Unable to upgrade Helm Chart when drivers are not installed

[badnetmask]

I'm gonna be honest and say that I am not entirely sure how to deal with this problem, hence why I am not just sending a PR.

I am using the Helm Chart to install the operator, but I have driver.enable=false set. For whatever reason, the node status shows as Install-In-Progress, as you can see here:

$ kubectl get deviceconfig -n amd-gpu-operator default -o jsonpath='{.status.nodeModuleStatus}' | jq
{
  "fw1": {
    "status": "Install-In-Progress"
  }
}

Because of that I cannot do a clean Helm Chart upgrade, because there is a pre-upgrade hook blocking it. The only way to proceed with the upgrade is to uninstall then re-install the chart.

So, I'm not entirely sure if the fix to the problem problem should be modifying the node status to show as "Completed", or something like that, or to add better checks to the pre-upgrade hook. I'll leave that decision to the developers.

[yansun1996]

Hi @badnetmask , thanks for reporting this issue.

I've figured out that this is an issue that is triggered by the upgrademgr, who is responsible for upgrading the Operator managed driver, it is wrongly putting the status and make it stuck at Install-In-Progress even if we don't need to trigger it.

Internal tickets has been created to track this, we will provide a fix asap.

Could you please let us know the start version and target version you want to upgrade to ?

[badnetmask]

The problem happened while upgrading from v1.4.0 to v1.4.1.

[coltoneshaw]

Facing the same issue here. I see a PR was merged to address this. When will a new release be cut? thank you!

[yansun1996]

Facing the same issue here. I see a PR was merged to address this. When will a new release be cut? thank you!

Hi @coltoneshaw , yep we have merged the fix into main branch, new release ETA is around end of March or early April, stay tuned.