AWS: Shrink an EBS configured in GPT mode

I don't really know why, but my image on my EC2 server was not configured in MBR but in GPT. All the guides I found for shrinking a root EBS was not made for GPT.

First of all, you cannot shrink an EBS. You can expand it but shrinking is not allowed. To achieve that goal, you have to copy the content on another EBS volume.

View current partitioning

Stop your EC2 instance and dettach your root volume (let's call it V1).
Create a new volume with the new target size (V2).
Create a new EC2 instance. Attach it V1 and V2.

SSH into your instance and type sudo lsblk

admin@ip-10-99-4-58:~$ sudo lsblk
NAME    MAJ:MIN RM    SIZE RO TYPE MOUNTPOINT
xvda    202:0    0      8G  0 disk
└─xvda1 202:1    0      8G  0 part /
xvdf    202:80   0     70G  0 disk
xvdg    202:96   0    300G  0 disk
├─xvdg1 202:97   0 1007.5K  0 part
└─xvdg2 202:98   0    300G  0 part

In my example, V1 is xvdg and V2 is xvdf

Install gdisk if it's not already installed in your OS.

View V1 partitioning with sudo gdisk -l /dev/xvdg

admin@ip-10-99-4-58:~$ sudo gdisk -l /dev/xvdg
GPT fdisk (gdisk) version 1.0.1

Partition table scan:
  MBR: protective
  BSD: not present
  APM: not present
  GPT: present

Found valid GPT with protective MBR; using GPT.
Disk /dev/xvdg: 629145600 sectors, 300.0 GiB
Logical sector size: 512 bytes
Disk identifier (GUID): D6CFA2DD-A1E7-425F-87B8-395D3361807E
Partition table holds up to 128 entries
First usable sector is 34, last usable sector is 629145566
Partitions will be aligned on 8-sector boundaries
Total free space is 0 sectors (0 bytes)

Number  Start (sector)    End (sector)  Size       Code  Name
   1              34            2048   1007.5 KiB  EF02  primary
   2            2049       629145566   300.0 GiB   8300

Please write down the first and the last sector of the first partition.

Create partitioning on the new disk

Replicate this partitioning onto V2. Use gdisk another time:

admin@ip-10-99-4-58:~$ sudo gdisk /dev/xvdf
GPT fdisk (gdisk) version 1.0.1

Partition table scan:
  MBR: not present
  BSD: not present
  APM: not present
  GPT: not present

Creating new GPT entries.

Command (? for help):

Go in expert mode (x), then press l to set alignement, enter 1, then press m to go back.

Command (? for help): x

Expert command (? for help): l
Enter the sector alignment value (1-65536, default = 2048): 1

Expert command (? for help): m

Create a new BIOS boot partition with n. Set the first and the last sector with the value you wrote down on step 6. In my case 34 and 2048. Set the type to EF02.

Command (? for help): n
Partition number (1-128, default 1): 1
First sector (34-146800606, default = 34) or {+-}size{KMGTP}: 34
Last sector (34-146800606, default = 146800606) or {+-}size{KMGTP}: 2048
Current type is 'Linux filesystem'
Hex code or GUID (L to show codes, Enter = 8300): EF02
Changed type of partition to 'BIOS boot partition'

Create another partition with Linux type. Leave all default values.

Command (? for help): n
Partition number (2-128, default 2):
First sector (2049-146800606, default = 2049) or {+-}size{KMGTP}:
Last sector (2049-146800606, default = 146800606) or {+-}size{KMGTP}:
Current type is 'Linux filesystem'
Hex code or GUID (L to show codes, Enter = 8300):
Changed type of partition to 'Linux filesystem'

Write the new partition table with w

Command (? for help): w

Final checks complete. About to write GPT data. THIS WILL OVERWRITE EXISTING
PARTITIONS!!

Do you want to proceed? (Y/N): y
OK; writing new GUID partition table (GPT) to /dev/xvdf.
The operation has completed successfully.

Create the filesystem and replicate the old UUID

Create an ext4 partition on /dev/xvdf2

admin@ip-10-99-4-58:~$ sudo mkfs.ext4 /dev/xvdf2
mke2fs 1.43.4 (31-Jan-2017)
Creating filesystem with 18349819 4k blocks and 4587520 inodes
Filesystem UUID: 68690b84-c1e5-4532-a467-33047975dc14
Superblock backups stored on blocks:
        32768, 98304, 163840, 229376, 294912, 819200, 884736, 1605632, 2654208,
        4096000, 7962624, 11239424

Allocating group tables: done
Writing inode tables: done
Creating journal (131072 blocks): done
Writing superblocks and filesystem accounting information: done

When you created the new ext4 partition, a new UUID has been generated. We want the old UUID back, otherwise the system won't boot. Get the UUID of /dev/xvdg2 with sudo blkid /dev/xvdg2
```
admin@ip-10-99-4-58:~$ sudo blkid /dev/xvdg2
/dev/xvdg2: UUID="5cfc0e59-b149-4ddc-8741-74bd7e71f91c" TYPE="ext4" PTTYPE="dos" PARTUUID="05266ce5-177c-4bab-8001-302ad2119d7c"
```

Copy it to /dev/xvdf2

admin@ip-10-99-4-58:~$ sudo tune2fs /dev/xvdf2 -U 5cfc0e59-b149-4ddc-8741-74bd7e71f91c
tune2fs 1.43.4 (31-Jan-2017)
Setting UUID on a checksummed filesystem could take some time.
Proceed anyway (or wait 5 seconds) ? (y,N) y

Copy the data

Mount both disk onto the system

admin@ip-10-99-4-58:~$ mkdir -p /source /target
admin@ip-10-99-4-58:~$ sudo mount /dev/xvdg2 /source/
admin@ip-10-99-4-58:~$ sudo mount /dev/xvdf2 /target

rsync the data (careful, no trailing / after /target):
```
sudo rsync -HAXxSa /source/ /target
```

Install GRUB

Prepare the chroot environment so that we can install GRUB in the BIOS boot partition of the new disk

admin@ip-10-99-4-58: sudo -i
root@ip-10-99-4-58:~$ cd /target/
root@ip-10-99-4-58:/target$ mount -t proc proc proc/
root@ip-10-99-4-58:/target$ mount -t sysfs sys sys/
root@ip-10-99-4-58:/target$ mount -o bind /dev dev/
root@ip-10-99-4-58:/target$ chroot .

Install GRUB

root@ip-10-99-4-58:/# grub-install /dev/xvdf

Leave chroot with exit, then unmount all

root@ip-10-99-4-58:/$ exit
root@ip-10-99-4-58:/target$ umount proc
root@ip-10-99-4-58:/target$ umount dev
root@ip-10-99-4-58:/target$ umount sys
root@ip-10-99-4-58:/target$ cd
root@ip-10-99-4-58:~$ umount /target
root@ip-10-99-4-58:~$ umount /source

Detach both EBS volume from you EC2 instance and mount V2 as /dev/xvda on your old EC2 instance. Boot it. If all went well, you have your new shrinked system !