On Sat, Apr 22, Joe Salmeri wrote:
Hi Richard,
I am a TW user and don't think I'll be using ALP but I installed MicroOS in a VM to get a feel for how the transactional-update process works.
Since the root fs is read only, I see that overlays are used for /etc with directories created under /var/lib/overlay which corresponding to each of the rootfs snapshots.
On my systems I am often creating my own snaphots for various reasons but since /etc is no longer part of the rootfs snapshot, it kind of breaks the snapshot/rollback process for snapshots that were not created as part of the transactional-update process.
If you create an snapshot of your own, it's 1:1 identical to the already existing read-only snapshot. Means it includes /etc and the configuration for the overlays.
I guess one could create the directory under /var/lib/overlay for snapshots not created by transaction-update but that seems like a hack.
Are there plans to address that issue or is that a method I am not aware of with transactional-updates to also make it easy for other admin created snapshots ?
If you can come up with a real good use case, why somebody need to create identical snapshots themself, we can look at it. Currently I don't see such an use case, except making the current code even more complex for duplicate data. If you want to keep a snapshot, there is no need to create a new one, just tell snapper to not delete the existing one when doing the cleanup. Thorsten
It seems that the same situation will exist with ALP ?
Joe
On 4/20/23 15:01, Richard Brown wrote:
On 2023-04-20 17:34, Henne Vogelsang wrote:
Hey,
On 20.04.23 10:27, Richard Brown wrote:
The question really though is, given the above, what does the openSUSE community want to build? (yes..I know, I'm repeating the main question from my original post..but we really do need to have some discussions on that else the answer will be 'nothing')
I'm not sure I agree with how you put this. We are *already* building a thing (Leap) that would be affected by ALP. I would argue we are building Leap by consensus that it's something we want to build.
So the more glaring question is: Can we build Leap from ALP or what do we do instead?
I put this like this for 2 reasons
1. Given ALP is a very different distribution from SLE, I expect any derivative of ALP that tries to be like Leap will be a lot more work. This appears to be confirmed by Simon's experiences working towards this for hackweek, hence you see his suggestion being something paired down from Leap, at least with the manpower he currently sees being interested in doing such.
2. ALP is a huge change to how we build stuff, which makes it an opportunity to do something very different, new, different from Leap, different from Tumbleweed, different from MicroOS.
Given 1 and 2 I wanted to put this out in a clear "the door is WIDE OPEN" kind of way, to see what responses we get.
Asking your version of the question puts us on a path I only want to see us walk if people really step up and want to build it.
So far we have Alberto's "build nothing, but work on adapting Tumbleweed to the ALP way" suggestion, and Simons "build something like Leap but focused on the lightweight xfce/enlightenment desktops"
That's a good start..I'm looking forward to others
-- Regards,
Joe
-- Thorsten Kukuk, Distinguished Engineer, Senior Architect, Future Technologies SUSE Software Solutions Germany GmbH, Frankenstraße 146, 90461 Nuernberg, Germany Managing Director: Ivo Totev, Andrew Myers, Andrew McDonald, Martje Boudien Moerman (HRB 36809, AG Nürnberg)