Carsten H. Pedersen schrieb am Samstag, 18. August 2001, 19:24:39:
> Sinisa writes:
>> Osus writes:
>> > ...
>> > Hi,
>> > SELECT * FROM table WHERE OK=1
>> > it gets about 4 or 5 second at least.
>> > I have indexes. That main table have 27 indexes to supply all
>> > possible searches.
> ...
>> Having 27 indexes and using queries with LIKE "%...%" is a nice
>> recepee for slowness.
> Why should having 27 indexes slow down a SELECT?
Also I notice that this kind of problem keeps popping up nearly
every day in the list in slightly different settings.
I don't suppose Osus had the problem right from the beginning.
When he gets 10.000 visits/day, he should be in business for
quite some time without it or he wouldn't have come that far.
And with the problem he won't have that many visitors for long
either.
We solved (or at least hope so) the problem with speeding up the
disk (hdparm). Others were not that fortunate, I read. If our
diagnosis is correct, the problem should occur when mysql suffers
from low i/o - I understand that there may be lots of reasons for
this.
If so, there should be a critical load when mysql slows down on
every machine, so the behaviour should be reproducible, right? It
would be nice if mysql could find out itself and give a warning
about this kind of condition.
When we had this problem, we had lots of reading and writing
errors in the error log (more reading, of course) - and these
seem to have not been really hardware reading or writing errors,
but maybe errors triggered by slow i/o. I can't read the source
to find out, but someone will. When is this error raised? What
does it _really_ mean?
BTW, we still have these errors, but not that often. Also, we
still have the other problems as stated in my posts to the list,
not that often as 30 minutes or 5 hours (after adding more RAM),
but one or two days, which still isn't what we want, of course.
Our queries are extremely fast, and when we had this kind of
state, they were slow. Just like that query which Osus gives as
an example:
> SELECT * FROM table WHERE OK=1
> it gets about 4 or 5 second at least.
> I have updated mysql and apache, checked, repaired, optimized
> tables. I run fschk on all the partitions.
That should be done in a tiny fraction of a second. And if it
is not, something weird is going on and I would like to see this
problem really understood and solved. It may turn out to be not a
problem of mysql proper, but who knows at the time being?
--
Herzlich
Werner Stuerenburg
_________________________________________________
ISIS Verlag, Teut 3, D-32683 Barntrup-Alverdissen
Tel 0(049) 5224-997 407 · Fax 0(049) 5224-997 409
http://pferdezeitung.de
---------------------------------------------------------------------
Before posting, please check:
http://www.mysql.com/manual.php (the manual)
http://lists.mysql.com/ (the list archive)
To request this thread, e-mail <[EMAIL PROTECTED]>
To unsubscribe, e-mail <[EMAIL PROTECTED]>
Trouble unsubscribing? Try: http://lists.mysql.com/php/unsubscribe.php
