On 12/31/18 9:44 PM, Theodore Y. Ts'o wrote: > But it seems to me that making cloud-init to use some kind of plug-in > architecture would be really helpful, even if it's not going to be a > solution for the immediate future.
Cloud-init already has a kind of plugin architecture. You can activate / deactivate many stuff in it, like updating your package list at boot time and many other stuff. > That, and cloud-init *badly* needs > to get away from the "run-and-done" model. That, we agree. >> There are also other considerations such as Windows, feature release >> co-ordination etc, runtime features (one thing Google's instance agent >> supports is injecting new SSH keys to instances, which cloud-init as a "run >> and done" doesn't support.) > > "Run-and-done" also doesn't support disk snapshots and disk resizing > properly. You didn't react to my suggestion about having some kind of signal sent by Qemu to the kernel. Isn't this a better idea than having a userland agent doing it? Cheers, Thomas Goirand (zigo)
