I've received email from aws stating that one of my ec2 instance is scheduled for retirement due to underlying hardware degradation on 1st September 2017. When I checked that instance after receiving the email, It was already unreachable and 0/2 health check passed. I did couple of reboot from aws web console after which instance got up and running with status 2/2 health check passed Now my question is Will it still be unreachable after 1st September 2017? I am having basic support plan and hence can't raise this with aws directly. I highly appreciate community support here.

Note: It's a EBS backed ubuntu linux instance. If you find this question meaningful kindly upvote.



"Will it still be unreachable after 1st September 2017"

You need to STOP and then START then instance to move it to new hardware. If you did a REBOOT then it will stop working on 1st September. A simple stop / start cycle will mean it will stay available.

You can read up on instance lifecycles here. It doesn't explicitly say about the hardware, I learned that during certification study. It's probably in the documentation somewhere, but there's so much documentation for AWS it can be difficult to locate.

| improve this answer | |
  • There's not a separate thing called "restart." There is only "reboot" and "stop/start." "Restart" is sometimes used in the documentation for "stop/start." If an instance scheduled for retirement came back to life after being in the 0/2 health check passed state, then almost certainly either a stop/start was already performed (not a reboot -- a reboot can't happen in the 0/2 state -- you can't reboot a system with an inaccessible hypervisor, and that's what the first check is testing). That, or the hypervisor is flaky and recovered coincident with (not because of) the reboot attempt. – Michael - sqlbot Aug 29 '17 at 6:58
  • There is an explicit reference to the hardware. "When you stop a running instance... in most cases, the instance is migrated to a new underlying host computer when it's started." The "most cases" caveat presumably would apply in a constrained capacity environment where there isn't any other hardware available, or to placement groups in some cases, or dedicated hosts... but a VM should never start up on a physical machine already flagged as degraded, so the caveat almost certainly never applies, here. – Michael - sqlbot Aug 29 '17 at 7:14
  • Yep you're right Michael, I said "restart" when I meant "reboot". I think of them as generally the same, so I tend to use the words interchangeably. I did clearly say stop then start is what needs to happen here. – Tim Aug 29 '17 at 7:39

According to AWS Docs:

An instance is scheduled to be retired when AWS detects irreparable failure of the underlying hardware hosting the instance. Your instance root device is an Amazon EBS volume so the instance will be stopped, and you can start it again at any time.

You have two options here:

  1. Wait for the scheduled retirement date, when the instance is stopped you can start it again at any time or
  2. Create an EBS-backed AMI from your instance, and launch a replacement instance. See here
| improve this answer | |
  • This doesn't answer the question, doesn't cover all the options, and your point #2 is only correct if the instance is based on ephemeral disk. – Tim Aug 28 '17 at 19:39
  • You're wrong. Look here – Prem Sompura Aug 29 '17 at 5:51
  • "An instance is scheduled to be retired when AWS detects irreparable failure of the underlying hardware hosting the instance." This sentence is copy/pasted from the documentation and presented here as if they are your own words -- without proper attribution. Please review the help section "How to reference material written by others." – Michael - sqlbot Aug 29 '17 at 7:25
  • Point #2 is technically possible for EBS-backed instances, but is unnecessary unless for some reason your degraded instance is an EBS-backed spot instance, in which case you can actually make an AMI from it after it fails, and often end up with a usable image. (EBS-backed spot instances, like instance-store, cannot be stopped, they can only be terminated.) Overlooked in this answer: stop/start now to move away from the degraded hardware and automatically clear the scheduled retirement event for the instance. – Michael - sqlbot Aug 29 '17 at 7:30
  • Though I'm certainly not always right, I don't believe I'm wrong here. #1 misses the main thing you need to do, which is stop and start the instance. #2 is largely unnecessary as you can use the EBS volume without creating an AMI. – Tim Aug 29 '17 at 7:38

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.