Hey there, I have taken some of the concepts and scripts that Colin used
to make the AWS AMI Builder (thank you!!), and used them to configure a
ZFS-on-Root image for GCP.
Here’s the script:
https://gist.github.com/patmaddox/1a0be973ecfce4f056c517abb7746286
I did it slightly differently from the AMI builder. Instead of working
in a mem disk, I just fetch base.txz and expand it to a dir, edit the
files, etc.
One other difference is that because base.txz doesn’t include a
kernel, I `cp -Rp /boot/kernel/ /mnt/boot/kernel/`.
It boots, I can SSH and do all the usual things, so far I am very happy.
My questions:
1. Is there anything I might be missing with the approach of extracting
base.txz, and copying over the current instance’s kernel?
2. If I change the GCP disk size and reboot, the partition table is
corrupt. I need to `geom part recover da0 && geom part resize -i 2 da0`
to resize it. How can I have that happen automatically during the boot
process, the way that the AMI does?
Pat
