Re: Happy success with LVM and RAID
On Mon, Dec 04, 2006 at 08:30:38AM +0100, Sven Luther wrote:
> On Sun, Dec 03, 2006 at 06:03:44PM -0500, Douglas Tutty wrote:
> > This juggling to keep all the balls (disks) in the air sounds familiar.
> > Non-partisan caution: before you settle on Reiserfs, read some recent
> > threads about Reiserfsck and Reiserfs vs JFS. I went to JFS when
> > Reiserfsck fscked up my drives. Never had a problem with JFS despite
> > many power failures.
> We made the choice for EXT3 ourselves, since it is the only filesystemm which
> can be grown online and shrinked offline, and reiserfs is not really a
> JFS/XFS can be grown online, but not shrinked, so they are problematic.
I went from ext3 to reiserfs because of corruption after power failures.
Ditto reiserfs to JFS. With JFS there have been no problems that I can
see. Once I get my new system running etch stable, I'll put in one of
the comprehesive file integrity monitors (current computer doesn't have
the juice to do that __and__ get anything accomplished) so I get a
notice if anything goes missing.
Now that I'm using LVM I'll want that ability to grow. But I doubt I'll
need to shrink.
I wish that there was a filesystem format that had built-in redundancy
for the data (not just the superblocks). Its OK to have raid1 to
protect against drive failure, but it would be nice to have something be
a little more concerned about data integrity. Reading all the docs, it
seems that all the fsck utils focus on getting the filesystem to a
mountable state; loss of data is a secondary concern.
There was a long thread here recently looking at this in some detail.