一、软硬件信息
1.服务器厂家:
2.沐曦GPU型号：C500
3.操作系统内核版本：5.15.0-119-generic
4.是否开启CPU虚拟化：
lsmod | grep kvm
kvm_intel 372736 0
kvm 1036288 1 kvm_intel
5.mx-smi回显：

mx-smi
mx-smi version: 2.3.1

=================== MetaX System Management Interface Log ===================
Timestamp : Thu Jul 16 17:12:01 2026

Attached GPUs : 8
+---------------------------------------------------------------------------------+
| MX-SMI 2.3.1 Kernel Mode Driver Version: 3.9.10 |
| MACA Version: 3.7.2.0 BIOS Version: 1.33.5.0 |
|------------------+-----------------+---------------------+----------------------|
| Board Name | GPU Persist-M | Bus-id | GPU-Util sGPU-M |
| Pwr:Usage/Cap | Temp Perf | Memory-Usage | GPU-State |
|==================+=================+=====================+======================|
| 0 MetaX C500 | 0 Off | 0000:06:00.0 | 0% Disabled |
| 71W / 350W | 46C P9 | 61938/65536 MiB | Available |
+------------------+-----------------+---------------------+----------------------+
| 1 MetaX C500 | 1 Off | 0000:07:00.0 | 0% Disabled |
| 74W / 350W | 46C P9 | 63240/65536 MiB | Available |
+------------------+-----------------+---------------------+----------------------+
| 2 MetaX C500 | 2 Off | 0000:08:00.0 | 0% Disabled |
| 71W / 350W | 46C P9 | 63238/65536 MiB | Available |
+------------------+-----------------+---------------------+----------------------+
| 3 MetaX C500 | 3 Off | 0000:09:00.0 | 0% Disabled |
| 75W / 350W | 45C P9 | 61940/65536 MiB | Available |
+------------------+-----------------+---------------------+----------------------+
| 4 MetaX C500 | 4 Off | 0000:95:00.0 | 0% Disabled |
| 68W / 350W | 39C P9 | 61942/65536 MiB | Available |
+------------------+-----------------+---------------------+----------------------+
| 5 MetaX C500 | 5 Off | 0000:96:00.0 | 0% Disabled |
| 70W / 350W | 39C P9 | 63238/65536 MiB | Available |
+------------------+-----------------+---------------------+----------------------+
| 6 MetaX C500 | 6 Off | 0000:97:00.0 | 0% Disabled |
| 69W / 350W | 39C P9 | 63234/65536 MiB | Available |
+------------------+-----------------+---------------------+----------------------+
| 7 MetaX C500 | 7 Off | 0000:98:00.0 | 0% Disabled |
| 70W / 350W | 40C P9 | 61940/65536 MiB | Available |
+------------------+-----------------+---------------------+----------------------+

+---------------------------------------------------------------------------------+
| Process: |
| GPU PID Process Name GPU Memory |
| Usage(MiB) |
|=================================================================================|
| 0 673394 sglang::schedul 61054 |
| 1 673395 sglang::schedul 62356 |
| 2 673396 sglang::schedul 62354 |
| 3 673397 sglang::schedul 61056 |
| 4 673398 sglang::schedul 61058 |
| 5 673399 sglang::schedul 62354 |
| 6 673400 sglang::schedul 62350 |
| 7 673401 sglang::schedul 61056 |
+---------------------------------------------------------------------------------+

End of Log

二、问题现象

C500节点无法被正确打标

kubectl logs -f metax-gpu-label-27gfr -n metax-operator
Defaulted container "gpu-labelgen" out of: gpu-labelgen, init-log-dir (init)
{"level":"info","ts":"2026-07-16T08:48:19Z","logger":"label-generator.entry","msg":"Starting manager"}
{"level":"info","ts":"2026-07-16T08:48:19Z","logger":"controller-runtime.metrics","msg":"Starting metrics server"}
{"level":"info","ts":"2026-07-16T08:48:19Z","logger":"controller-runtime.metrics","msg":"Serving metrics server","bindAddress":":8080","secure":false}
{"level":"info","ts":"2026-07-16T08:48:19Z","msg":"Starting EventSource","controller":"node","controllerGroup":"","controllerKind":"Node","source":"kind source: *v1.Node"}
{"level":"info","ts":"2026-07-16T08:48:19Z","msg":"Starting Controller","controller":"node","controllerGroup":"","controllerKind":"Node"}
{"level":"info","ts":"2026-07-16T08:48:19Z","msg":"Starting workers","controller":"node","controllerGroup":"","controllerKind":"Node","worker count":1}
{"level":"info","ts":"2026-07-16T08:48:19Z","logger":"label-generator.UpdateEvent","msg":"label changed","diff":{"metax.com/gpu.installed":"nil => false"}}

operator版本：
NAME NAMESPACE REVISION UPDATED STATUS CHART APP VERSION
metax-operator metax-operator 4 2026-07-15 11:42:51.857938402 +0800 CST deployed metax-operator-0.15.2 0.15.2

helm value配置：

metax-operator values

Registry for container images

registry: "hub.xxx.com/metax"

