Hi,
I've got a Fedora 28 image running on GCP which is basically just the raw disk image found in Fedora-Cloud-Base-28-1.1.x86_64.raw.xz. For the record, here's how I instantiated the VM:
xz --decompress Fedora-Cloud-Base-28-1.1.x86_64.raw.xz mv Fedora-Cloud-Base-28-1.1.x86_64.raw disk.raw tar cfz Fedora-Cloud-Base-28-1.1.x86_64.tar.gz disk.raw --sparse gsutil cp Fedora-Cloud-Base-28-1.1.x86_64.tar.gz gs://fedora-cloud-base/ gcloud compute images create --source-uri gs://fedora-cloud-base/Fedora-Cloud-Base-28-1.1.x86_64.tar.gz fedora-cloud-base-28 gcloud compute instances create fedora28 --machine-type f1-micro --image fedora-cloud-base-28 --zone us-east1-b
So far so good. Everything works as expected and I can log into the VM with 'gcloud compute ssh fedora@fedora28'.
However, as soon as I run 'dnf update', or to be more specific, as soon as I update the 'kernel-core' package and reboot the VM, it takes hours (yes, literally) for it to come up again. The log output I can grab over the serial port is not particularly helpful (https://paste.fedoraproject.org/paste/reRXgIX-z0bz0W~dw33b8A).
I'm quite sure I'm just missing the obvious here. How does one usually update the kernel inside a GCE VM?
Cheers!