yast / yast-storage-ng

Rewrite of https://github.com/yast/yast-storage
http://yast.github.io/
GNU General Public License v2.0
13 stars 19 forks source link

Handling of YAST_REUSE_LVM linuxrc option #1365

Closed mchf closed 5 months ago

mchf commented 5 months ago

Problem

https://jira.suse.com/browse/PED-6407

The Storage Proposal in SLE 15 allows for creating a layout based on LVM. For historical reasons, it tries to reuse the existing LVM volume group for new installations. This sometimes produces a non-optimal, non-logical or even confusing disk setup.

Solution

linuxrc option to disable lvm reuse

Testing

coveralls commented 5 months ago

Coverage Status

coverage: 97.762% (+0.002%) from 97.76% when pulling eb6aa7d2d784f40b973dabc2371be6bdecf5af1a on yast_lvm_reuse into 4253f2b8408eee70fd24be5e9fb18e9248e91807 on SLE-15-SP6.

ancorgs commented 5 months ago

Review was sent when I was still working on the comment (I clicked the wrong button, I guess).

I was saying that I like the idea of apply_user_enforced although I think the current implementation does not actually sets the value for the setting.

That being said, I have doubts about adding the setting to the control file. I'm not saying is a bad idea, I literally mean I have doubts whether is a good idea. On the bright side, we will be prepared for configuring the default value per-product in the future without code changes. On the other hand, adding things to the control file definition is a live-lasting contract signed in blood. :wink:

mchf commented 5 months ago

Review was sent when I was still working on the comment (I clicked the wrong button, I guess).

I was saying that I like the idea of apply_user_enforced although I think the current implementation does not actually sets the value for the setting.

That being said, I have doubts about adding the setting to the control file. I'm not saying is a bad idea, I literally mean I have doubts whether is a good idea. On the bright side, we will be prepared for configuring the default value per-product in the future without code changes. On the other hand, adding things to the control file definition is a live-lasting contract signed in blood. 😉

I had an idea that we can easily switch default behavior later on if we want to. Question is if it pays of to do it in a creepy way or we should hope in big bang. Adding a feature into control file even changing default proposal, both is radical. Which way brings us more benefits? I'm not sure ... but with control file it at least won't require our assistance in that decision (hope).

yast-bot commented 5 months ago

:heavy_check_mark: Internal Jenkins job #5 successfully finished :heavy_check_mark: Created IBS submit request #321652