On Fri, 2007-03-02 at 00:45 +0100, Christian Hammers wrote: > Hi > > On 2007-02-26 Jeroen Coekaerts wrote: > > The init.d script does a kill -9 if the server has not shutdown afther 10 > > seconds. This causes table corruption. I have had an unrepairable mysql > > user table afther a mysql restart. > > The script first tries the normal "mysqladmin shutdown". If that for > some reason gives an error message a "kill -15" is tried (internally the > same as the shutdown command). The server then has to stop immediately by > request of the system. Only if that does not work, too, it is killed as > mysqld usually hangs in an endless loop in such situations. > > The only thing to do after a SIGTERM is to flush those buffers that contain > modified data and have not yet been written to disk. And a database is > usually very eager to quickly write its data to disk where it's save. > > So what situation did you have where it needed more time? > > And BTW, were there any modifications to the user table in the last hour > before the shutdown? Because else, it was corrupted anyway but just not > checked or used so that you did not notice :)
Hello, Maybe the usertable was already corrupt and that caused mysqladmin to fail. Somebody was adding a field to a big table (8gb) when the restart was issued. Maybe that makes it to slow ;-) It happened twice but i can't reproduce it anymore, guess it was just an already corrupted table. Jeroen -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
