Ubuntu 22.04 EC2实例重启后无法通过EC2 Instance Connect连接

问题详情 #

• 使用Ubuntu 22.04 AMI的EC2实例，执行sudo reboot后无法正常运行，无法通过EC2 Instance Connect连接
• 实例状态检查结果：1/2 checks passed
• 已创建实例快照并挂载到新实例，仍无法连接

系统启动日志 #

[[0;32m  OK  [0m] Finished [0;1;39mEC2 Instance Connect Host Key Harvesting[0m.
[[0;32m  OK  [0m] Finished [0;1;39mGRUB failed boot detection[0m.
[[0;32m  OK  [0m] Finished [0;1;39mRemove Stale Onliâ¦ext4 Metadata Check Snapshots[0m.
[[0;32m  OK  [0m] Started [0;1;39mA high performanceâ¦er and a reverse proxy server[0m.
[[0;32m  OK  [0m] Started [0;1;39mUser Login Management[0m.
[[0;32m  OK  [0m] Started [0;1;39mUnattended Upgrades Shutdown[0m.
[[0;32m  OK  [0m] Started [0;1;39mDispatcher daemon for systemd-networkd[0m.
[[0;1;31mFAILED[0m] Failed to start [0;1;39mEC2 instance hibernation setup agent[0m.
See 'systemctl status hibinit-agent.service' for details.
[[0;32m  OK  [0m] Started [0;1;39mSnap Daemon[0m.
         Starting [0;1;39mWait until snapd is fully seeded[0m...
         Starting [0;1;39mTime & Date Service[0m...
[[0;32m  OK  [0m] Started [0;1;39mTime & Date Service[0m.
[[0;32m  OK  [0m] Finished [0;1;39mWait until snapd is fully seeded[0m.
[[0;32m  OK  [0m] Finished [0;1;39mService for snap application lxd.activate[0m.
[[0;32m  OK  [0m] Stopped [0;1;39mService for snap aâ¦on-ssm-agent.amazon-ssm-agent[0m.
[[0;32m  OK  [0m] Started [0;1;39mService for snap aâ¦on-ssm-agent.amazon-ssm-agent[0m.
[[0m[0;31m*     [0m] A start job is running for Wait for⦠to be Configured (32s / no limit)
M
[K[[0;1;31m*[0m[0;31m*    [0m] A start job is running for Wait for⦠to be Configured (33s / no limit)
M
[K[[0;31m*[0;1;31m*[0m[0;31m*   [0m] A start job is running for Wait for⦠to be Configured (33s / no limit)
M
[K[ [0;31m*[0;1;31m*[0m[0;31m*  [0m] A start job is running for Wait for⦠to be Configured (34s / no limit)
M
[K[  [0;31m*[0;1;31m*[0m[0;31m* [0m] A start job is running for Wait for⦠to be Configured (34s / no limit)
M
[K[   [0;31m*[0;1;31m*[0m[0;31m*[0m] A start job is running for Wait for⦠to be Configured (35s / no limit)
M
[K[    [0;31m*[0;1;31m*[0m] A start job is running for Wait for⦠to be Configured (35s / no limit)
M
[K[     [0;31m*[0m] A start job is running for Wait for⦠to be Configured (36s / no limit)
M
[K[    [0;31m*[0;1;31m*[0m] A start job is running for Wait for⦠to be Configured (36s / no limit)
M
[K[   [0;31m*[0;1;31m*[0m[0;31m*[0m] A start job is running for Wait for⦠to be Configured (37s / no limit)
M
[K[  [0;31m*[0;1;31m*[0m[0;31m* [0m] A start job is running for Wait for⦠to be Configured (37s / no limit)
M
[K[ [0;31m*[0;1;31m*[0m[0;31m*  [0m] A start job is running for Wait for⦠to be Configured (38s / no limit)

解决方案 #

1. 定位核心问题 #

日志中重复出现的Wait for... to be Configured是关键，这是systemd-networkd-wait-online.service在等待所有网络接口配置完成时超时卡住，导致系统无法完成启动流程。hibinit-agent.service启动失败属于次要问题，不影响实例核心连接。

2. 修改故障卷的系统配置（基于已挂载快照到新实例的操作） #

在挂载了故障卷的正常Ubuntu实例上执行以下步骤：

• 挂载故障卷的根分区（假设挂载点为/mnt/faulty-root）：

  sudo mount /dev/xvdf1 /mnt/faulty-root
  

• 进入挂载的系统环境：

  sudo chroot /mnt/faulty-root
  

• 修改systemd-networkd-wait-online.service配置，使其只需等待任意网络接口就绪：

  systemctl edit systemd-networkd-wait-online.service
  

  在打开的编辑器中添加以下内容：

  [Service]
  ExecStart=
  ExecStart=/lib/systemd/systemd-networkd-wait-online --any
  

  保存退出后，重新加载配置：

  systemctl daemon-reload
  

• （可选）禁用hibinit-agent.service以消除启动错误：

  systemctl disable hibinit-agent.service
  

• 退出chroot环境并卸载故障卷：

  exit
  sudo umount /mnt/faulty-root
  

3. 恢复实例 #

将修改后的卷挂载回原EC2实例，或基于该卷创建新实例，启动后即可正常通过EC2 Instance Connect连接。

内容的提问来源于stack exchange，提问作者Kunal Koli