Hello list, we use MySQL 4.1.13 / InnoDB on Xeon em64t running Debian, kernel 2.6.8.
After a while, queries run slower and slower, although the CPU is >70% idle and diskio is nearly zero. A simple select by primary key takes up to one second. After doing a FLUSH TABLES the speed is up again. I also noticed that 4.1.15 seems to run slower than 4.1.13 but that might be subjective. As far as I know, FLUSH TABLES reopens the tables and clears the query cache. Query cache is enabled, but does not seem to cause the slow down, as I can observe the same effect without the query cache. So it looks as if some buffers run full, or a lock is blocked or something. The queries running do not hang in any specific stage (such as "waiting for lock"). Even stranger: we have 4 servers of that type (hardware, software, configuration all identical) and two of them need a flush tables every now and then, the other two (which have more load, 5000 q/s compared to 1500) are performing without problems. Well, if anyone has an idea what causes the slowdown, and/or why a flush tables helps, I'd be glad to hear from you ;) Thanks, Moritz -- MySQL General Mailing List For list archives: http://lists.mysql.com/mysql To unsubscribe: http://lists.mysql.com/[EMAIL PROTECTED]
