[clug] EXT4 Reliability

Daniel Pittman daniel at rimspace.net
Sun Oct 4 17:26:16 MDT 2009

Ian McLeod <ianmcleod75 at gmail.com> writes:

> So EXT4 is more of a performance than stability enhancement then

...so far.

Let me emphasize that again: today, ext4 is higher performance, but less
stable, than ext3.  *TODAY*  This *IS* going to change, and ext4 will replace
ext3 as standard in due time.

As an example: the next stable kernel release will fix a (fairly rare) kernel
oops caused by a bug in the ext4 code.  It has been quite some time since ext3
hit many of those.

Given time — a few years of people testing it with easily replaced[1] data —
ext4 will be a very compelling replacement for ext3.  It just needs that extra
time to mature.

> I should have used EXT3 on my backup drive - oh well..  But if performance
> improves then it would be good on this laptop I just installed with
> encrypted EXT4 filesystems.

*nod*  Alternately, you could use XFS, which has many of the same advantages
and drawbacks[2], but has many years more testing behind it, so is a more
conservative choice to gain the same advantages.


I use XFS on an encrypted disk, on MD-RAID10,f2, which gives quite good
performance on this laptop.

[1]  ...and a sad number testing it with their only copy of critical data, but
     we should all try to discourage that, /before/ they lose that data.

[2]  ...because that performance /does/ have some costs, like the risk that
     unwritten extents and no more sync-every-five-seconds presents.

✣ Daniel Pittman            ✉ daniel at rimspace.net            ☎ +61 401 155 707
               ♽ made with 100 percent post-consumer electrons
   Looking for work?  Love Perl?  In Melbourne, Australia?  We are hiring.

More information about the linux mailing list