Skip to content

Latest commit

 

History

History

image

README.md

Build custom AMI

  1. Make sure you have read/write access to a S3 bucket where you can store transient data and scripts. Replace occurrence of s3://hpc-scripts1234 with the actual S3 bucket reference in install-image.sh, image-config.yaml and install-r.sh.
  2. Check versions in install-image.sh
  3. Finally, run
./build-image.sh <IMAGENAME> [<BUCKETNAME>]

where <IMAGENAME> is the desired name of the new AMI and <BUCKETNAME> the name of the S3 bucket, e.g. hpc-scripts1234. <BUCKETNAME> is an optional argument. If missing, the script will look in .bucket.default to read the default bucket name.

Useful for debugging

Creating your own S3 bucket

aws s3api create-bucket --bucket <BUCKETNAME> --region <REGION> --create-bucket-configuration LocationConstraint=<REGION>

where <BUCKETNAME> is the desired name of the s3 bucket and <REGION> the region intended for use.

Cleanup

If you want to get rid of all AVAILABLE images, run

for i in `pcluster list-images --image-status AVAILABLE | grep imageId | awk '{print $2}' | sed 's#"##g' | sed 's#,##'`; do pcluster delete-image -i $i ; done

Get information about image

pcluster describe-image -i <IMAGENAME>
pcluster  list-image-log-streams -i <IMAGENAME>
pcluster get-image-log-events  -i <IMAGENAME> --log-stream-name <AWSPCVERSION>/1

where <AWSPCVERSION> is the version of AWS parallelcluster used (e.g. 3.11.1)