I want to check if there are any issues with the agent installation or configuration,

Question

I want to check if there are any issues with the agent installation or configuration,

眞仁田博之(Hiroyuki Manita) 20

問題の説明

Azure Arc で管理されているサーバーが自動オンボードされなくなっており、エージェントのインストールや設定に問題があるのか、または通信経路に障害があるのかを確認したいです。特に、Amazon Linux 2023 上で azcmagent コマンドやパッケージが存在しない状態です。

環境

Azure Arc enabled servers, Amazon Linux 2023, 日本リージョン

既に試したこと

Azure Arc エージェントの診断や、サーバーのネットワーク状態、証明書の検証、インスタンスプロファイルの確認などを行いました。ただし、azcmagent コマンドやパッケージが存在せず、インストール状況が不明です。

現在の状態

対象サーバーでのエージェントのインストールと設定の状況を詳細に調査したいです。特に、通信経路や必要なエージェントのインストール方法についてアドバイスを求めています。

Moderator translated message:

Problem Explanation

I want to check if there are any issues with the agent installation or configuration, or if there are issues with the communication route, since servers managed by Azure Arc are no longer automatically onboarded. In particular, the azcmagent command or package does not exist on Amazon Linux 2023.

environment

Azure Arc enabled servers, Amazon Linux 2023, Japan region

What I've already tried

We conducted diagnostics of the Azure Arc agent, verified server network status, certificates, and instance profiles. However, the azcmagent command and package do not exist, so the installation status is unknown.

Current Status

I want to investigate in detail the status of agent installation and configuration on the target server. In particular, they are seeking advice on communication routes and how to install the necessary agents.

0 comments

1 answer

Your answer

Answer 1

Hi 眞仁田博之(Hiroyuki Manita) & thx for join me at Q&A platform,

first thing I'd focus on is that azcmagent and its package are both missing from the Amazon Linux 2023 instance. That points more toward an agent deployment failure than an existing Arc agent losing connectivity.

Amazon Linux 2023 is supported by Azure Arc, so the OS itself shouldn't be the blocker.

You can confirm the installation state with rpm -q azcmagent and ls -l /opt/azcmagent/bin/azcmagent. If neither finds anything, there's no point troubleshooting azcmagent check yet because the agent isn't installed.

The next step depends on how automatic onboarding was configured. If you're using the Azure Arc multicloud connector for AWS EC2 onboarding, the deployment relies on AWS Systems Manager (SSM). I'd check whether the affected EC2 instance appears as an online managed node in AWS Systems Manager and whether the SSM agent is running:

sud0 systemctl status amazon-ssm-agent

Check amazon-ssm-agent.log and the Systems Manager Run Command execution history for installation failures. Those should help distinguish an SSM execution problem from a failure to download or install the Arc package. On the Azure side, open your multicloud connector and check Test Permissions, the Arc onboarding solution, configured AWS regions, and EC2 tag filters. Also verify that the instance has the required ArcForServerSSMRole IAM role assigned. An instance can be successfully discovered in inventory without the Arc agent being installed.

If SSM successfully executes the installation but the package download fails, then investigate outbound HTTPS connectivity, DNS, proxy configuration, and access to Microsoft's package repository. That's where the network troubleshooting becomes relevant.

You can install the Connected Machine agent manually using Microsoft's Linux installation script from https://aka.ms/azcmagent, but I'd avoid doing that immediately if your goal is to restore automatic onboarding. Manual installation could make the instance appear fixed while leaving the actual automation problem unresolved.

Once the agent is installed, sud0 /opt/azcmagent/bin/azcmagent check --location japaneast and azcmagent show can help diagnose connectivity and registration. Use japanwest instead if that's the actual target Azure region.

Are these AWS EC2 instances being onboarded through the Azure Arc multicloud connector, or through your own automation such as SSM scripts, Terraform, or Ansible? Are the affected instances visible in Azure as discovered AWS resources but missing from Azure Arc Machines, or have previously connected machines changed to Disconnected?

That distinction would help identify whether the failure is occurring during discovery, agent deployment, or connection to Azure.

rgds,

Alex

p.s. if my answer is help - pls mark it as an answer