From: Sergei Golubchik Date: February 24 2012 8:16pm Subject: Re: handlerton info method locking List-Archive: http://lists.mysql.com/internals/38461 Message-Id: <20120224201629.GA2605@meddwl.fritz.box> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Hi, John! On Feb 24, John Esmet wrote: > Any insight on this? I'm wondering if it's necessary to do my own > table-level locking (say, for shared variables among threads on the > same table) inside the handlerton's info method. Yes, it is. It could be something simple, like a mutex that protects shared data, but you need to take care of locking it. There's HA_STATUS_NO_LOCK though. But it doesn't mean that all locking is done for you. See the comment in my_base.h. Regards, Sergei > On Thu, Feb 23, 2012 at 3:07 PM, John Esmet wrote: > > Hello all, > > > > What is the expected locking behavior of 'show table status' and the > > ::info method in the handlerton? I'm observing unlocked access to > > ::info when other threads have TL_WRITE locks on the table, so I'm > > curious if it is normal/expected for threads to call ::info on a > > write-locked table.