Crashing repeatedly: 6.2-RELEASE-p5 and MySQL 5.0.41
primeroz.lists at googlemail.com
Wed Feb 6 09:31:08 UTC 2008
Yes i agree with everything. Definetly mysql need to be tuned for InnoDB and
in general .
As stated in the previous post my a collegue of mine i had to install a new
kernel to have a consistent crash coredump.
Anyway still in my mind is that even if not tuned mysql should not cause my
kernel to panic, i could expect a mysql crash or very very poor performance
... but the kernel panic leave me confused.
I will post on this thread my coredump as soon as the server crash again
(and it will) ... surely i will tune mysql then and see if and how much this
On Feb 5, 2008 7:37 PM, Tom Samplonius <tom at samplonius.org> wrote:
> ----- "Primeroz lists" <primeroz.lists at googlemail.com> wrote:
> > Hi all,
> > we are experiencing repeated crash on a Dell PowerEdge 2950 (rev 1 or
> > 2).
> > FBSD release is 6.2-RELEASE-p5 , AMD64. 2xXeon QuadCore and 8G of Ram.
> > MySQL Version is 5.0.41 with following configuration settings:
> > set-variable = key_buffer=768M
> > set-variable = table_cache=800
> > set-variable = sort_buffer=24M
> > set-variable = myisam_sort_buffer_size=256M
> > set-variable = record_buffer=16M
> > set-variable = max_allowed_packet=10M
> > set-variable = thread_stack=128K
> > set-variable = join_buffer=512M
> > set-variable = max_heap_table_size=256M
> > set-variable = max_connections=300
> > set-variable = tmp_table_size=384M
> > set-variable = query_cache_size=402653184
> > set-variable = query_cache_limit=134217728
> > set-variable = read_rnd_buffer_size=10M
> > set-variable = ft_min_word_len=1
> > pid-file = /var/db/mysqld.pid
> > tmpdir = /var/tmp
> > ft_stopword_file = ''
> > set-variable = thread_cache_size=80
> > set-variable = myisam_stats_method=nulls_equal
> Also, myslq is not really well tuned.
> The query cache is a kludge. It is helpful, if you have stupid
> application that issues the same query over and over again, even though the
> database has not changed. If you don't have this problem, it just adds
> overhead. And quite a lot, if it is big. Generally, the query cache should
> be 20 to 100M at most, if not disabled. If you have a smart web application
> (anything using memcached), the query cache should just be turned off. It
> will actually be faster.
> You should give us much storage as possible to the database engine, for
> it cache actual data, not query results. It is weird that you are
> apparently are heavily using Innodb, but you have just set various myiasam
> Here is something useful:
More information about the freebsd-stable