Endpoint with scheme problem - storage class provisions, but unable to mount

[davidpechcz]

geesefs -v
geesefs version 0.35.4

We are unable to mount volume correctly after being provisioned, we are using helm chart and storageClass:

mounter: geesefs
options: ' --no-systemd --memory-limit 1000 --dir-mode 0777 --file-mode 0666 --debug
--debug_fuse --debug_s3'

and secret.endpoint: https://fradozn9lozb.compat.objectstorage.eu-frankfurt-1.oraclecloud.com - this provisions the bucket, but does not mount it correctly - with this error:

Jun 23 13:58:42 metro-acc-node1 kubelet[885]: E0623 13:58:42.082579 885 nestedpendingoperations.go:348] Operation for "{volumeName:kubernetes.io/csi/ru.yandex.s3.csi^makro-bin-acc/pvc-6977b81b-884b-43c4-84f1-fafcca195753 podName: nodeName:}" failed. No retries permitted until 2023-06-23 14:00:44.082522884 +0000 UTC m=+195623.760539894 (durationBeforeRetry 2m2s). Error: MountVolume.MountDevice failed for volume "pvc-6977b81b-884b-43c4-84f1-fafcca195753" (UniqueName: "kubernetes.io/csi/ru.yandex.s3.csi^makro-bin-acc/pvc-6977b81b-884b-43c4-84f1-fafcca195753") pod "mocz-nginx-c68bf674b-zgrk6" (UID: "3d6c8e35-22b4-4bb2-8148-1e661965a8f2") : rpc error: code = Unknown desc = Error fuseMount command: geesefs
Jun 23 13:58:42 metro-acc-node1 kubelet[885]: args: [--endpoint https://fradozn9lozb.compat.objectstorage.eu-frankfurt-1.oraclecloud.com -o allow_other --log-file /dev/stderr --setuid 65534 --setgid 65534 --memory-limit 1000 --dir-mode 0777 --file-mode 0666 makro-bin-acc:pvc-6977b81b-884b-43c4-84f1-fafcca195753 /var/lib/kubelet/plugins/kubernetes.io/csi/ru.yandex.s3.csi/7ca88b01d8604c6cbabde677ea3418367b1a9047aa128a8139988732712f0046/globalmount]
Jun 23 13:58:42 metro-acc-node1 kubelet[885]: output:

All examples contain scheme in the endpoint. When we try to use endpoint without 'http://', we get an error before provisioning: (kubelet):
Jun 23 21:14:42 metro-acc-node1 kubelet[885]: E0623 21:14:42.071632 885 nestedpendingoperations.go:348] Operation for "{volumeName:kubernetes.io/csi/ru.yandex.s3.csi^makro-bin-acc/pvc-e7ca2fb3-43e6-4140-b7b4-3bb3240b195d podName: nodeName:}" failed. No retries permitted until 2023-06-23 21:16:44.071594657 +0000 UTC m=+221783.749611665 (durationBeforeRetry 2m2s). Error: MountVolume.MountDevice failed for volume "pvc-e7ca2fb3-43e6-4140-b7b4-3bb3240b195d" (UniqueName: "kubernetes.io/csi/ru.yandex.s3.csi^makro-bin-acc/pvc-e7ca2fb3-43e6-4140-b7b4-3bb3240b195d") pod "mkcz-nginx-5ffbb57f8b-mzwk6" (UID: "a33aee50-1a1c-4960-bcd9-25c10b3c8028") : rpc error: code = Unknown desc = failed to initialize S3 client: Endpoint: does not follow ip address or domain name standards.

From inside container (kubectl exec kube-system/csi-s3-9x685, container csi-s3) - when trying manually:

endpoint with "https" does not work, without it - works.

/ # export AWS_ACCESS_KEY_ID=XX
/ # export AWS_SECRET_ACCESS_KEY=YY
/ # geesefs --endpoint https://fradozn9lozb.compat.objectstorage.eu-frankfurt-1.oraclecloud.com -o allow_other --log-file /dev/stderr --setuid 65534 --setgid 65534 --
memory-limit 1000 --dir-mode 0777 --file-mode 0666 makro-bin-acc:pvc-e7ca2fb3-43e6-4140-b7b4-3bb3240b195d /var/lib/kubelet/plugins/kubernetes.io/csi/ru.yandex.s3.csi/
f03642a6dc198d3dd3165af841807f18095bf8eebcdf5c749940438e6cffa049/globalmount
2023/06/23 21:36:32.759654 main.ERROR Unable to access 'makro-bin-acc': bucket makro-bin-acc does not exist
2023/06/23 21:36:32.759813 main.FATAL Mounting file system: Mount: initialization failed
2023/06/23 21:36:33.760127 main.FATAL Unable to mount file system, see syslog for details

without 'https' -> works correctly:

/ # geesefs --endpoint fradozn9lozb.compat.objectstorage.eu-frankfurt-1.oraclecloud.com -o allow_other --log-file /dev/stderr --setuid 65534 --setgid 65534 --memory-l
imit 1000 --dir-mode 0777 --file-mode 0666 makro-bin-acc:pvc-e7ca2fb3-43e6-4140-b7b4-3bb3240b195d /var/lib/kubelet/plugins/kubernetes.io/csi/ru.yandex.s3.csi/f03642a6
dc198d3dd3165af841807f18095bf8eebcdf5c749940438e6cffa049/globalmount
2023/06/23 21:36:43.590764 s3.ERROR Unable to access 'makro-bin-acc': unsupported protocol scheme ""
2023/06/23 21:36:43.800459 main.INFO File system has been successfully mounted.

[Reonaydo]

I have same error. I am unable to use geesefs with csi driver. Driver requires endpoint with schema but geesefs able to mount only without schema

[Reonaydo]

I've found a strange workaround for me.
I edited csi-s3-secret and added region: us-east-1 (base64 encrypted of course)

kubectl -n kube-system edit secrets csi-s3-secret

Added
region: dXMtZWFzdC0x

And geesefs started working

p.s. I installed csi driver via helm and it's strange that there no region variable in the helm template