I installed harvester via DEV method.

I applied harvester-pcidevices-controller to activate GPU-passthrough function. I installed windows server R12 and attached a nvidia 2080 Ti to this VM. It works. Why do I use window server 2012? Because I can't attach nvidia GPU to windows 10 VM which stuck at 'starting' phase, some libvirt error in logs as well. By the time, it's ok. Now when i backup this 2012 server VM and try 'restore for new' function, I get a blue screen from web console, meanwhile, via window server rdp(3389) i'm able to reach the VM without any proble.

this blue screen is unable to response to ctrl-alt-del.

i wanna web console comes back. The original VM works fine, web console + windows remote desktop + nvidia 2080 Ti working( nvidia-smi output ok). Why backup/restore break it?

trying to get harvester on a node that will pxe boot 100% of the time (or can lay down a disk image). I don't think harvester supports this deployment method right now the system is in a reboot/install loop, re-installing constantly...

is there a way to generate a disk image for harvester itself, say to boot in a virtual env or otherwise (dd to disk, basically)

Is it possible to install harvester VMs on top of harvester cluster? Could be very good for tests later.

I would like to test Harvester in my Lab. Currently i’ve only a lot of raspberry pis. Due the fact that harvester currently dosn’t run on ARM, can some recommend some intel nuc or other mini pcs that are suitable for a harvester lab environment? Otherwise is there currently already an alpha or dev release that runs on arm?

Is there already CHANGELOG for upcoming v1.1.2 version ? Thanks

Hi everyone. Trying to ask a question of the people who build the harvester terraform module. Is that a whole separate thing, or part of this community?

Hello, I came across this issue and need your help, I’ve set up Harvester v1.1.1 and registered it in Rancher, 2.7.1 then I provisioned a new RKE cluster, using Harvester. Harvester created the VMs, deployed masters and workers according to the pools I specified, and the cluster was created successfully. Then I added an additional pool for ingress (I made sure to disable the built-in ingress during the cluster creation) and deployed ingress using Flux and Helm The ingress pod is running (with node selector and taints for the ingress pool) and there are no errors. However I’m not able to access the ingress service in 443/80 or any other port beside 22. It seems that the ip 192.168.226.X which is the VM ip on which the ingress is running only opened to TCP 22 (SSH) I should mention that Linkerd is also installed but I see the same issue without Linkerd. the ingress ports are not exposed. I expected that if I will curl “http://192.168.226.x” then I’ll get 404 but the port is not even opened. What am i missing here ? I should also mentioned that without Harvester, same RKE + ingress + Linkerd works fine for me

Hello, There are lots of helm operation pod in error status.

helm upgrade --history-max=5 --install=true --namespace=cattle-system --reset-values=true --timeout=5m0s --values=/home/shell/helm/values-rancher-webhook-2.0.0-up0.3.0.yaml --version=2.0.0+up0.3.0 --wait=true rancher-webhook /home/shell/helm/rancher-webhook-2.0.0-up0.3.0.tgz 2023-04-06T124113.408410296Z Error: UPGRADE FAILED: another operation (install/upgrade/rollback) is in progress

I'm wondering which trigger this operation.

Hi folks, I just installed Harvester bare metal on one of my homelab machines for evaluation and I'm really excited at the possibilities! It is a clean install and I'm having an issue pulling images from a URL. Since it is a clean install at the latest version, I'm not sure if it is an issue with my configuration or a bug. Wanted to do a sanity check here before filing a bug. When I go to download an image, I get the error

Head "<https://github.com/rancher/k3os/releases/download/v0.20.6-k3s1r0/k3os-amd64.iso>": x509: certificate is not valid for any names, but wanted to match <http://github.com|github.com>

This happens for different domains and images as well. If I try with just the http protocol, I get

Got 404 status code from <http://github.com/rancher/k3os/releases/download/v0.20.6-k3s1r0/k3os-amd64.iso>

. No VMs are running. I have the default management network and a VM network (both are reporting as active). It is connected to my home LAN with nothing fancy like proxies going on. In advanced settings, I don't have anything set for

additional-ca

,

http-proxy

,

ssl-certificates

,

ssl-parameters

, or

vip-pools

. I'm not really sure how to debug this further. Is this a bug that I should file or is there an issue with my configuration?

is there an option when running the installer (kernel command line or otherwise) to add the node to an existing cluster as a controlplane node. instead of just a worker node. say when adding as a second or third node

Loadbalancer question ... it seems that Harvester will create loadbalancer on management network, how do I tell him to create load balancer on custom "oob" network attached to VMs that are used to host RKE2 cluster?

We are trying to enable KMS for secrets encryption using AWS KMS Provider or trousseau on a harvester install. We got the provider running and are ready to commit the encryption-configuration. For that we understand we need to pass a flag the api-service. This is where we are stuck. Can anyone help here?

Hi all, Is it possible to schedule VM backups in Harvester to an S3 or NFS storage?

