cancel
Showing results for 
Search instead for 
Did you mean: 

Edge Integration Cell Setup with AWS

HuyNguyen
Explorer
0 Kudos
781

Hi

I setup Edge Integration Cell with AWS Kubenetest Cluster (EKS). When run ELM.exe I get issue:

Validation Error The check for "JSON_POSIX_COMPLIANT_STORAGE_CLASSES" failed. Details No usable storage class found. Configure a storage class with a POSIX-compliant file system:
ext3,ext4,xfs,zfs,btrfs
For analysis and known fixes for failures related to "JSON_POSIX_COMPLIANT_STORAGE_CLASSES" see the Troubleshooting Guide
(https://help.sap.com/viewer/elm_bridge_troubleshooting/cd2c59cdb1a144d692721fe63335dc5d.html)

This is my configure AWS cluster with storage class

1/ Add on Amazone EFS CSI Driver

2/ Storage Class 

HuyNguyen_0-1715942875022.pngHuyNguyen_1-1715942903957.png

3/ Issue here

HuyNguyen_2-1715942954499.png

Kindly help to fix this issue.

Thanks.

 

 

Accepted Solutions (0)

Answers (2)

Answers (2)

istoykov
Associate
Associate
0 Kudos

Hi,

when you register an Edge Node with monitoring and logging via Edge Lifecycle Management, we will deploy Prometheus as part of our application stack. 

Prometheus requires a persistence volume with POSIX compliant filesystem. The Amazon's EFS is an NFS type file system (NFSv4 protocol). Despite Amazon's claims that it is POSIX compliant, Prometheus have stated that they do not support it.

This is the reason we don't support it and you get the "No usable storage class found" error.

In order to register an Edge Node, you have to set up a storage class, which can provision POSIX compliant volumes.

Best regards,
Ivaylo

MattiaPerego
Associate
Associate
0 Kudos

Hi @HuyNguyen ,did you finally make it work or are you still struggling with this error?

Let us know

Regards

Mattia

HuyNguyen
Explorer
0 Kudos

Dear Mattia

I have not fixxed yet. Any advise ?

MattiaPerego
Associate
Associate
0 Kudos

In the end we were able to solve the issue (not on AWS but on SUSE Rancher in a private landscape).

The issue was related to cluster configuration that wasn't fully compliant with longhorn requirements with SUSE Rancher. My suggestion is to double check the cluster configuration to verify if everything is fine.