Skip to content

Health Monitoring Agent helm chart missing ap-southeast-3 region mapping causes ImagePullBackOff #359

@KeitaW

Description

@KeitaW

Describe the bug

The Health Monitoring Agent (HMA) helm chart is missing the ap-southeast-3 (Jakarta) region in the ECR account mapping, causing HMA pods to fail with ImagePullBackOff errors when deploying HyperPod EKS clusters in that region.

Root Cause

The region-to-ECR account mapping in helm_chart/HyperPodHelmChart/charts/health-monitoring-agent/templates/_helpers.tpl does not include ap-southeast-3.

When a region is not found in the mapping, the template defaults to using account 767398015722 but constructs the ECR URI using the deployment region, resulting in an invalid endpoint.

This ECR repository does not exist in ap-southeast-3, causing a 403 Forbidden error.

Error Messages

Warning  FailedCreatePodSandBox  Failed to find plugin "aws-cni" in path /opt/cni/bin
Warning  ErrImagePull            403 Forbidden
Warning  ImagePullBackOff        Back-off pulling image "767398015722.dkr.ecr.ap-southeast-3.amazonaws.com/sagemaker-hyperpod-health-monitoring-agent:1.0.1038.0_1.0.305.0"

Expected Behavior

HMA should successfully pull the container image and start normally in all AWS regions where HyperPod is supported.

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions