I will try that. But the symptoms are a little bit different - the server works fine for 3-4 hours, but when it gets slow, neither cpu nor disk-io reach the limit.
As far as I know a cache/memory-bottleneck should turn cpu to 100%, like you have observed. In our case I guess it's a locking issue, will grab the output of INNODB STATUS the next time it happens. Moritz -----Original Message----- From: Gabriel PREDA [mailto:[EMAIL PROTECTED] Sent: Wednesday, July 12, 2006 11:22 AM To: Moritz Möller Subject: Re: mysterious speedup after doing "FLUSH TABLES" See http://lists.mysql.com/mysql/199572 As lower as you can afford... The same thing happened here... MySQL will work fine for let's say some hours then it begins struggling... then the CPU load increases and machines works like a bull in the Spanish-style bullfighting... We have it now at 2 and the highest load we've seen it's 35%... at 900 simultaneous connections... Since the implementation of 'memcached' the workload on the MySQL dropped significantly... now 900 simultaneous connections is enough to keep the website at it's best ! -- Gabriel PREDA Senior Web Developer -- MySQL General Mailing List For list archives: http://lists.mysql.com/mysql To unsubscribe: http://lists.mysql.com/[EMAIL PROTECTED]
