On 03 Sep 1999 14:33:31 -0700, Philippe Troin <phil@fifi.org> said:
> This box is hit by quite heavy NFS traffic. There seem to be no file
> corruption (so far).
Please test. It is really easy to miss file corruption: the metadata
errors here
> This is the error patterns (slightly reformatted):
> Sep 1 22:22:54: attempt to access beyond end of device
> Sep 1 22:22:54: 08:08: rw=0, want=1145390407, limit=16208895
> Sep 1 22:22:54: dev 08:08 blksize=1024 blocknr=1145390406
> sector=-2004186484 size=1024 count=1
are spotted for you by the kernel, so I'd like to be _sure_ that the
silent corruptions which would happen if there were bit-flips in data
too are definitely absent. md5sum or cmp can stress file content
checking quite easily.
Is knfsd a common factor in people seeing these problems, btw? We had a
couple of cases in the past where knfsd was causing metadata funnies, so
if nobody is seeing these without knfsd I'll have a look to see if there
are any cases we are still getting wrong in there.
--Stephen
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/