On Thursday 02 April 2015 19:04:56 Anton Aylward wrote:
What you described happened to me a long time ago, many revisions age. You don't' say which version of the kernel you use, BtrFS + Tools. Its in constant revision and along the way there have been ... Diversions ... That have been less than satisfactory.
I suppose one might say that one should never turn off the machine :-) Certainly every problem I've ever had has involved turning the machine back on :-) What? Disk crash from head retraction? BTDT a couple of times :-(
When this happened to me it was on a /home partition. I've never had it happen on a ROOT partition. Not even at the time I was also running the BtrFS-ificated /home. I did speculate that the BtrFS driver was not re-entrant, but was told that is not so.
So, what benefit is there to BtrFS other than its new and shiny?
My interest in it was to avoid the fixed number of inodes situation with ext4. I'd been bitten by that to many times in the last century and this. I had been using ReiserFS and was happy with that but people keep telling me its unsupported and won't be around.
ZFS is getting to look viable and does what BtrFS claims, volume management, spindle management over and above the simple partition model. Its reputed to be one of the really good things that came out of SUN. I can believe that. Is it production ready for Linux yet? I can't speak to that.
Linda recommends XFS. It seems very fiddly to me.
In the long run I wouldn't give up on BtrFS. KDE4 was a true Open Source project in that it was a 'ship early and get feedback' project. The mistake was that the early version were treated as production not as development. It was released as part of the main system, replacing KDE3 completely, before it was ready. Yes, it forced users to try it out, push its limits and report bugs. Perhaps if it hadn't been forced on us its development would have been slower. Perhaps too there would have been less antagonism.
BtrFS seems to have gone the same way, forced on us, and without a lot of consideration about such matters as "whole disk" vs partition, an issue which is key to the design assumptions; without adequate explanation of 'snapper'; without adequate explanation of subvolumes. All of these are new concepts. The questions that Don Fisher has posed are indicative, I think, of how the Joe Sixpack Linux users are dazed by these new concepts. I'm not sure how to handle them so have them all disabled.
Linda recommends XFS. Personally I find it too fiddly.
And there's JFS out of Veritas and IBM. I don't know anything about that.
Interesting summary, makes a lot of sense. And repetition of Linda and her XFS makes it little psychodelic :) For me btrfs works quite good so far, except for USB 3.0 SSD drive, where it constantly loses free space cache, which in turn hangs the systemd for all the eternity on umount and I have to use hard poweroff. At first I was using clear_cache mount option as default, but couple of days ago gave up and reformatted everything to ext4. At least it does not have the same issues. On the other hand, I've set btrfs on encrypted /home despite yast not allowing that. Was not so hard actually, and works much much better then XFS. And XFS was a complete disaster. I believe they have some kind of delayed flush, or something like that. Unpacking of 500 MB tar.xz tarball with the sources and git repository inside finished suspiciously fast, but in couple of seconds the system freezes and is absolutely not responsible for next 3-4 minutes, even mouse does not move. Yes, btrfs maybe not so fast, but at least the system remains usable the whole time. That is the one thing I miss from good old SUSE 8-10 days, when the choice of the file system was not a question at all, ReiserFS ruled them all. -- Regards, Stas -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org