Mounting ebs volume via bootstrap-containers

[bartoszdabr]

Hi!
I have setup with karpenter that creates bottlerocket nodes with the following blockDevicesMappings setting

blockDeviceMappings:
# xda is the OS partition for bottlerocket and should only be 4Gi; read only
- deviceName: /dev/xvda
ebs:
volumeType: gp3
volumeSize: 4Gi
deleteOnTermination: true
encrypted: true
# Bottlerocket data volume; read-write. This stores images etc
- deviceName: /dev/xvdb
ebs:
volumeType: gp3
volumeSize: 256Gi
deleteOnTermination: true
encrypted: true
# Cache node drive; read-write.
- deviceName: /dev/xvdc
ebs:
volumeType: gp3
volumeSize: 200Gi
deleteOnTermination: true
encrypted: true

I want the /dev/xvdc to be mounted on node so my pods can use hostPath volume for on-node cache mechanism. I've read that bootstrap-containers is the way to go for me so I've tried to set it up by passing userData on the EC2NodeClass resource

[settings.bootstrap-containers.cache]
source = "public.ecr.aws/amazonlinux/amazonlinux:2023"
essential = false
mode = "once"
user-data = "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"

The b64 user-data is exactly:
#!/usr/bin/env bash
set -ex

yum install -y e2fsprogs parted

# The name of the disk we want to manage
DISK=/.bottlerocket/rootfs/dev/nvme2n1
# Sentry file to check if the disk was already partitioned and formatted
PARTITIONS_CREATED=/.bottlerocket/bootstrap-containers/current/created
# Mounts from this mount point will propagate accross mount namespaces
BASE_MOUNT_POINT=/.bottlerocket/rootfs/mnt

# If the disk hasn't been partitioned, create the partitions and format them
if [ ! -f $PARTITIONS_CREATED ]; then
  parted -s $DISK mklabel gpt 1>/dev/null
  parted -s $DISK mkpart primary ext4 0% 100% 1>/dev/null
  mkfs.ext4 -F ${DISK}p1
  # Create sentry file once the disk is partitioned and formatted
  touch $PARTITIONS_CREATED
fi

# We make sure the target mount points exist
mkdir -p $BASE_MOUNT_POINT/cache

# Always mount the partitions
mount ${DISK}p1 $BASE_MOUNT_POINT/cache

it's from this discussion #2168 with a small adjustment for my case.
Do you see anything that I'm doing wrong or how can I at least debug that? I tried getting into node via aws ssm -> apiclient exec admin bash -> sheltie and running journalctl but all I get to find there is that the /usr/bin/boostrap-containers command has been run and no longs from inside the container that would suggest anything wrong. The one thing that I've noticed that although I set ebs as /dev/xvdc it's always nvme2n1 in admin shell so I've changed that device name in script

[bartoszdabr]

Nevermind, I've just realized that the script must be a part of dockerfile instead of passing it through user-data, once I've pasted it to image and set it as entrypoint everything started working. I guess you were my rubber duck