not for this case yet.

This also has some overlap with #39, i would love a command to diagnose problems with the host.

[linuxem]

Still having this error
Docker is running I can run docker ps
But kine create cluster gives an error

Error

even the docker gives error when running this
docker ps -q -a --no-trunc --filter label=io.k8s.sigs.kind.cluster --format {{.Names}}
see this:

"just information about this docker installation it is Docker Desktop 4.20.0"
Any suggestions?

[linuxem]

update to the newest kind fixed the problem.

/close

[stmcginnis]

Based on the previous comment, going to reclose this issue. I think further improvements can be tracked somewhat with #39.

/close

[k8s-ci-robot]

@stmcginnis: Closing this issue.

In response to this:

Based on the previous comment, going to reclose this issue. I think further improvements can be tracked somewhat with #39.

/close

Instructions for interacting with me using PR comments are available here. If you have questions or suggestions related to my behavior, please file an issue against the kubernetes-sigs/prow repository.