Comment 0 for bug 2048876

Currently, the default chrony.conf configures a set of pools. Confirmed this on a focal and jammy instance on GCP. If one wishes to use only a specific server/server pool or not use a server at all they will need to modify /etc/chrony/chrony.conf. This will possibly lead to a prompt during an Ubuntu release upgrade and during an unattended chrony security upgrade.
We are trying to move all configuration changes to their respective *.d directories. See: https://bugs.launchpad.net/livecd-rootfs/+bug/1968873
We test for modified chrony config file by invoking `sudo md5sum --quiet --check /var/lib/ucf/hashfile`.

Listing the cases that I know where we are not able to move chrony configuration changes to a *.d config
1. Azure: Azure needs all default pool entries in chrony.conf disabled. This is currently done by commenting out the pool entries in /etc/chrony/chrony.conf. There doesn't seem to be an alternative way to reset the pool set used by chrony through a configuration in *.d directory.
2. Google: GCP images need to set a single server source entry. This is done indirectly through the ntp cloud-init module configuration. The ntp module replaces the default /etc/chrony/chrony.conf with another file that has required server entry and no pool entries. I believe this cannot be done through an override in *.d directory without touching /etc/chrony/chrony.conf.

This request perhaps can be extended to ensure that "negating" a configuration in the default /etc/chrony/chrony.conf should be possible through a configuration in /etc/chrony/*.d directory.