On Mon, May 31, 2004 at 02:03:09PM +0200, Michael Arndt wrote: > > Problem for this variant: merge table has to be dropped and recreated > periodically. > during the short lag interval the log merge table > does not exist
You don't have to drop and recreate it in MySQL. You can do it under MySQL's feet without it noticing. The .MRG definition file is simply a list of tables that make up the merge, with a couple of optional configuration likes; the only one I use is #INSERT_METHOD. You can create a new .MRG file very easily in Perl, consisting of the new range of tables, then move it over the old .MRG, so in one atomic operation the merge table now has a new subtable; in order to get MySQL to notice however you have to do a flush tables. You can use mysqladmin flush-tables to do that. For insertion, this is where #INSERT_METHOD comes in handy. MySQL can either insert to the first or last table: $ tail -3 MessageIDs.MRG MessageIDs_20040529 MessageIDs_20040530 #INSERT_METHOD=LAST Now INSERT INTO merge_table will be inserting into MessageIDs_20040530, which is today. Just after midnight, my Perl runs and makes a new .MRG, and after the flush MySQL is inserting into the next table without even realising. -- Chris -- MySQL General Mailing List For list archives: http://lists.mysql.com/mysql To unsubscribe: http://lists.mysql.com/[EMAIL PROTECTED]
