use device serial number to construct id

[peachest]

fix #16

The device id in the device information reported and registered to the node by the current dcu-vgpu-device-plugin is constructed by adding the device index to the DCU- prefix.

If there is only one DCU node in the cluster, there is no effect on HAMi scheduling. However, if there are multiple DCU nodes in the cluster, the scheduler will not be able to schedule properly, for example, use use-uuid to limit the scheduling range.

The dcu-vgpu device plugin originally integrates dcgm to get device information. At the same time, dcgm provides a ShowSerialNumber function to get the serial number of the device based on the specified device index. Usage:

deviceSerialInfos, err := dcgm.ShowUId([]int{0, 1, 2, 3, 4, 5, 6, 7})

根据查询出来的序列号构造出设备 id，格式为 DCU-<serial number>。影响范围包括如下两点

1. 设备插件注册到节点上的 Annotation，以及
2. 为容器分配设备时需要根据序列号反向找回对应的设备序号，以挂载正确的设备文件

测试结果如下：

根据日志时间戳，调用 dcgm.ShowSerialNumber 的时间不超过 1s，时间开销少。

节点 Annotation 如下：

在一个 8 卡集群中申请 0~3 号这 4 张卡，调度成功的 Pod Annotation 如下：

容器内部实际挂载设备：

结论：可以在不修改调度器的情况下正常调度，并且可以正常使用根据 uuid 指定调度范围的特性。

[hami-robot]

Welcome @peachest! It looks like this is your first PR to Project-HAMi/dcu-vgpu-device-plugin 🎉

[Nimbus318]

Thanks for the contribution! Could you also add an example for specifying UUID in the following documentation paths?

• [EN] docs/userguide/Hygon-device/examples
• [ZH] i18n/zh/docusaurus-plugin-content-docs/current/userguide/Hygon-device/examples

You can refer to the NVIDIA examples below for formatting and content:

• [EN] docs/userguide/NVIDIA-device/specify-device-uuid-to-use.md
• [ZH] i18n/zh/docusaurus-plugin-content-docs/current/userguide/NVIDIA-device/specify-device-uuid-to-use.md

No need to update anything in versioned_docs; just adding it under the current docs will ensure it's visible in the latest (Next) version. Thanks!

[peachest]

Thanks for the contribution! Could you also add an example for specifying UUID in the following documentation paths?感谢您的贡献！您能否在以下文档路径中添加一个指定 UUID 的示例？

• [EN] docs/userguide/Hygon-device/examples  [英文] docs/userguide/Hygon-device/examples
• [ZH] i18n/zh/docusaurus-plugin-content-docs/current/userguide/Hygon-device/examples  [中文] i18n/zh/docusaurus-plugin-content-docs/current/userguide/Hygon-device/examples

You can refer to the NVIDIA examples below for formatting and content:您可以参考下面的 NVIDIA 示例来了解格式和内容：

• [EN] docs/userguide/NVIDIA-device/specify-device-uuid-to-use.md  [英文] docs/userguide/NVIDIA-device/specify-device-uuid-to-use.md
• [ZH] i18n/zh/docusaurus-plugin-content-docs/current/userguide/NVIDIA-device/specify-device-uuid-to-use.md  [中文] i18n/zh/docusaurus-plugin-content-docs/current/userguide/NVIDIA-device/specify-device-uuid-to-use.md

No need to update anything in versioned_docs; just adding it under the current docs will ensure it's visible in the latest (Next) version. Thanks!无需在 versioned_docs 中更新任何内容;只需将其添加到当前文档下即可确保它在最新（Next）版本中可见。谢谢！

The corresponding PR Project-HAMi/website#87 has been submitted, please check it out

[peachest]

According to the metrics returned by the official dcu-exporter provided by hygon, the device id is obviously also using the serial number.
This further illustrates the correctness of using serial numbers to construct dcu device ids in HAMi.

[archlitchi]

/lgtm

[hami-robot]

[APPROVALNOTIFIER] This PR is NOT APPROVED

This pull-request has been approved by: archlitchi, peachest

The full list of commands accepted by this bot can be found here.

Details

Needs approval from an approver in each of these files:

• OWNERS

Approvers can indicate their approval by writing /approve in a comment
Approvers can cancel approval by writing /approve cancel in a comment