Interesting Blog Post about 8.3 names

Jeremy Allison jra at samba.org
Mon Nov 26 12:08:09 MST 2012


On Sun, Nov 25, 2012 at 02:51:33AM +0900, TAKAHASHI Motonobu wrote:
> From: Jeremy Allison <jra at samba.org>
> Date: Tue, 20 Nov 2012 08:56:36 -0800
> 
> > On Tue, Nov 20, 2012 at 09:40:53AM +0100, Christian Ambach wrote:
> >> In case you are not following Jose's blog, there is an interesting post
> >> about the performance impacts of 8.3 names on NTFS and how to disable
> >> them to get more speed:
> >>
> >> http://blogs.technet.com/b/josebda/archive/2012/11/13/windows-server-2012-fi\
> le-server-tip-disable-8-3-naming-and-strip-those-short-names-too.aspx
> >>
> >> What would be the matching Samba options to disable 8.3 names
> >> completely? Any idea how much that speed up Samba directory enumerations?
> >
> > We don't have one (currently). We don't store 8.3 names on disk,
> > but always generate them from a hashing function, so it wouldn't
> > change anything w.r.t. disk access or lookups. Removing them
> > would save some CPU cycles though - might be worth it.
> >
> > Jeremy.
> 
> To set "mangled names = no", Samba does not generate 8+3 names.

Well we still generate *something* to put in that field when
"mangled names = no", we just do it badly :-).

        /* name mangling can be disabled for speed, in which case
           we just truncate the string */
        if (!lp_manglednames(p)) {
                strlcpy(out, in, 13);
                return True;
        }

Jeremy.


More information about the samba-technical mailing list