Cannot start server node: klipper helm permission denied #4281
Replies: 3 comments 3 replies
-
|
You appear to have some sort of security module (probably selinux) enabled that is preventing pods from running correctly; did you install RKE2 using RPMs, install the rke2-selinux package, and enable selinux support? |
Beta Was this translation helpful? Give feedback.
-
|
With the same installation steps and configuration files, I can start normally under Ubuntu20.04 Kernel 5.4. You can also reproduce the problem by creating a new user under alpine and su - /home/klipper-helm |
Beta Was this translation helpful? Give feedback.
-
|
Hmm, what are the permissions on / when you do this? I do suspect that it is something related to the kernel blocking a syscall, and not actually anything to do with the filesystem permissions or container runtime. |
Beta Was this translation helpful? Give feedback.
-
|
I followed the air gap install from tarballs. So I did not try installing the selinux rpm. I will try and let you know! What did though, was to disable selinux, Weirdest thing is that containerd 1.7.0, installed on the same machine as mentioned in their “getting started”, runs the same container with no permission issues. Apart from selinux, do you have any other clue of what it might be? |
Beta Was this translation helpful? Give feedback.
-
|
Hi @daniel88ferreira - did you get anywhere with this, I'm having the same issue on both redhat8 and redhat9? |
Beta Was this translation helpful? Give feedback.
-
re: is that a network mount? don't do that. if it's not a network mount please be advised that the rke2-selinux policy does not support custom rke2 data-dir without further work by an operator, human or otherwise, that understands enough selinux to install/apply/enforce custom file-context policy for your custom location. take a look at the if none of this applies, consider posting a containerd log and security log (if selinux is in play, please paste the results of |
Beta Was this translation helpful? Give feedback.
-
Environmental Info:
RKE2 Version:
rke2 version v1.27.1+rke2r1 (9ce3825)
go version go1.20.3 X:boringcrypto
Node(s) CPU architecture, OS, and Version:
Cluster Configuration:
Describe the bug:
Cannot get the control plane to work since helm related pods fail immediately:
And these are the logs from the container itself:
Also trying to run the container directly I get the same error:
Note: I disabled SELinux and re-installed RKE2 but got the same behaviour.
Steps To Reproduce:
1- follow the guide for an air-gapped install here: https://docs.rke2.io/install/airgap
Expected behavior:
Control plane starts correctly.
Actual behavior:
Control plane helm pods fail
Additional context / logs:
Thank you!
Beta Was this translation helpful? Give feedback.
All reactions