On Saturday 12 July 2003 22:22, Dan Nelson wrote:
> In the last episode (Jul 12), Tim Fountain said:
> > This may be a silly question but what can be done to stop
> > load-intensive MySQL processes taking down a server? Things like
> > adding fulltext indexes to very large tables, or selects on very
> > large (multi-million-row) tables just completely kill the box until
> > they complete.
> >
> > I don't mind how long these things take but the box shouldn't become
> > unresponsive while they are running.
>
> Are you swapping? Insufficient memory causing the OS to swap processes
> out is the only thing I can think of that could cause this. You
> usually have to be *REALLY* low on memory (like a 512MB process on a
> 128MB RAM system) to cause total unresponsiveness, though. Try adding
> more RAM or limiting mysql's use of it by lowering buffer sizes in
> my.cnf.
However MySQL does have a tendency to grab all the cpu-power it can get, and
running un-niced (which it is supposed to) it would reduce the responsiveness
of ssh (or console-sessions)...
The best way is to add another CPU (ie. SMP or more) that way the thread
adding an index (or somethingelse CPU-intensive) would just grab one CPU
leaving the other(s) to do system and other stuff... (serving other requests
for instance).
--
Andreas D. Landmark / noXtension
