dell / csm

Dell Container Storage Modules (CSM)
Apache License 2.0
68 stars 15 forks source link

[FEATURE]: Enable/disable automatic SDC deployment along with driver installation. #663

Closed anurajpd closed 3 weeks ago

anurajpd commented 1 year ago

Bug Description

Powerflex csi driver installation failing when sdc image is not provided in the values file, when using the manual sdc install.

Logs

values.yaml version: v2.5.0 images: driverRepository: dellemc powerflexSdc: certSecretCount: 0 logLevel: "debug" logFormat: "TEXT" kubeletConfigDir: /var/lib/kubelet defaultFsType: ext4 imagePullPolicy: IfNotPresent fsGroupPolicy: File controller: volumeNamePrefix: csi-k8s02 controllerCount: 2 snapshot: enabled: true resizer: enabled: true

---- install error ------

Error: INSTALLATION FAILED: execution error at (csi-vxflexos/templates/node.yaml:287:20): Must provide the PowerFlex SDC container image.

Screenshots

No response

Additional Environment Information

No response

Steps to Reproduce

Install the powerflex csi driver using helm and remove the sdc image from teh values file for manual sdc deployment

Expected Behavior

Should not try to install the sdc as manual sdc installation is used and the sdc is already installed

CSM Driver(s)

CSI Driver Powerflex 2.5

Installation Type

Helm

Container Storage Modules Enabled

No response

Container Orchestrator

Kubernetes 1.25.6

Operating System

RHEL 8.4

csmbot commented 1 year ago

@anurajpd: Thank you for submitting this issue!

The issue is currently awaiting triage. Please make sure you have given us as much context as possible.

If the maintainers determine this is a relevant issue, they will remove the needs-triage label and assign an appropriate priority label.


We want your feedback! If you have any questions or suggestions regarding our contributing process/workflow, please reach out to us at ses.csm.engineering@dell.com.

bharathsreekanth commented 1 year ago

@anurajpd Can you confirm the host OS in this case ?

bharathsreekanth commented 1 year ago

It appears that this is due to a day1 decision when SDC was made available as an init container. Although SDC init container was only supporting subset of host OSes, the init container would gracefully exit on non-supported OS. The current architecture would have to be tweaked to enable this. https://github.com/dell/csm/issues/992 is related to this enhancement and hence closing this as this will be prioritized in our backlog accordingly and addressed in a future release.

bharathsreekanth commented 1 year ago

It appears that there is a bit more to https://github.com/dell/csm/issues/992 than just mandating SDC installation, re-opening. The crux of this requirement is now to make SDC installation via init container optional. On environments where SDC is manually installed, users should be able to proceed with driver install without providing SDC image value.

rbo54 commented 1 year ago

Hi Bharath, that addresses part of the issue, a way to disable attempting the SDC install. I've always done manual install. However the first time I tried the new Dell csm-helm chart, I let it try to install, and it failed. I just used whatever default images is in the chart. I don't know where to get the images from (I didn't see that documented in the instructions.) The default image in the chart: dellemc/sdc:3.6.1 is probably way too old. I have newer RHEL versions 8.7 and 9.0 and I find I have to run something like EMC-ScaleIO-sdc-3.6-2000.117.el8.x86_64. I've taken to getting these out of the Powerflex build repositories. Sometimes I have to try multiple ones to find one that works.

rbo54 commented 1 year ago

The image selection needs to be dynamic based on the OS and OS version. And only the Powerflex team knows for sure AFIK what image runs with what OS.

bharathsreekanth commented 11 months ago

The image selection needs to be dynamic based on the OS and OS version. And only the Powerflex team knows for sure AFIK what image runs with what OS.

There is enhancement coming in future from PowerFlex team (probably Q2 '24 and we will need to make some edits on our end to enable that) that will help compile an SDC image dynamically based on the host OS kernels that will solve this particular problem. The latest helm charts update as part of CSM 1.9 will include the correct SDC image, v4.5, that is compatible with all the current versions we support. The charts are updated in the release-1.9 branch, and will be part of main once released.

bharathsreekanth commented 11 months ago

I propose closing this as this is technically not a defect and was a design consideration at the time of initial CSI driver deployment. Those considerations need to be revisited in light of newer enhancements to SDC and CSI Driver over time. https://github.com/dell/csm/issues/992 tracks the enhancements, so proposing to close this out. @rbo54 @anurajpd thoughts?

Prabhu-Dell commented 11 months ago

Reopening and considering this as feature to address following usecases.

suryagupta4 commented 2 months ago

link: 19353

gallacher commented 2 months ago

@Prabhu-Dell, please update the description to reflect that this is a feature and not a bug. Also, please ensure the description is clear and concise, reflecting the value this feature will deliver. Thanks!

tdawe commented 3 weeks ago

Feature completed