Hello Heikki,

On Wed, 2003-06-04 at 16:27, Heikki Tuuri wrote:

> actually, it probably is reporting the right query :). It is the TEXT column
> `FEEDBACK` text. TEXT is essentially the same as a BLOB.
> 
> 
> > I have exorcised the code we have that updates and interacts with this
> > table in order to try and cause the server to crash to no avail.
> >
> > What's next?
> 
> I have added some diagnostic code to 4.0.14 which would reveal a leak in the
> mutex.
> 
> I am also running a simulation of your workload on our SuSE Linux 4-way
> server.
> 
> select MEET, count(*) from RATINGS_WHENU where MEET in ('N','Y') and SITE
>  = '21' group by MEET;
> +------+----------+
> | MEET | count(*) |
> +------+----------+
> | N    |      137 |
> | Y    |      136 |
> +------+----------+
> 
> The assertions are so strange that I suspect thread stacks get corrupt. That
> can be a mysqld bug, an out-of-memory problem, or a Linux/glibc bug.  An
> upgrade to Linux-2.4.20 is recommended.
> 

We are already running 2.4.20. :/

> Also try adding that index to SITE. That changes the query plan and might
> mask the bug whatever it is.

I have done this, and will report if it diminishes or masks the problem.

We cannot upgrade to 4.0 right now as we have several large deployments
and it's a large undertaking to recompile everything with the right
client libraries etc.

Please report back with any new findings you have, as will I.

Thank you,

Richard

-- 
Richard F. Rebel
[EMAIL PROTECTED]
t. 212.239.0000


-- 
MySQL General Mailing List
For list archives: http://lists.mysql.com/mysql
To unsubscribe:    http://lists.mysql.com/[EMAIL PROTECTED]

Reply via email to