NVMe volume mounts do not persist upon reboot

[kislyuk]

This happens because systemd interferes with mount -a and unmounts the mountpoint specifies in /etc/fstab:

systemd[1]: dev-xvdz.device: Job dev-xvdz.device/start timed out.
systemd[1]: Timed out waiting for device /dev/xvdz.

The device name /dev/xvdz is a symlink that we create in cloudinit code to the actual device node. This apparently breaks systemd logic and it starts to unmount the device immediately even when we mount it manually after this.

Address this by disabling this systemd mount "helper" behavior.

[kislyuk]

See systemd/systemd#1741

[kislyuk]

See https://docs.aws.amazon.com/AWSEC2/latest/UserGuide/block-device-mapping-concepts.html