Tim:
Can you bring your libc to the latest patch level?
Not necessary. I resolved the problem:
binlog_cache_size was set to 32MB
I didn't realise that this would automatically be allocated to every thread, even if there are no InnoDB or BDB tables in the entire instance. This explains why --skip-innodb "fixed" the problem; without InnoDB, MySQL knew that there would be no transactions occurring, and so would not need the binlog cache for each thread.
The documentation probably needs clarifying that this is another per-thread buffer, and it is always allocated to every connection thread if the server supports transactional table types.
As a followup question; what happens to the binlog cache if a thread requires more? Does it automatically increase it as needed (up to an eventual limit of max_binlog_cache_size)?
Tim
-- Dr Tim Cutts Informatics Systems Group Wellcome Trust Sanger Institute Hinxton, Cambridge, CB10 1SA, UK
-- MySQL General Mailing List For list archives: http://lists.mysql.com/mysql To unsubscribe: http://lists.mysql.com/[EMAIL PROTECTED]
