FATAL error starting the kubehound stack: dependency failed to start: container kubehound-release-kubegraph-1 is unhealthy. Please make sure your Docker is correctly installed and that you have enough disk space

[GurdipS5]

Hi,
I am running Kubehound on an ubuntu host. I have 150gb on the disk, but keep getting this error:

15:10:06 INFO Loading application configuration from default embedded app=kubehound
15:10:06 WARN No local config file was found app=kubehound file=kubehound.yaml
15:10:06 INFO Using file for default config app=kubehound path=
15:10:06 INFO Initializing application telemetry app=kubehound
15:10:06 WARN Telemetry disabled via configuration app=kubehound
15:10:06 INFO Loading backend from default embedded app=kubehound
15:10:06 INFO Spawning the kubehound stack app=kubehound
[+] Running 3/3
✔ Container kubehound-release-ui-jupyter-1 Started11.1s
✔ Container kubehound-release-mongodb-1 Recreated10.7s
✘ Container kubehound-release-kubegraph-1 Error57.2s
15:11:03 FATAL error starting the kubehound stack: dependency failed to start: container kubehound-release-kubegraph-1 is unhealthy. Please make sure your Docker is correctly installed and that you have enough disk space available. app=kubehound
main.main
/src/cmd/kubehound/main.go:19
runtime.main
/usr/local/go/src/runtime/proc.go:283

However, if I run df -h, i see:

/dev/mapper/ubuntu--vg-ubuntu--lv 72G 8.5G 60G 13% /

So there is enough disk space.

[Minosity-VR]

Thanks for reporting, I will investigate!

Can you please post here the logs of the docker? Something like docker logs kubehound-release-kubegraph-1
edit: if the container failed to start, there isn't probably any logs, but let's just make sure it's not an error inside the container 😄

Also, if you are running docker desktop, can you check how much disk you allowed docker to use? It's the "Disk usage limit" setting under the "Resources" tab. It often takes much less disk space than you total disk and could explain why you are seing the failed to start error. You can additionally check the total size used by docker with docker system df

[GurdipS5]

Just checked logs, I found it was complaining about lack of memory, so let me try bumping that up!

…

On Thu, Oct 2, 2025 at 4:31 PM Simon Maréchal ***@***.***> wrote: *Minosity-VR* left a comment (DataDog/KubeHound#358) <#358 (comment)> Thanks for reporting, I will investigate! Can you please post here the logs of the docker? Something like docker logs kubehound-release-kubegraph-1 Also, if you are running docker desktop, can you check how much disk you allowed docker to use? It's the "Disk usage limit" setting under the "Resources" tab. It often takes much less disk space than you total disk and could explain why you are seing the failed to start error. You can additionally check the total size used by docker with docker system df — Reply to this email directly, view it on GitHub <#358 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AU2JXHGUYAMVVYEKFLEGSRL3VVALTAVCNFSM6AAAAACIDYW2QSVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZTGNRRHA2TEMBUGI> . You are receiving this because you authored the thread.Message ID: ***@***.***>

[GurdipS5]

Ok so re ran this after giving the server 8gb ram, now get this error:

16:39:10 FATAL health check: missing cluster name app=kubehound
main.main
/src/cmd/kubehound/main.go:19
runtime.main
/usr/local/go/src/runtime/proc.go:283

Btw, does this tool work with k3s?

[Minosity-VR]

This looks like a config issue, it seems kubehound is not finding your cluster.

KubeHound internally uses the various kubernetes packages, so it should work with any framework/distribution if they follow the Kubernetes spec (for instance, it uses kind for the system tests)

There might be difference on where k3s stores its config. KubeHound uses Kubernetes' clientcmd package, which I believe follows the same order as kubectl.

k3s apparently creates a kubeconfig file in /etc/rancher/k3s/k3s.yaml on linux, but I saw people mentioning ~/k3s.yaml (in the home directory).

Can you try copying this file to ~/.kube/config, or run kubehound with the KUBECONFIG env var set to the k3s config path?

export KUBECONFIG=~/k3s.yaml # or export KUBECONFIG=/etc/rancher/k3s/k3s.yaml
kubehound

[GurdipS5]

Hey, same thing, but I have my kubeconfig file added, and can browse my cluster.

I also added this command:

export KUBECONFIG=/home/x/.kube/config

I should add that I am putting kubehound on its own server.

[Minosity-VR]

Tested on a Ubuntu EC2:

• installed k3s with their script: curl -sfL https://get.k3s.io | sh -
• installed docker following their instructions
• made docker available sudoless to my user
• install kubehound from release

Please tell me if I didn't correctly reproduce your setup, but I had to use sudo to access k3s config files.
Did you perhaps ran kubehound in sudo too? If so, you need to add -E so share the environment to the sudo process.
Additionally, if you run KH with sudo, the golang kube client looks for the config file under /root/.kube/config (I don't understand why yet, as I thought sudo was preserving $HOME, but 🤷. I had to rebuild it with more logging to find out).

I was able to run it in two ways:

1. Run sudo with -E: KUBECONFIG=/etc/rancher/k3s/k3s.yaml sudo -E ./kubehound --skip-backend
2. Copy the config file to the user's home, change the owner from root to user, run kubehound: cd ~ && mkdir .kube && sudo cp /etc/rancher/k3s/k3s.yaml ~/.kube/config && sudo chown ubuntu:ubuntu ~/.kube/config && ./kubehound --skip-backend

(the --skip-backend is too avoid rebuilding the containers if they are already present)

Hope this will fix your problem!

[GurdipS5]

Ok so the deployment succeeds and it states the url to access the UI.

However, with my firewall off and if I do http://x.x.x.x:8888 (where x.x.x.x is my server's IP), the page doesn't load?