Skip to content

v1.20

Compare
Choose a tag to compare
@sharkinsspatial sharkinsspatial released this 24 Aug 19:07
· 151 commits to master since this release
30d7944

Temporary fix for upstream issue with the most recent ECS Optimized AMI release. From the AWS Batch team

Dear Customer,

Upon some further researching, I have some more findings regarding the issue you have encountered.

We are aware of an issue when launching Amazon Linux 2 (AL2) AMIs and passing a cloud-init boot hook via EC2 userdata. If the boot hook attempts to upgrade the chrony package, some instances will fail to boot. The root cause is a faulty update to the chrony package that was pushed to the yum repositories on August 19, 2022. The workaround for this issue is to NOT update the chrony package from the boot hook, by excluding the package as below:

yum update -y --exclude='chrony*'

I could see from the script that has EFS calls included has "cloud-init-per once yum_update yum update -y" commands which could possibly affected by the issue.

Alternatively, you may also consider moving the "yum update" from 'cloud-boothook' section to the 'shellscript' section.

Our ETA for a new ECS optimized AMI to resolve this issue is August 30, 2022.

Notes there was a time where ssh was not possible as "cloud-bookhook" and "cloud-config" are the sections executed by cloud-init before any other script, hence any failure here can lead to subsequent runlevels to be skipped, hence daemons (like ssh) fail to start. [1]