docs: fix bottlerocket node template for GPU volume size #2400
Conversation
the GPU bottlerocket AMI requires at least 3GB. otherwise you run into:
> 2022-08-30T13:27:52.897Z ERROR controller.provisioning Provisioning failed, launching node, creating cloud provider instance, with fleet error(s), InvalidBlockDeviceMapping: Volume of size 3GB is smaller than snapshot 'snap-0c5c999fe56c48b6f', expect size >= 4GB {"commit": "639756a"}
The normal bottlerocket AMI size is still 2GB, just GPU is a bit bigger.
There was also a typo with the volume size value.
✅ Deploy Preview for karpenter-docs-prod canceled.
|
| @@ -23,7 +23,7 @@ spec: | |||
| - deviceName: /dev/xvda | |||
| ebs: | |||
| volumeType: gp3 | |||
| volumeSize: 2Gi | |||
| volumeSize: 2G # if using GPU, replace with 4G since the GPU bottlerocket AMI requires at least 4G | |||
There was a problem hiding this comment.
Gi is the correct format, right?
the typo is in the one with 20GiBs
There was a problem hiding this comment.
They both work, G is a power of 10 version and Gi is the power of 2 version. https://kubernetes.io/docs/concepts/configuration/manage-resources-containers/#meaning-of-memory
There was a problem hiding this comment.
right... but helps if we align with what AWS EC2 and AMIs provide.
If EBSs with 20 gigabytes are 20Gi, that is what I feel we should use, since if we were to use 20G, the volume would be smaller than the AMI and fail.
Similar for RAM. If an ec2 has 16 gigabytes of RAM, using 16G is better than Gi ?
and yeah, I'm aware that HDD and SSD use base 2 and base 10 each.... OEMs ....
Description
the GPU bottlerocket AMI requires at least 3GB. otherwise you run into:
The normal bottlerocket AMI size is still 2GB, just GPU is a bit bigger.
There was also a typo with the volume size value.
How was this change tested?
Does this change impact docs?
By submitting this pull request, I confirm that my contribution is made under the terms of the Apache 2.0 license.