Driver payload configuration

driver:
payload:
version: 3.8.0.10-amd64

Maca payload configuration

maca:
payload:
registry: "hub.sudoinfotech.com/metax"
images:
- "maca:3.8.0.11-ubuntu22.04-amd64"

vendor:
vendorID: "9999"
driver: "metax"
virtDriver: "vfio-pci"
charDev: "mxcd"
domain: "metax.com"

openshift:
deploy: false

尊敬的开发者您好，请参考developer.metax-tech.com/api/client/document/preview/1273/k8s/01_scheduler.html#gpu-scheduler

helm install ./k8s/metax-operator-0.15.1.tgz \
  --create-namespace -n metax-operator \
  --generate-name \
  --wait \
  --set registry=DOMAIN/PROJECT \
  --set gpuScheduler.deploy=true

请问一下 0.15.2版本是暂时不支持volcano的方式部署吗？

gpu scheduler安装后gpu label依然无法正常打标：

kubectl get pods -n metax-operator
NAME READY STATUS RESTARTS AGE
metax-gpu-label-5l7cd 1/1 Running 0 5m8s
metax-gpu-label-cgz9p 1/1 Running 0 4s
metax-gpu-scheduler-bdcc887d5-l8rfx 2/2 Running 0 51s
metax-operator-64cf4b48b8-7fbhx 1/1 Running 0 5m26s
kubectl logs -f metax-gpu-label-cgz9p -n metax-operator
Defaulted container "gpu-labelgen" out of: gpu-labelgen, init-log-dir (init)
{"level":"info","ts":"2026-07-17T06:45:03Z","logger":"label-generator.entry","msg":"Starting manager"}
{"level":"info","ts":"2026-07-17T06:45:03Z","logger":"controller-runtime.metrics","msg":"Starting metrics server"}
{"level":"info","ts":"2026-07-17T06:45:03Z","msg":"Starting EventSource","controller":"node","controllerGroup":"","controllerKind":"Node","source":"kind source: *v1.Node"}
{"level":"info","ts":"2026-07-17T06:45:03Z","logger":"controller-runtime.metrics","msg":"Serving metrics server","bindAddress":":8080","secure":false}
{"level":"info","ts":"2026-07-17T06:45:03Z","msg":"Starting Controller","controller":"node","controllerGroup":"","controllerKind":"Node"}
{"level":"info","ts":"2026-07-17T06:45:03Z","msg":"Starting workers","controller":"node","controllerGroup":"","controllerKind":"Node","worker count":1}
{"level":"info","ts":"2026-07-17T06:45:03Z","logger":"label-generator.UpdateEvent","msg":"label changed","diff":{"metax.com/gpu.installed":"nil => false"}}
^C

尊敬的开发者您好，支持部署，请参考developer.metax-tech.com/api/client/document/preview/1273/k8s/01_scheduler.html#volcano

就是参考的这个文档，出现上述问题。

最先是报vender参数解析的问题：
helm upgrade --install metax-operator ./packages/metax-operator-0.15.2.tgz -f ./values/metax-operator.yaml -n metax-operator
Release "metax-operator" does not exist. Installing it now.
I0717 14:51:05.514079 1811828 warnings.go:107] "Warning: spec.template.spec.affinity.nodeAffinity.preferredDuringSchedulingIgnoredDuringExecution[0].preference.matchExpressions[0].key: node-role.kubernetes.io/master is use \"node-role.kubernetes.io/control-plane\" instead"
Error: server-side apply failed for object /cluster-operator gpu.metax-tech.com/v1alpha1, Kind=ClusterOperator: ClusterOperator.gpu.metax-tech.com "cluster-operator" is invalid: [spec.vendor.charDev: Invalid value: "null": spec.vendor.charDev in body must be of type string: "null", spec.vendor.domain: Invalid value: "null": spec.vendor.domain in body must be of type string: "null", spec.vendor.driver: Invalid value: "null": spec.vendor.driver in body must be of type string: "null", spec.vendor.vendorID: Invalid value: "null": spec.vendor.vendorID in body must be of type string: "null", spec.vendor.virtDriver: Invalid value: "null": spec.vendor.virtDriver in body must be of type string: "null"]

指定如下的vendor后，helm 部署成功，但是gpu-label无法识别 C500 卡：

vendor:
vendorID: "9999" # PCI Vendor ID (lspci shows Device [9999:4001])
driver: "metax" # Kernel driver module (Kernel driver in use: METAX)
virtDriver: "vfio-pci" # VFIO driver for GPU passthrough to KVM/QEMU VMs
charDev: "mxcd" # Character device (/dev/mxcd - MetaX Control Device)
domain: "metax.com" # K8s extended resource prefix (e.g., metax.com/gpu)

尊敬的开发者您好，Maca payload configuration
maca:
payload:
registry: "hub.sudoinfotech.com/metax"
images:
- "maca:3.8.0.11-ubuntu22.04-amd64"

vendor:
vendorID: "9999"
driver: "metax"
virtDriver: "vfio-pci"
charDev: "mxcd"
domain: "metax.com"，请将vendorID9999改为0x9999尝试

尊敬的开发者您好，麻烦提供一下./values/metax-operator.yaml配置文件

我把vendorID改成0x9999就正常了，感谢