Jeremy writes:
> I've seen a small number of people currently using ext3. A lot of
> distributions are making the option available, if not in their binary
> release, but such things as kernel source rpm's are making ext3 an option.
>
> The thing is, in all the accounts I know of where people are using ext3, I
> haven't anything negative about ext3. I have yet to try it out, but I'm
> in the process. Just curious on the development track and maybe some good
> reasons not to use ext3 currently.
As of ext3-0.0.2f there are no problems with journal recovery that I'm
aware of. This doesn't mean that ext3 is as bulletproof as ext2, because
the ext3 code doesn't handle a lot of uncommon error conditions (such as
disk read/write errors, data corruption, OOM, etc) in a graceful manner.
In fact, the current ext3/jfs code has lots of assertions in it that will
purposely cause an oops if we get in some wierd state, rather than doing
something reasonable such as abort with an error.
The other current drawback of ext3 is the I/O performance is slower
than that of ext2 because of the lack of metadata-only journalling.
Not necessarily half as fast as ext2 (because of cache and I/O ordering),
but it can't be used where you are already pushing a box to the I/O limit.
However, since most systems never get disk I/O bound for any length of
time, ext3 is still a good choice.
Cheers, Andreas
--
Andreas Dilger \ "If a man ate a pound of pasta and a pound of antipasto,
\ would they cancel out, leaving him still hungry?"
http://www-mddsp.enel.ucalgary.ca/People/adilger/ -- Dogbert
