Forgive me if this is a linux and/or system/os-specific question. I am supporting a MySQL 4.0.x replication setup (2 slaves) that handle ~10K qps and ~5000 persistent connections each.
It's obvious to me that system interrupts and context-switching will be high with 2000+ mysqld procs, but generally performance is very good. Occassionally though, the load average of one slave surges up to 10-20 and stays there. My load average (each box has two CPUs w/ HT enabled) is typically <2. Two questions: 1. Is this an artifact of simply having too many active processes vying for CPU time? 2. Are the "2nd" and "3rd" mysqld processes "dispatch" processes? such that having too many mysqld threads, cause these to thrash? I always assumed these two were the slave and i/o threads. By 2nd and 3rd, i mean the first two threads that are spawned by the root/ parent mysqld proc. Only 50% of real memory is utilized, so no swapping. I tried to strace the first few procs, but I'm not sure what I'm looking at there. Thanks. -- MySQL General Mailing List For list archives: http://lists.mysql.com/mysql To unsubscribe: http://lists.mysql.com/[EMAIL PROTECTED]
