Merge brlock.tdb and locking.tdb
slow at samba.org
Thu Nov 2 10:37:24 UTC 2017
On Wed, Nov 01, 2017 at 10:45:41AM -0700, Jeremy Allison wrote:
> On Mon, Oct 30, 2017 at 09:06:44PM +0100, Volker Lendecke wrote:
> > On Mon, Oct 30, 2017 at 07:19:21PM +0100, Ralph Böhme via samba-technical wrote:
> > > yeah, I was afraid that would be the reason. On the downside, by
> > > using seperate dbs we have do the update-num-read-oplocks
> > > mumbo-jumbo, but I guess that's still less overhead then unpacking a
> > > locking.tdb record for the mandatory brls. Hmpf.
> > Just brainstorming: We have the locking.tdb record cache now. Does
> > that help here? Maybe the real performance penalty for the
> > non-contended workload isn't as bad anymore as it used to be. Of
> > course, if a lot of byte range locking activity is going on, things
> > go down, but does that matter in real world cases?
> Only benchmarking/tests can tell :-).
Ralph Boehme, Samba Team https://samba.org/
Samba Developer, SerNet GmbH https://sernet.de/en/samba/
More information about the samba-technical