generate EC2 AMI from compressed images

[thomas-roos]

question on the upload script ... couldn't we e.g. use a compressed file
upload to s3 and inject an uncompression step ?

I'm playing with images with extra space (aka "0"s ) which are e.g. 14GB in
size but would compress to like 800MB with xz.
Doing the upload compressed would save so much time.

I looked at the import commands, but they do only take uncompressed vhd vhdx
and raw.

W/O changing that tooling, would it be possible to upload compressed and
execute an unxz on an s3 object ?

[thomas-roos]

maybe https://www.cloudtechsimplified.com/unzip-files-s3-bucket-aws-lambda/
might this helps: https://github.com/aws4embeddedlinux/meta-aws-ewaol?tab=readme-ov-file#limitations growing the partition when started?
Other than this I did not find a easy solution. There is something that extracts a zip in S3 (Lambda function) but this needs to be created in the template (https://github.com/aws4embeddedlinux/meta-aws/blob/master/scripts/ec2-ami/vmimport-cfn.yml)
Easiest would be a supported compressed VDI format, but I did not see this supported at first sight as well.

[thomas-roos]

May auto resize the partition on first boot ist an easy first step!?

[thomas-roos]

If the partition is a data partition it can be solved as here, to grow the partition on first boot: https://github.com/aws4embeddedlinux/meta-aws-demos/blob/d10a7010c9d64e6bf3b1465a7255e72c6da65b37/layers/bsp/meta-rauc-aws-ec2-x86-64/recipes-core/rauc/files/rauc-grow-data-partition.service#L8