> By the way, Jeremy's original answer was more correct than mine, because > he noted where I did not that the sort buffer is a per-client resource > and is allocated once for each client -- or at least for each client that > issues queries requiringn sorting, such as those with ORDER BY clauses.
Thanks Paul and Jeremy for the replies. That helped out a great deal and I very well may recompile the MySQL installs on my FreeBSD boxes using LinuxThreads. Now I'm wondering just how optimised (or non-optimised as the case may be) my configurations are. Below are some settings specified in the my.cnf of a linux box with 2Gb of memory that I'd say roughly 75%-80% of its purpose in life is dedicated to MySQL: set-variable = key_buffer=256M set-variable = max_allowed_packet=1M set-variable = table_cache=64 set-variable = sort_buffer=2M set-variable = net_buffer_length=8K set-variable = myisam_sort_buffer_size=2M set-variable = max_connections=1000 set-variable = thread_concurrency=10 innodb_data_file_path = ibdata1:10M:autoextend set-variable = innodb_buffer_pool_size=40M set-variable = innodb_additional_mem_pool_size=40M set-variable = innodb_log_file_size=5M set-variabl e = innodb_log_buffer_size=5M innodb_flush_log_at_trx_commit=1 set-variable = innodb_lock_wait_timeout=50 Am I allocating too little memory to table_cache, sort_buffer_size, and innodb_buffer_pool_size given the 2Gb of memory? So I can also adjust these settings on other boxes, is there a way to mathematically determine what percentage of totally memory to set these variables to? Thanks again. Cheers, Mark -- MySQL General Mailing List For list archives: http://lists.mysql.com/mysql To unsubscribe: http://lists.mysql.com/[EMAIL PROTECTED]
