This message was deleted.
# harvester
a
This message was deleted.
g
(Harvester 1.4.0, in case that helps)
I also tried dropping a host into maintenance mode for this, but that doesn't seem to make any difference either. I'm guessing that there's a log somewhere that I can get more context from?
all I see from journalctl -b -f is
Copy code
Jan 20 12:35:12 harvester1 kernel: Console: switching to colour dummy device 80x25
PCIDeviceClaim YAML
Copy code
apiVersion: <http://devices.harvesterhci.io/v1beta1|devices.harvesterhci.io/v1beta1>
kind: PCIDeviceClaim
metadata:
  creationTimestamp: '2025-01-20T13:12:07Z'
  finalizers:
    - <http://wrangler.cattle.io/PCIDeviceClaimOnRemove|wrangler.cattle.io/PCIDeviceClaimOnRemove>
  generation: 1
  managedFields:
    - apiVersion: <http://devices.harvesterhci.io/v1beta1|devices.harvesterhci.io/v1beta1>
      fieldsType: FieldsV1
      fieldsV1:
        f:metadata:
          f:ownerReferences:
            .: {}
            k:{"uid":"abedf977-162c-4e67-bd17-a7fe7ef7186b"}: {}
        f:spec:
          .: {}
          f:address: {}
          f:nodeName: {}
          f:userName: {}
      manager: harvester
      operation: Update
      time: '2025-01-20T13:12:07Z'
    - apiVersion: <http://devices.harvesterhci.io/v1beta1|devices.harvesterhci.io/v1beta1>
      fieldsType: FieldsV1
      fieldsV1:
        f:metadata:
          f:finalizers:
            .: {}
            v:"<http://wrangler.cattle.io/PCIDeviceClaimOnRemove|wrangler.cattle.io/PCIDeviceClaimOnRemove>": {}
      manager: pcidevices
      operation: Update
      time: '2025-01-20T13:12:07Z'
  name: harvester1-000000020
  ownerReferences:
    - apiVersion: <http://devices.harvesterhci.io/v1beta1|devices.harvesterhci.io/v1beta1>
      kind: PCIDevice
      name: harvester1-000000020
      uid: abedf977-162c-4e67-bd17-a7fe7ef7186b
  resourceVersion: '10575175'
  uid: f85272ad-d07c-4812-8b3d-ad2315f69ad7
spec:
  address: '0000:00:02.0'
  nodeName: harvester1
  userName: admin
t
I had a similar issue where I needed to a. reboot, b. disable pic passthrough all together, C. reboot, D. reenable pci
g
thanks @thousands-advantage-10804 - I'll give that a shot for sure
Looks like it didn't even need to be disabled - I rebooted the host and it's now showing as available
t
nice
g
will see if I get the same on the other 3, if so, I'll file a bug, as this doesn't seem to match the documented behaviour
thanks Andy!
🙌 1
t
Yeah in my experience a couple reboots are/can be needed to get gpu passthrough to show up or work.