Comment 0 for bug 1840686

CPC team has recently converted Xenial images to use GPT instead of MBR. However, after booting an instance that has a disk size of 2049 GB or higher, we hang on the next subsequent boot (Logs indicate it hanging on "Booting Hard Disk 0".

This works on Bionic, but what makes it strange is that they have the same kernel revision - 4.15.0-1-37.

patrick_viafore@patviafore-test-3072-xenial:~$ lsb_release -rd
Description: Ubuntu 16.04.6 LTS
Release: 16.04
patrick_viafore@patviafore-test-3072-xenial:~$ sudo dpkg -l | grep linux-gcp
ii linux-gcp 4.15.0.1037.51 amd64 Complete Google Cloud Platform (GCP) Linux kernel and headers
ii linux-gcp-headers-4.15.0-1037 4.15.0-1037.39~16.04.1 amd64 Header files related to Linux kernel version 4.15.0

To reproduce:

1) Create an image with a disk size of 3072 using a serial that has GPT
gcloud compute instances create test-3072-xenial --image daily-ubuntu-1604-xenial-v20190731 --image-project ubuntu-os-cloud-devel --boot-disk-size 3072

Reboot the instance

2) It will hang on reboot and you cannot connect

3) Please note that later serials have the GPT change reverted.

You can replace xenial with bionic in the above commands to get a bionic instance instead.

To test this out in a more slower fashion:

1) Create an image with a disk size of 2048 using a serial that has GPT
gcloud compute instances create test-2048-xenial --image daily-ubuntu-1604-xenial-v20190731 --image-project ubuntu-os-cloud-devel --boot-disk-size 2048

2) Resize the disk to 3072

3) Issue growpart /dev/sda 1

4) Issue resize2fs /dev/sda1

5) Issue rsize2fs /dev/sda1 instead

On the second resize2fs, it tries to resize again, but on a working instance, it says there's nothing to resize.

I've tried starting from a Xenial instance and doing a do-release-upgrade to get to bionic and then doing the growpart/resize2fs, but the issue still shows up.