System-upgrade-controller not working on Windows nodes

Rancher 2.x
1

Hi,

I have deployed a custom RKE2 cluster from Rancher 2.6.8. I’m having issues with rancher/system-upgrade-controller:v0.9.1 on Windows worker nodes, it doesn’t complete.

I would like to completely remove or disable upgrades on that nodes.

I have tried to scale down system-upgrade-controller to 0 instances, delete upgrade plans, etc… even deleting deployment but it gets recreated every few seconds.

Is there any way to disable it or prevent it from running on Windows nodes?

Step failing is rancher/wins:v0.4.7. At logs I can see this:

Creating C:\etc\rancher\wins
Sun, Oct 30 2022 12:41:32 pm Write-EventLog : The source name “rancher-wins” does not exist on computer
Sun, Oct 30 2022 12:41:32 pm “localhost”.
Sun, Oct 30 2022 12:41:32 pm At C:\run.ps1:103 char:5
Sun, Oct 30 2022 12:41:32 pm + Write-EventLog -LogName “$LogName” -Source “$Source” -Message "$M …
Sun, Oct 30 2022 12:41:32 pm + ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Sun, Oct 30 2022 12:41:32 pm + CategoryInfo : InvalidOperation: ( : ) [Write-EventLog], InvalidO
Sun, Oct 30 2022 12:41:32 pm perationException
Sun, Oct 30 2022 12:41:32 pm + FullyQualifiedErrorId : Microsoft.PowerShell.Commands.WriteEventLogComma
Sun, Oct 30 2022 12:41:32 pm nd

Thanks!

/Marcos

2

Found a workaround for making Rancher stop trying upgrade Windows nodes.

Add label to Windows workers

key: plan.upgrade.cattle.io/system-agent-upgrader-windows
value: random id from upgrade plan

This way Rancher upgrader thinks nodes are up to date and doesn’t try it again.

Not a solution but at least it stopped creating hundreds of PODs.

If anyone can help with a proper solution (making upgrader work on Windows nodes) I would appreciate it.

Thanks

/Marcos

3

I have the same problem, in the mixed cluster(Linux&Windows nodes) the cluster is stuck in update status. The error in Windows node:

error applying plan – check rancher-system-agent.service logs on node for more information