Closed dirkmueller closed 2 weeks ago
You are right this code path was not ported to the kiwi-ng version. Interesting that it pops now.
is there a different way to achieve this nowadays?
I suggest to add a patch to kiwi and add the free blocks scanner back. It would be good to add this as a capability of the filesystem class and support it for the filesystems for which such a tooling exists. imho extX and XFS only.
It would be good to get some numbers though because I don't know if there are really that many blocks to clear in the process of an image build. If your time permits could you check on one of the integration test images e.g Virtualization:Appliances:Images:Testing_x86:tumbleweed/ if you postprocess the filesystem with zerofree and how much it saves us on compressing the result ?
We shouldn't need it in the current kiwi, since we don't create the disk image until the very end and sync the files over to it.
That was also my understanding, but maybe @dirkmueller has a workflow in mind where it would pay off. That's why I'm waiting for his feedback too. Thanks
I will do experiments and see if there's a gain. I haven't determined yet whether zerofree would bring a gain
Do we still need this, @dirkmueller? At least from my own tests and from an incidental analysis on the Fedora side, I think we don't need this.
I'm closing this issue due to lack of information and lack of apparent need of this.
fine by me, thanks for the cleanup!
Problem description
zerofree is a tool to zero out XFS and ext2/3/4 filesystem blocks that are no longer allocated but maybe have been written to in the past. this improves compression later somewhat.
legacy kiwi had some code to deal with this:
https://github.com/OSInside/kiwi-legacy/blob/1d4f7b139c481f063e4e008059439bdcf03ff482/modules/KIWIBoot.pm#L2734
I don't find this in the kiwi git tree and I find no plugin. how / where did that move? is there a different way to achieve this nowadays?
Expected behaviour
Steps to reproduce the behaviour
OS and Software information