Hello,

Thanks for your answer!
I've re-formatted the whole server and I'm re-installing Gentoo 32 bits
instead of 64 bits.
I want to see if I face the same problem.
I'll keep you posted.

Regards,
Thomas.

On 1/18/07, kashani <[EMAIL PROTECTED]> wrote:

Thomas Balthazar wrote:
> Hello,
>
> I'm using Gentoo Base System version 1.6.14 on a x86_64 Intel(R)
> Celeron(R) CPU 2.66GHz.
> I've added "dev-db/mysql innodb berkdb" to my package.use then I've run
> emerge -1 dev-db/mysql.
>
> I've installed PHPMyAdmin that is up and running (MySQL 5.0.26).
> When I try to create a Innodb table, I get an error :
> #2013 - Lost connection to MySQL server during query
>
> After that, I cannot stop or start my MySQL server.
> Everything seems to be corrupted, and all I can do is to erase all the
> content of /var/lib/mysql and restart from scratch.
>
> Has anyone heard of problems with MySQL/InnoDB/Gentoo?
>
> Any help would be much appreciated!
> Thanks in advance,
> Thomas.

Couple of things on this.

        This whole community vs enterprise is making things a bit weird at
the
moment for ebuilds. For Innodb I highly recommend going with the
enterprise build, dev-db/mysql which you've already installed, and using
the ~arch version of 5.0.32. It fixes a number of high concurrency/multi
thread issues in Innodb and I'd move to it sooner rather than later.

        Have you modified your my.cnf at all? The default Innodb settings
are
TINY. Assuming you have at least a 1 GB of RAM in you machine I'd bump
the following setting up so that you can fit real tables into Innodb.

#innodb_buffer_pool_size = 16M
innodb_buffer_pool_size = 128M

        Innodb buffers and general Mysql buffers like key, sort, etc are
managed separately. If you're starting to migrate things into Innodb
from Myisam you might need to decrease some of the current buffers if
you've got limit RAM.

        The Gentoo Mysql startup script is a bit retarded when starting
Mysql
with Innodb tables turned on the first time, at least with large tables
and log files. I use two 512M log files in production and the startup
script fails though Mysql is actually running, it's just pausing to
write the log files and initial ibdata files out. In your case I'd start
and stop Mysql a few times before trying to create an Innodb table just
to be sure that Mysql is finished with all the file writes.

I suspect the issues is Innodb not having enough memory assigned to it
rather than the binary being borked. You might also try creating a
simpler table in Innodb and see if you have the same issues.

I'd also recommend adding the setting, innodb_file_per_table, so that
each table gets it's own ibdata file in the form of
lib/mysql/$db/$table.idb. It performs better and it is a bit easier to
tell how big your db is on disk or which db is using all your disk.

kashani
--
gentoo-user@gentoo.org mailing list


Reply via email to