[SGVLUG] SUSE abandoning ReiserFS

[Mike Fedyk]

Dustin Laurence wrote:
> This surprised me, maybe it will surprise someone else too:
>
> http://linux.wordpress.com/2006/09/27/suse-102-ditching-reiserfs-as-it-default-fs/
I'm not surprised.

A typical requirement for a core VFS patch in the kernel is to convert 
all callers of that call.  Reiser3 has been left out of that requirement 
several times because a lot of the kernel hackers don't like the reiser3 
code.  This is the "maintenance" reason Jeff used in his post, and left 
out probably because he's either being polite or political.

The revelation that reiser3 uses the BKL everywhere only makes the 
argument stronger.

I have used reiser3 before in the past and liked it until I ran into a 
lot of problems with the reiserfsck tools.  The last time I used 
reiserfsck, it was anything but robust with a 50% chance of needing to 
do a meta-data reconstruction.

Also add to this that it didn't do well in the fragmentation area at 
all.  You keep the metadata in a tree, but allocate blocks from the 
beginning of the disk with no consideration of locality between inode 
and data blocks (except for tail packing of course).  This was fixed by 
Chris Mason IIRC.  And Hans fought tooth and nail to keep it out because 
"reiser3 is in maintenance and everyone should use reiser4".

I don't think SUSE had a choice.

I like XFS and I don't know nearly as much about it as ext2/3 and 
reiser3, but there are a lot of known ordering issues with it similar to 
what ext2 does to the point that several developers have said concerning 
bad blocks and and power outages: "don't use xfs unless it is on RAID 
and powered by a UPS".  It is an "enterprise" filesystem and it scales 
for that use, but it doesn't handle the low-end reliability issues 
nearly as well as ext2/3.  And the xfs development community is much 
smaller than ext2/3 so suse would be moving from a filesystem in low 
standing to a filesystem in mid standing.

That said, I have no qualms with using xfs, and will be learning more 
about it more in the near future.

Mike