Thanks for the hints, however, I can't try it anymore. The latest freeze has left my data partition with a corrupt metadata tree and -- rebuild-tree gave the filesystem it's rest.
After some research on Reiser4 and the surprising twist in Hans Reiser's personal fate (he was arrested with murder charges last October), I'm having my doubts as far as further development/ bugfixing is concerned and therefore decided to switch back to ext3 before playing back the backup.
http://en.wikipedia.org/wiki/Hans_reiser
Cheers, -sven
Sven Schwyn wrote:
Thanks for the hints, however, I can't try it anymore. The latest freeze has left my data partition with a corrupt metadata tree and --rebuild-tree gave the filesystem it's rest.
After some research on Reiser4 and the surprising twist in Hans Reiser's personal fate (he was arrested with murder charges last October), I'm having my doubts as far as further development/bugfixing is concerned and therefore decided to switch back to ext3 before playing back the backup.
I'm using SGI's XFS for any kind of data filesystem and only use ext3 for / . So far, no problems at all.
Just my 2 cents,
En/na Patrick Cernko ha escrit:
I'm using SGI's XFS for any kind of data filesystem and only use ext3 for / . So far, no problems at all.
I also use XFS exclusively, but I have the "random zeroed files" when the system crashes/power goes off unexpectedly.
Bye
On Tuesday 24 April 2007 19:46, Luca Olivetti wrote:
En/na Patrick Cernko ha escrit:
I'm using SGI's XFS for any kind of data filesystem and only use ext3 for / . So far, no problems at all.
I also use XFS exclusively, but I have the "random zeroed files" when the system crashes/power goes off unexpectedly.
I always thought exactly this should be avoided by a journalling FS?
A pity that Tux2 never made it (because the author thought of patent issues) ... or is there similar technology available now, so many years after?
Guido Fiala wrote:
On Tuesday 24 April 2007 19:46, Luca Olivetti wrote:
En/na Patrick Cernko ha escrit:
I'm using SGI's XFS for any kind of data filesystem and only use ext3 for / . So far, no problems at all.
I also use XFS exclusively, but I have the "random zeroed files" when the system crashes/power goes off unexpectedly.
Oh sorry, I missed that. But something like that must be expected on every system on crashed at least if you do not mount the FS "in sync" (like sugested for a totally safe qmail mailserver :-) ).
I always thought exactly this should be avoided by a journalling FS?
A pity that Tux2 never made it (because the author thought of patent issues) ... or is there similar technology available now, so many years after?
The journal normally only protects the filesystem metadata so that the OS can still "recognize" the the block-dev as a filesystem. Some filesystems also allow "data journaling" but at a cost of writing the data twice to the disc(s).
So long,
On Tuesday 24 April 2007 21:19, Patrick Cernko wrote:
The journal normally only protects the filesystem metadata so that the OS can still "recognize" the the block-dev as a filesystem. Some filesystems also allow "data journaling" but at a cost of writing the data twice to the disc(s).
Well - tux2 was different - it simply wrote the inode-structure separately for the new data/files and did an atomic switchover from one inode-structure to the other. As far as i gathered it, it was a quite unobstrusive patch to ext2...
Sven Schwyn wrote:
After some research on Reiser4 and the surprising twist in Hans Reiser's personal fate (he was arrested with murder charges last October), I'm having my doubts as far as further development/bugfixing is concerned and therefore decided to switch back to ext3 before playing back the backup.
Reiser's personal issues are strange, however I don't think that this is relevant for the file system. Reiser3 is stable and working fine in my eyes, even though it has a bad reputation because of some early bugs.
Reiser4 sounds like an excellent performer on the parallel-recording task, and its very sad that Reiser4 is waiting for kernel integration for several years now. Hopefully the remaining developers continue on it.
Ext3 by the way is one of the slowest on parallel-recording, one of the reasons why ext4 is on heavy development. If you really want to avoid Reiser3, I'd suggest XFS or other high performing journaling FS'.
Cheers,
Udo
-
Guido Fiala -
Luca Olivetti -
Patrick Cernko -
Sven Schwyn -
Udo Richter