Date: Sat, 2 Oct 2010 15:30:05 -0700 From: Jeremy Chadwick <freebsd@jdc.parodius.com> To: Rumen Telbizov <telbizov@gmail.com> Cc: freebsd-stable@freebsd.org, Steven Hartland <killing@multiplay.co.uk> Subject: Re: MySQL performance concern Message-ID: <20101002223005.GA78136@icarus.home.lan> In-Reply-To: <AANLkTi=WmvuDEkzbLZQ8BRLpCZtq8R8jKgjPUZmFc4fy@mail.gmail.com> References: <AANLkTikGHByF0dJ-hj6zPoRhV6YoHGSrW3g0audama3M@mail.gmail.com> <ACD11509724249559BD89DDD26F67C62@multiplay.co.uk> <AANLkTi=WmvuDEkzbLZQ8BRLpCZtq8R8jKgjPUZmFc4fy@mail.gmail.com>
next in thread | previous in thread | raw e-mail | index | archive | help
On Sat, Oct 02, 2010 at 01:18:20PM -0700, Rumen Telbizov wrote: > Hello everyone, > > Here's the requested information below: The tunings between your Linux and FreeBSD instances differ severely, and some of the variables don't even exist any longer (example: table_cache is now known as table_open_cache as of MySQL 5.1.3, and probably key_buffer vs. key_buffer_size too). Can you please rule out MySQL tunings being responsible for the problem? Here are your configuration bits, more sanely written: FreeBSD Linux -------------------------- -------------- ------------------ MySQL version 5.1.51 5.1.50 -------------------------- -------------- ------------------ my.cnf tuning FreeBSD Linux -------------------------- -------------- ------------------ key_buffer_size 8 GB ???? key_buffer ???? 4 GB max_allowed_packet 16 MB 1 MB table_open_cache 2048 ???? table_cache ???? 64 sort_buffer_size 64 MB 512 KB read_buffer_size 8 MB 256 KB read_rnd_buffer_size 16 MB 512 KB net_buffer_length ???? 8 KB myisam_sort_buffer_size 256 MB 8 MB thread_cache_size 64 ???? query_cache_size 32 MB ???? thread_concurrency 8 ???? max_heap_table_size 6 GB 6 GB -------------------------- -------------- ------------------ Can you also please provide "top" output for the mysqld process on FreeBSD? > * As for the ZFS. Here's the pool configuration: If you move things to UFS2, does the problem disappear? You might not be seeing any disk I/O on the filesystem with gstat because ZFS ARC could have all of the data in it. -- | Jeremy Chadwick jdc@parodius.com | | Parodius Networking http://www.parodius.com/ | | UNIX Systems Administrator Mountain View, CA, USA | | Making life hard for others since 1977. PGP: 4BD6C0CB |
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20101002223005.GA78136>