rsharpe at richardsharpe.com
Tue Dec 9 03:35:36 GMT 2008
On Mon, 8 Dec 2008, Jeremy Allison wrote:
> Here's a really interesting paper from Intel
> that they recently brought to my attention.
This _is_ an interesting paper, but I disagree with the following
"If the server OS uses a filesystem like NTFS, these pre-allocate writes
are used to allocate contiguous blocks on the disk. NTFS assumes that all
files will be filled with data, so the small write to a high offset forces
the filesystem to allocate a set of contiguous blocks up to that offset.
When the valid data is written to NTFS, the blocks are already allocated,
simplifying and speeding the data writes."
I have seen this behavior before, and it is most likely because the client
has an exclusive OpLock on the file and the writes are going into the
client cache, however, to present any sort of reasonable semantics with
respect to ENOSPACE they have to attempt to extend the file to the end of
It just so happens that this is good for other reasons on Windows as well.
Interestingly, I have been seeing other problems with large non-sparse
files in a VTL implementation (many GB sequential files) and Ext3, so this
might be what I need to switch to XFS.
Richard Sharpe, rsharpe[at]richardsharpe.com, rsharpe[at]samba.org,
More information about the samba-technical