Jay Miller wrote:
> I have read that --skip-locking is enabled by default on Linux systems
> because of a bug in the Linux file locking mechanism. This leads me to
> wonder about when locking is important with MySQL.
> I have one MySQL daemon running on Slackware Linux 2.2.5. There are several
> tables and many client machines, but there are two clients that both INSERT
> into the same large (+3 million records) table multiple times per second.
> This table becomes corrupt somewhat regularly. Is there potentially a
> problem with more than one MySQL thread updating the table at the same time?
> Or does MySQL handle preventing this sort of problem?
> Jay Miller
The --skip-locking has nothing to do with multiple client access.
Mysql provides a locking mechanism for this in itself.
This flag only prevents locking of the data files on the filesystem level.
This filesystem locking only is useful, if two mysqld servers are using the same data
With two servers running there is a chance, that both servers try to UPDATE, INSERT or
DELETE the same table file.
This only can be syncronized on filesystem level, because the two mysql servers don't know
If you don't run two mysql servers (or one mysql server and an isamchk) on the same data
files, you never need this filesystem locking mechanism.