MySQL hangs server completely
aimass at yabarana.com
Wed May 22 18:11:30 UTC 2013
On Wed, May 22, 2013 at 2:04 PM, Michael Ross <gmx at ross.cx> wrote:
> On Wed, 22 May 2013 15:52:45 +0200, Alejandro Imass <aimass at yabarana.com>
>> We've been having this problem with a customer for a while and it
>> seems that some funky query makes MySQL use 100% of CPU. Nevertheless,
>> even though you can see in top that it's only 1 CPU in 100% (out of 8)
>> the server eventually becomes useless and stops responding completely.
>> So my question is, how does a user process hang the whole server? What
>> system resources could MySQL be draining to make the server stop
>> responding completely?
> In laymans terms - can't do better - MySQL racing itself to obtain a ( table
> | memory | file ) lock?
> I know I can death-stall the MySQL server at a customer's site if I give it
> a big enough query ( like, DROPping a table, recreating it and pushing
> backup data inside ) while cron's hourly backup-dump is running on the
> database. Just the MySQL server, the machine itself hasn't stalled yet - but
> I'm sitting at the console while doing this, so I don't know what would
> eventually happen if I'd let it sit for a while.
Right on the money. It doesn't immediately hang the server but in time
it drains it to the point the shell stops responding and no more ssh
access and even snmp stops responding! It doesn't happen immediately,
but only after a while that MySQL has one of the CPUs at 100%.
What I don't understand is how it manages to crash the whole server.
More information about the freebsd-questions