Re: hunt for 2.6.37 dm-crypt+ext4 corruption? (was: Re: dm-crypt barrier support is effective)

[Chris Mason <chris.mason@oracle.com>]

Excerpts from Jon Nelson's message of 2010-12-07 15:25:47 -0500:
> On Tue, Dec 7, 2010 at 2:02 PM, Chris Mason <chris.mason@oracle.com> =
wrote:
> > Excerpts from Jon Nelson's message of 2010-12-07 14:34:40 -0500:
> >> On Tue, Dec 7, 2010 at 12:52 PM, Chris Mason <chris.mason@oracle.c=
om> wrote:
> >> >> postgresql errors. Typically, header corruption but from the li=
mited
> >> >> visibility I've had into this via strace, what I see is zeroed =
pages
> >> >> where there shouldn't be.
> >> >
> >> > This sounds a lot like a bug higher up than dm-crypt. =C2=A0Zero=
s tend to
> >> > come from some piece of code explicitly filling a page with zero=
s, and
> >> > that often happens in the corner cases for O_DIRECT and a few ot=
her
> >> > places in the filesystem.
> >> >
> >> > Have you tried triggering this with a regular block device?
> >>
> >> I just tried the whole set of tests, but with /dev/sdb directly (a=
s
> >> ext4) without any crypt-y bits.
> >> It takes more iterations but out of 6 tests I had one failure: sam=
e
> >> type of thing, 'invalid page header in block ....'.
> >>
> >> I can't guarantee that it is a full-page of zeroes, just what I sa=
w
> >> from the (limited) stracing I did.
> >
> > Fantastic. Now for our usual suspects:
> >
> > 1) Is postgres using O_DIRECT? =C2=A0If yes, please turn it off
>=20
> According to strace, O_DIRECT didn't show up once during the test.
>=20
> > 2) Is postgres allocating sparse files? =C2=A0If yes, please have i=
t fully
> > allocate the file instead.
>=20
> That's a tough one. I don't think postgresql does that, but I'm not a=
n
> expert here.
>=20
> > 3) Is postgres using preallocation (fallocate)? =C2=A0If yes, pleas=
e have it
> > fully allocate the file instead
>=20
> As far as strace is concerned, postgreql is not using fallocate in
> this version.

Well, the only other usual suspect would be mmap.  Does the strace show
that you're using read/write for file IO or is it doing a lot of mmaps
on the files?

-chris
--
To unsubscribe from this list: send the line "unsubscribe linux-ext4" i=
n
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html