I understand it'll be a lot of work, but wouldn't this be an excellent moment to introduce grub2, which has btrfs support already, iirc? Wzzrd 05:37, 29 May 2011 (UTC)
I think we should be good to keep desktop integration in mind here too. It is no good to have filesystem features exposed in anaconda, and then have no UI in the installed system to deal with it. --mclasen 18:21, 1 June 2011 (UTC)
Now would also probably be a good time to start setting up QA test cases. Jdulaney 008, 2 June 2011 (UTC)
Not losing functions, which worked with ext4
- Is there plan to utilize dm-crypt or do you want to reimplement whole encryption in btrfs? On the wiki page are a little misleading information about btrfs/dm-crypt https://btrfs.wiki.kernel.org/index.php/Gotchas
- Missing swap file support in combination with missing encryption removes very important Fedora feature (fully encrypted laptops). mmaslano 15:25, 7 June 2011 (GMT)
- initramfs - Grub2 cannot boot from LUKS/encrypted disk. Despite experimental patches exists, these are not upstream and are very problematic (it need reimplementation of every new algorithm, LUKS version, hash etc inside grub2). So, the reality is that you need initramfs even with grub2 if you want full disk encryption. mmaslano 16:13, 7 June 2011 (GMT)
- Working with quota must be still possible. Missing quota means that some installations (like ftp servers etc) has no ability to limit space for users. mmaslano 15:25, 7 June 2011 (GMT)
- Ask anaconda guys, how would they handle it. Preferably, add easier change of fs in installer. mmaslano 19:21, 3 June 2011 (GMT)
- basic support is possible, but no cool btrfs features. It's not feature as a future strategy: https://fedoraproject.org/wiki/Anaconda/Features/BtrfsSupport mmaslano 14:44, 8 June 2011 (GMT)
- According to some info, there is no plan to implement raid5/6 directly in btrfs but reuse common library with MD. What are the plans? ETA?
- Can btrfs do all features like MD? (Online reshaping, monitoring, regular RAID resync, etc. Pointer to documentation?)
- btw Page: "The change should be largely invisible to users." -> The RAID maintenance (using mdadm) is completely different, the same applies for LVM.
- fsck must work. It should be tested for a longer period to see if it's really working. Probably good test suite would be nice. When fsck is planned to be commited to upstream btrfs-progs? http://git.kernel.org/?p=linux/kernel/git/mason/btrfs-progs-unstable.git;a=summary Another thing, why there are no recent commits and why Fedora has patches, which are not in upstream? mmaslano 15:25, 7 June 2011 (GMT)
- Is there instruction how to handle situations like disk failure?
Stability & performance of filesystem
All btrfs bugs should be reviewed and the important will be added into tracking bug for blockers of btrfs.
- https://bugzilla.redhat.com/show_bug.cgi?id=710534 mmaslano 19:21, 3 June 2011 (GMT)
- https://bugzilla.redhat.com/show_bug.cgi?id=689127 virtualization killer mmaslano 15:37, 7 June 2011 (GMT)
I've sketched a few high-level use cases from a desktop perspective [] - are those feasible at all with btrfs in F16, F17, or down the road ? --mclasen 15:40, 8 June 2011 (UTC)
Could you create performance tests? ext4 vs btrfs on different type of systems? Inspired by http://thread.gmane.org/gmane.comp.file-systems.btrfs/11401 mmaslano 12:21, 21 June 2011 (GMT)