On Thu, Apr 22, 2010 at 9:40 PM, Michael Torrie <torr...@gmail.com> wrote:
> On 04/22/2010 09:19 PM, Robert LeBlanc wrote: > > I lurk on the btrfs list for fun ( I think I need to get out more ). > > Although the on-disk format should not change, there are quiet a large > > number of changes in the code since I've been on the list. I really don't > > like how they are planning to implement quotas and that you can't do > > recursive snapshots. They are planning to implement quotas like ZFS does, > > basically it's a new file system with it's own quota for the entire file > > system. This is great until you wish to host hundreds of home spaces, and > > you want 10 snapshots so they can restore their own files. > > My ZFS pools typically have hundreds, and in one case, thousands, of > file systems and each of them have quotas and I typically keep 7 > snapshots of each file system. As long as you can do recursive > snapshotting, this method of quotas works great. In fact, given the > nature of ZFS and BtrFS I don't know how else one would implement > quotas. On ZFS, snapshots count against your quota, as they should. > This occasionally causes problems when a user runs out of space and then > deletes files, thinking he's freeing up space. > > > <snip> > > Basically, I'm lurking on the btrfs list so that I can speak up on > snapshots > > and quotas when the topics come up. I like the on the fly compression and > it > > feels pretty snappy. On my Debian Squeeze system, it won't automatically > > mount, but I'm sure it's something I've done wrong. Linus is using btrfs > as > > his root file system on his machines (Yikes, I'd hate to be Chris. That's > a > > lot of pressure). > > How would you like snapshots and quotas done? > > As of right now, btrfs will not recurse down to do snapshots, so if you have your root and all your file systems under that root and you want to snapshot each one, you can't just snap the root, you have to snap each individual file system. As far as quotas, for me right now, the way you describe ZFS could work for us. I'm thinking about flexibility for the future. I do like that on a file system, I can pull a quota report and see a user's usage across multiple group shares for instance. I could also give each group share a quota and set per user quotas for all groups. I can't think of how to do that with ZFS's quota/file system methodology. If you can tell me how that could be accomplished, I'd be happy to know so that I can try it on btrfs. So far, I'm stuck on the EXT2/3/4, XFS, ReiserFS type of quotas to be able to accomplish that. Thanks for the clarification on ZFS. I tried it in Fuse on Linux and hung the system I think because I created some kind of mount loop. Since the code was so old and abandoned from what I could tell (lead developer went to Lustre when he graduated), I couldn't tell if it was me or the code. Robert LeBlanc Life Sciences & Undergraduate Education Computer Support Brigham Young University
-------------------- BYU Unix Users Group http://uug.byu.edu/ The opinions expressed in this message are the responsibility of their author. They are not endorsed by BYU, the BYU CS Department or BYU-UUG. ___________________________________________________________________ List Info (unsubscribe here): http://uug.byu.edu/mailman/listinfo/uug-list
