amazon-eks-user-guide
amazon-eks-user-guide copied to clipboard
Update s3-csi.md - Prerequisite
Updated one more prerequiste which I tested on my own as well.
Issue #, if available: Without EFS CSI Driver installed, S3 CSI Driver won't work and will throw below error:
0/2 nodes are available: pod has unbound immediate PersistentVolumeClaims. preemption: 0/2 nodes are available: 2 Preemption is not helpful for scheduling..
Description of changes: Addition of a point i.e., "While using S3 CSI Driver, we require EFS CSI Driver as storage won't be provisioned to get PersistentVolume mounted." I have tested this as well today and it's working.
Checkout related Issue: https://github.com/awslabs/mountpoint-s3-csi-driver/issues/185#issuecomment-2097980816
By submitting this pull request, I confirm that you can use, modify, copy, and redistribute this contribution, under the terms of your choice.
Thanks @Ramneek-kalra. Changes have been staged.
@Ramneek-kalra @chrisnegus Is this change suggesting that the EFS CSI driver is a prerequisites to using the S3 CSI driver?
The documentation here seems to suggest the EFS driver is not provisioned
This change shouldn't be required, Mountpoint S3 CSI Driver does not require EFS CSI Driver to be installed.
If somethings not working and installing EFS CSI Driver seems to help, please do open an issue on the Mountpoint S3 CSI Driver repository so the team can investigate. I saw there was already a comment in https://github.com/awslabs/mountpoint-s3-csi-driver/issues/185#issuecomment-2097980816 but we should have a dedicated issue to dig deeper, as we'd need logs to investigate.
I did some testing based on a Customer requirement who want to use S3 and EFS CSI Driver parallel in the same Cluster. The below Consideration seems invalid. Document "To get PersistentVolume mounted while using the Mountpoint for Amazon S3 CSI driver, we require that the Amazon EFS CSI driver won't be provisioned."
EKS Version: 1.29 S3-CSI-Driver: v1.7.0 EFS-CSI-Driver: v2.0.6
Only S3-CSI Driver Installed in the Cluster: POD is running fine. S3+EFS CSI both Driver Installed in the Cluster: Both POD using S3 and EFS running Fine. S3+EFS CSI Driver both Driver Installed in the Cluster: Both POD using S3 and EFS running Fine in the Same WorkerNode as well.
I don't see any relation between S3 and EFS CSI Driver.
Thank you @towhidul1987, I confirmed with several people that this statement was misleading and it has been removed. The main documentation website will have that consideration removed in about 2 hours.