On Fri, 16 May 2008 15:53:31 -0500
Eric Sandeen [off-list ref] wrote:

Andrew Morton wrote:

quoted

On Fri, 16 May 2008 14:02:46 -0500
Eric Sandeen [off-list ref] wrote:

quoted

A collection of patches to make ext3 & 4 use barriers by
default, and to call blkdev_issue_flush on fsync if they
are enabled.

Last time this came up lots of workloads slowed down by 30% so I
dropped the patches in horror.

I actually did a bit of research and found the old thread, honestly.  I
thought this might not be a shoo-in.  :)  Seems worth hashing out, though.

quoted

I just don't think we can quietly go and slow everyone's machines down
by this much.  The overhead of journalling is already pretty horrid.

But if journali[zi]ng guarantees are thrown out the window by volatile
caches on disk, why bother with the half-solution?  Slower while you
run, worthless when you lose power?  Sounds like the worst of both
worlds.  (well, ok, experience shows that it's not worthless in practice...)

quoted

If we were seeing a significant number of "hey, my disk got wrecked"
reports which attributable to this then yes, perhaps we should change
the default.  But I've never seen _any_, although I've seen claims that
others have seen reports.

Hm, how would we know, really?  What does it look like?  It'd totally
depend on what got lost...  When do you find out?  Again depends what
you're doing, I think.  I'll admit that I don't have any good evidence
of my own.  I'll go off and do some plug-pull-testing and a benchmark or
two.

But, drive caches are only getting bigger, I assume this can't help.  I
have a hard time seeing how speed at the cost of correctness is the
right call...

Yeah, it's all so handwavy.  The only thing which isn't handwavy is
that performance hit.

quoted

There are no happy solutions here, and I'm inclined to let this dog
remain asleep and continue to leave it up to distributors to decide
what their default should be.

Do we know which distros are enabling barriers by default?

SuSE does (via patch for ext3).  Red Hat & Fedora don't, and install by
default on lvm which won't pass barriers anyway.  So maybe it's
hypocritical to send this patch from redhat.com  :)

And as another "who uses barriers" datapoint, reiserfs & xfs both have
them on by default.

I suppose alternately I could send another patch to remove "remember
that ext3/4 by default offers higher data integrity guarantees than
most." from Documentation/filesystems/ext4.txt  ;)

We could add a big scary printk at mount time and provide a document?