rancher-users #harvester-dev

harvester-dev

quiet-carpenter-8833

09/07/2022, 3:01 PM

Hi everyone, are there any docs on how to renew the self-signed harvester certificate for a single node harvester deployment? Unable to launch any VMs as the initial cert has expired.

quiet-carpenter-8833

09/07/2022, 3:02 PM

The node in question is a harvester 1.0.2 upgraded to 1.0.3, TLS secrets are now 110d old

limited-breakfast-50094

09/12/2022, 4:58 PM

Hey @millions-restaurant-90347

limited-breakfast-50094

09/12/2022, 4:59 PM

I'm working on applying your workaround to my kubevirt setup

millions-restaurant-90347

09/12/2022, 5:00 PM

Batteries included. Just apply that yaml (it's a common mutating webhook).

👍 1

limited-breakfast-50094

09/12/2022, 5:27 PM

Harvester cluster booted up and ready. Going to try applying the yamll

future-address-23425

09/12/2022, 5:35 PM

Do you try this on harvester master or 1.0.3?

future-address-23425

09/12/2022, 5:36 PM

It should work on both (kubevirt versions) anyway.

limited-breakfast-50094

09/12/2022, 5:37 PM

I am using a version of 1.0.3 with my own changes on top

limited-breakfast-50094

09/12/2022, 5:37 PM

Although I have been running into this nodeaffinity error

(combined from similar events): Error creating pod: Pod "virt-launcher-pcitest2-rt8x4" is invalid: spec.affinity.nodeAffinity.requiredDuringSchedulingIgnoredDuringExecution.nodeSelectorTerms: Required value: must have at least one node selector term

There's no PCI device attached yet

limited-breakfast-50094

09/12/2022, 5:38 PM

In the UI I had it assigned to a specific node, but I keep getting this FailedCreate error

future-address-23425

09/12/2022, 5:43 PM

If you don't have a reason to select a specific node, why not let the device plugin do it for you? It looks like a validation error. We haven't come across this.

future-address-23425

09/12/2022, 5:44 PM

If on every node the devices of your selector are assigned to

vfio-pci

, you won't have a problem wherever it sits.

limited-breakfast-50094

09/12/2022, 5:45 PM

Yeah, I was trying to keep it on the one host with gpus, but I can test PCI passthrough on either (the other host in my 2 node cluster has a 10Gb PCI NIC

limited-breakfast-50094

09/12/2022, 5:54 PM

Thread for Kubevirt hack (to avoid filling up the channel)

future-address-23425

09/14/2022, 6:09 PM

I was trying to enable the

vfio-pci

driver through the Harvester configuration, using this:

os:
  modules:
  - vfio-pci

, with no luck. I made it work with:

os:
  write_files:
  - content: |
      vfio-pci
    path: /etc/modules-load.d/vfio-pci.conf

Any thoughts, am I missing sth? Shouldn't the first one work (do the same thing)? I believe that

vfio-pci

should be anyway enabled by default.

crooked-scooter-58172

09/21/2022, 5:49 PM

Team: I need to build a harvester client in go language which can create VM. Any suggestions pls?

crooked-scooter-58172

10/18/2022, 5:30 PM

Team: Any idea from where I can download the kubevirt-->VirtualMachine CRD? Thanks in Advance!!

crooked-scooter-58172

10/20/2022, 3:59 PM

Team: We used to have an API docs in the Harvester Site. I am not finding anymore. Anyone know the new location?

crooked-scooter-58172

11/04/2022, 2:47 AM

@here: We need to implement the the version management for VM images. Just checking what tool and process you guys are using to store VM images (I.e. git…etc)

mysterious-boots-59774

11/13/2022, 5:57 PM

You can store VM images in a S3 server that’s local to Harvester cluster. On the Harvester side, internally the VM images are stored in Longhorn.

crooked-scooter-58172

11/17/2022, 9:50 PM

@Team: Any idea what is the use of the "vip-pools" under "settings" ?

crooked-scooter-58172

11/18/2022, 5:59 PM

Team: Do we have any production ready harvester-go-client which we can use to create Virtual Machines from my Go Code?

crooked-scooter-58172

11/21/2022, 7:20 PM

Team: I have a two node cluster. My primary (First) Node went down. However, the secondary node is not picking-up and showing "Not ready" in console. Is this a limitation of harvester? Isn’t a single point of failure? Any suggestion on where to look into logs? Looks like a complete down?

crooked-scooter-58172

12/07/2022, 7:47 PM

Team, It seems that Harvester doesn’t consider “_*requests.memory*_” param while assigning memory for a VM. It always use the limit memory (minus some overhead) and assign that much of memory to the VM. I am just wondering whether others faced the similar issue. If yes, then any suggestions?

crooked-scooter-58172

12/08/2022, 2:33 PM

Hello team, I have downloaded the harvesterhci source code. Just wondering whether anyone can direct me to the function which gets called when Harvester UI triggers a create VM call. Thanks in advance.

crooked-scooter-58172

12/21/2022, 7:38 PM

Team, one of our nodes is keep giving below error. Any advice please? Failed to create pod sandbox: rpc error: code = Unknown desc = failed to get sandbox image "index.docker.io/rancher/pause:3.6": failed to pull image "index.docker.io/rancher/pause:3.6": failed to pull and unpack image "docker.io/rancher/pause:3.6": failed to resolve reference "docker.io/rancher/pause:3.6":

crooked-scooter-58172

02/07/2023, 10:45 PM

We are using a 3-node cluster for our development. Setup is working fine, however sometime a specific Node is keep getting into “*Cordoned*” state with the message “*Kubelet stopped posting node status*”. I tried a lot but haven't been able to nail down the actual issue with that specific Node. Can someone please help me with some pointers towards the troubleshooting? Do we have any tool I can use for the troubleshooting?

witty-honey-18052

02/14/2023, 2:27 PM

Hi folks, just wondering if I could ask a couple of UI-dev related questions? Having a couple of challenges with Extensions, and trying to follow the harvester extension work. Seems Vue version related. Hoping that some of the harvester dev expertise can help put 2 and 2 together there (there is no channel for extensions yet)

crooked-scooter-58172

02/26/2023, 7:31 PM

After a few weeks of running Harvester, we started getting Kubelet disk pressure errors. When looking further we noticed that there were >50GBs generated by thousands of tmp directories in the /usr/local/.state/var-lib-rancher.bind/agent/tmp directory. Just checking whether anyone faced the similar issue? Any idea on how to fix this?