跳至內容

t4g server hangs to dead without any reason

0

I've started instance type c8g.2xlarge, started the docker build process using just 1 thread. After 5 minutes, the server is completely dead; it is not responsive, and it is not possible to log in using SSH or using the Amazon console. Out of memory is not possible: just 1 build thread and 16 GB of RAM is completely enough.

Instance state is green, and status check passed. But all monitoring graphs stopped because it is not possible to grab any information from a dead server.

Instance ID: i-0f99d60722953ff45, region: eu-central-1. Can you please check this machine using a physically connected serial console?

Also, please refund the money; we should not pay anything for the broken hardware.

已提問 1 年前檢視次數 94 次

1 個回答
0

Based on the information provided, it appears your c8g.2xlarge instance has become unresponsive despite showing a "running" state and passing status checks. This situation is similar to other cases where instances appear healthy from AWS monitoring perspective but are actually unreachable.

While I can't directly check your specific instance using a physically connected serial console (as this would require AWS Support intervention), I can suggest some troubleshooting steps:

  1. The issue might be related to the underlying system rather than broken hardware. Sometimes instances can become unresponsive due to disk errors that cause the filesystem to be mounted in read-only mode, or due to kernel-level issues.

  2. Since you mentioned you were running a Docker build process, it's possible that despite having sufficient RAM, there could be other resource constraints or system-level issues triggered by the build process.

  3. You could try stopping and starting the instance (not just rebooting) which would migrate it to different underlying hardware.

  4. If possible, create a snapshot of the instance's root volume before terminating it, which would allow you to investigate the issue further or recover data if needed.

Regarding your refund request, you would need to contact AWS Support directly to discuss billing adjustments for service issues. They can review your specific case and determine if a refund is appropriate based on their service level agreements.
Sources
statuscheckfailed_instance still unreachable after reboot and shutdown restart | AWS re:Post

已回答 1 年前

專家

已審閱 1 年前

您尚未登入。 登入 去張貼答案。

一個好的回答可以清楚地回答問題並提供建設性的意見回饋,同時有助於提問者的專業成長。