Closed KyleSanderson closed 1 year ago
@KyleSanderson Can you provide your env information?
Kubernetes distro (e.g. RKE/K3s/EKS/OpenShift) and version:
Number of management node in the cluster:
Number of worker node in the cluster:
Node config
OS type and version:
SELinux enabled (true/false):
CPU per node:
Memory per node:
Disk type(e.g. SSD/NVMe):
Underlying Infrastructure (e.g. on AWS/GCE, EKS/GKE, VMWare/KVM, Baremetal):
Did you try #283 (comment)?
Yes, no impact. This is very easy to reproduce...
Kubernetes distro (e.g. RKE/K3s/EKS/OpenShift) and version: v1.24.10+rke2r1 Number of management node in the cluster: SINGLE NODE OS type and version: 5.14.0-162.12.1.el9_1.0.1.x86_64 SELinux enabled (true/false): Enforcing CPU per node: 92 Memory per node: 1024 Disk type(e.g. SSD/NVMe): NVMe Underlying Infrastructure (e.g. on AWS/GCE, EKS/GKE, VMWare/KVM, Baremetal): Metal
Kubernetes distro (e.g. RKE/K3s/EKS/OpenShift) and version: v1.24.10+rke2r1 Number of management node in the cluster: SINGLE NODE OS type and version: 5.14.0-162.12.1.el9_1.0.1.x86_64 SELinux enabled (true/false): Enforcing CPU per node: 92 Memory per node: 1024 Disk type(e.g. SSD/NVMe): NVMe Underlying Infrastructure (e.g. on AWS/GCE, EKS/GKE, VMWare/KVM, Baremetal): Metal
I see.
Will the helper pod work correctly if you disable the SELinux?
BTW, what's your OS distro (uname -a
)?
Confirmed, it's selinux. RHEL9.1 retail.
Confirmed, it's selinux. RHEL9.1 retail.
Thanks for your information. I will investigate this issue. It might be caused by the SELinux. You can disable the SELinux and check it can work. If yes, I can work on it with more confidence.
Yes, it is working now, and I have a replaced filesystem underneath (where it continues to work). Just need those environment variables to make this usable from a "real deployment" perspective. Every other user is going to encounter this on modern ELs.
@KyleSanderson Fixed in https://github.com/rancher/local-path-provisioner/pull/295. Will release 0.0.24 soon. Thank you.
Thank you. If the variables stuff could get some love https://github.com/rancher/local-path-provisioner/issues/291 that would be very appreciated. My NAS provider barely supports folder names longer than 30 characters.
Thank you. If the variables stuff could get some love #291 that would be very appreciated. My NAS provider barely supports folder names longer than 30 characters.
Sure. Will check it. Thanks for the feedback.
Thank you. If the variables stuff could get some love #291 that would be very appreciated. My NAS provider barely supports folder names longer than 30 characters.
The challenge of the issue is the backward compatibility. Need to investigate more.
Kind of unclear what's going on here as there's reports going back saying there's issues with OpenShift and similar. Retail RKE2 install on 1.24. Making the "helper" pod privileged creates the path, but every other pod still can't use it.