Summary of Changes

Hello @haitwang-cloud, I'm Gemini Code Assist¹! I'm currently reviewing this pull request and will post my feedback shortly. In the meantime, here's a summary to help you and other reviewers quickly get up to speed!

This pull request resolves issues related to device allocation in Kubernetes pods that utilize both init containers and regular containers. The changes ensure that device requests are correctly processed and mapped to the appropriate containers by fixing indexing logic and preventing the premature skipping of container device entries. This improves the robustness of device allocation for complex pod configurations.

Highlights

• Fixes device allocation for multi-container pods with init containers: Addressed the 'device request not found' error that occurred when allocating devices to pods containing both init containers and regular containers.
• Corrected container index mapping: Implemented logic to properly map device requests to init containers and regular containers, accounting for the index offset between them.
• Preserved empty ContainerDevices entries: Ensured that empty ContainerDevices entries are no longer skipped during decoding, which was previously breaking the index mapping for multi-container pods.
• Enhanced pending pod detection: Modified the logic to identify pending pods for device binding, now also considering pods in the 'DeviceBindSuccess' phase, which is crucial for multi-container scenarios where some containers might be allocated while others are still pending.

🧠 New Feature in Public Preview: You can now enable Memory to help Gemini Code Assist learn from your team's feedback. This makes future code reviews more consistent and personalized to your project's style. Click here to enable Memory in your admin console.

U can reproduce the bug by the following yaml

          initContainers:
            - name: cuda-sample-vector-add
              image: nvcr.io/nvidia/k8s/cuda-sample:vectoradd-cuda11.7.1-ubuntu20.04
              command:
                - /bin/bash
                - '-c'
                - '--'
              args:
                - /cuda-samples/vectorAdd && exit 0 || exit 1
              resources:
                limits:
                  hami.com/gpu: '1'
              imagePullPolicy: IfNotPresent
          containers:
          - name: cuda-sample-vector-add
            image: nvcr.io/nvidia/k8s/cuda-sample:vectoradd-cuda11.7.1-ubuntu20.04
            command:
              - /bin/bash
              - '-c'
              - '--'
            args:
                - /cuda-samples/vectorAdd && exit 0 || exit 1
            resources:
              limits:
                hami.com/gpu: '1'
            imagePullPolicy: IfNotPresent

@DSFans2014 Done,PTAL again

Codecov Report

✅ All modified and coverable lines are covered by tests.

Flags with carried forward coverage won't be shown. Click here to find out more.

... and 1 file with indirect coverage changes

@Shouren @DSFans2014 @archlitchi PTAL, and let me know if u have any concerns

@Shouren PTAL again, I have address all the UT error.

@Shouren PTAL again, I have address all the UT error.

@haitwang-cloud I am attending KubeCon EU this week. I will review it next week.

[APPROVALNOTIFIER] This PR is APPROVED

This pull-request has been approved by: haitwang-cloud, Shouren

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

The pull request process is described here