I have gone through many iterations to get rancher manager to manage a harvester hosted kubernetes cluster. Due to disk pressure and not wanting to have to continually access the 'secret' longhorn UI to change the number of replicas I have added my third Intel NUC like server as a harvester 1.1.1 node and am installing Rancher Manager 2.6.9 as a docker container on an Ubuntu 20.04.6 VM within the harvester cluster. Rancher has successfully imported the harvester cluster using virtualization management but when using cluster management to create a new kubernetes cluster on harvester the UI presents me with a "clusters.management.cattle.io "v1" not found" when trying to retrieve information (such as the image to use for the machine pool) and the associated operation in developer tools is "GET https://rancher.roguestarr.com/k8s/clusters/v1/harvesterhci.io.virtualmachineimages 404 (Not Found) I am exploring options for official support from SUSE as I have taken up enough of your time here on slack that deserves some $$$ to get back into your pockets to make this the vmware killer that I am truly hoping it will become. In the meantime if anyone can see from the attached logs what I may have been doing wrong it would be greatly appreciated.

Hi, I'm trying to get Harvester installed on a diskless baremetal host - the boot disk is a 300g block volume delivered over iSCSI - I tried a couple of different ways of doing this - a. If I try network boot the installer and then switch over the the command line and mount the iscsi volume - the installer sees it and attempts to install to it but hangs during the install process. b. If I pre-install Harvester in a VM where the block volume appears like a local disk and then attach this block volume over iSCSI to the BM host - the boot process fails trying to switch root. Any tips on how to get this to work would be appreciated, thanks.

👋 Folks, does the

harvester-csi-driver

support CSI Snapshots? I've been trying to make these work but no luck so far 😕 This is my volume snapshot class:

apiVersion: <http://snapshot.storage.k8s.io/v1|snapshot.storage.k8s.io/v1>
kind: VolumeSnapshotClass
metadata:
  name: default
driver: <http://driver.harvesterhci.io|driver.harvesterhci.io>
deletionPolicy: Retain
parameters:
  type: bak

(I also tried

<http://driver.longhorn.io|driver.longhorn.io>

) And I test it with

apiVersion: <http://snapshot.storage.k8s.io/v1|snapshot.storage.k8s.io/v1>
kind: VolumeSnapshot
metadata:
  name: test-snapshot
spec:
  volumeSnapshotClassName: default
  source:
    persistentVolumeClaimName: home-assistant-data-home-assistant-0

But it just gets stuck in waiting for snapshot to be created by csi driver:

2023-04-16 12:00:54	I0416 19:00:54.824012       1 snapshot_controller.go:638] createSnapshotContent: Creating content for snapshot home-assistant/test-snapshot through the plugin ...
2023-04-16 12:00:54	I0416 19:00:54.837378       1 snapshot_controller.go:918] Added protection finalizer to persistent volume claim home-assistant/home-assistant-data-home-assistant-0
2023-04-16 12:00:54	I0416 19:00:54.847442       1 event.go:285] Event(v1.ObjectReference{Kind:"VolumeSnapshot", Namespace:"home-assistant", Name:"test-snapshot", UID:"2bf391e6-ff0f-466b-97d7-2d2a13ef2342", APIVersion:"<http://snapshot.storage.k8s.io/v1|snapshot.storage.k8s.io/v1>", ResourceVersion:"32744110", FieldPath:""}): type: 'Normal' reason: 'CreatingSnapshot' Waiting for a snapshot home-assistant/test-snapshot to be created by the CSI driver.
2023-04-16 12:00:54	I0416 19:00:54.861963       1 snapshot_controller.go:964] Keeping PVC home-assistant/home-assistant-data-home-assistant-0, it is used by snapshot home-assistant/test-snapshot

hi team, I running a docker container (specifically a kind k8s cluster) inside a vm created by harvester. I noticed that the network inside the container are pretty slow (but the network is fine in the vm layer.) anyone has experienced similar issues before?

Hi 👋 Any ideas why the VM metrics in Harvester report a almost 100% memory utilization while my kubernetes node-exporter reports way less? • Harvester: 32.5 GiB usage • Kube Node Exporter: around 10 GiB usage

I tried to restore a vm backup from one instance to a new cluster, but it say backing image not found, however I know what is the backing image ( ubuntu cloud ), is there a way to restore the backup ?

Another question, I use harvester 1.1.1 with longhorn default install version ( 1.3.2 ), in longhorn dashboard it say new version available ( 1.4.1 ), can I upgrade or it have breaking changes ?

Are peoples using harvester in production cluster ?

Do we have a date or estimate for a new harvester version with latest things in it ?

Hey everyone. I am using Rancher 2.7.2 and I have a harvester 1.1.1 (so both latest releases). But I can't select Harvester as a cloud provider when creating an RKE2 cluster:

Hello everyone. I am standing up a new Harvester cluster and one of my node keeps joining with blank storage. I have 7 nodes, I delete the node and rebuild it but the nodes keeps joining with no storage… anything. I join it with the same name everytime so I am not sure it there is an issue with the control nodes hanging on to some info or if there is an issue with the node itself. I have attached screen shots: