tdb performance

Volker Lendecke Volker.Lendecke at SerNet.DE
Sun Feb 4 23:32:28 GMT 2007

On Mon, Feb 05, 2007 at 10:21:37AM +1100, tridge at wrote:
> Are you sure this would be a real win? The way I usually test this
> sort of idea is this:
>  1) remove the lock/unlock pair, leaving the code unsafe

You mean the lock/unlock pair on the freelist? It's not the
uncontended case I'm worried about. It is just that I've
seen these huge >5000 user servers on multiprocessor
machines suffer from considerable time spent waiting in the
fcntl that locks the freelist. I have to think about how I
benchmark those and still keep a sane freelist :-)

-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: not available
Url :

More information about the samba-technical mailing list