If you are using AWS and EC2 instances, a reboot of most those instances is on the horizon. Amazon’s AWS informed of this reboot that is scheduled between 02:00 GMT on September 26th and 23:59 GMT on September 30th.
Read more about this reboot on Gigaom and Rightscale. Technical Forums on AWS and other sites are already buzzing with lot of traffic, discussing the potential impact and how to ensure that the services are not impacted.
Given the urgency and magnitude of the instances that are impacted, it looks like the patch is going potentially going to address a security vulnerability. The actual details of the patch and the issues that are fixed by it will be known around October 01st.
Summarizing various discussions on related forums, here is a quick summary of what to watch out for during this AWS / EC2 instance reboot
- The reboot is not limited to any single availability zone. It spawns across all the availability zones
- Good news is that the EC2 instances on all availability zones are not rebooted at the same time. So if your instances spawn across multiple availability zones, you are on a relatively safer side.
- The reboot does not impact instances of the type T1, T2, M2, R3, and HS1. However, if the patch fixes issues on these instance types too, then you might be on your own. We will know more around October 1st.
Here are a few quick checks for those who are getting impacted.
- Check your mailbox for a notice from AWS and it is likely to give more details about the reboots, impact and schedules
- Ensure that the key services on your instances are configured for auto restart when the system boots up. It looks silly, but I have seen code that takes good care of newly spawned instances but doesn’t address reboots that well.
- Ensure that your network paths (non-Elastic IPs, Route 53 entries, S3 buckets) survive reboot of the instances.
- For those whose instances are NOT rebooted by AWS, watch out for the issues fixed by AWS during this reboot and evaluate their impact on your instances. Take corrective measures as soon as possible.
For those who can afford to be heroic enough – why wait till AWS reboots your instances? Reboot these on your own in each availability zone and test the resilience.