Kasthuri Ilankamban [EMAIL PROTECTED] writes:
Hi, We are running mysql version 4.1.7 with innodb on i686 running
2.4.26 linux kernal with 8G memory. Mysql crashes consistently during
heavy usage with fatal innodb error. We are running a high volume front
end application which inserts 50M
Kasthuri,
Maybe it's time to re-think your application architecture? A 200-meg
BLOB is quite large for a highly-concurrent system, considering that
MySQL will have to read/save it in its entirety _and_ allocate network
buffers for it, so essentially you're allocating _400_ megs or so
_per_
Thank a lot for your quick reply. SInce we are not using myisam
tables ( except for the system tables), I deallocated memory from
myisam and allocated to innodb. When I allocated close to 1.7G to
innodb buffer size, mysql used to crash more often. So I decreased
innodb_buffer_size to 1G.
-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1
Kasthuri Ilankamban wrote:
[snip]
The table that's it's complaining about is a session table which
stores web session information. Basically the data in that table is a
throw away data and gets deleted when the user logs out. The session
A big part of your problem seems to be that 32bit OS's can not allocate
more than 2G memory to a single process, and with each client connection
needing so much buffer space, you quickly exceed that limit. Short of
moving to a 64bit OS, the only solution I have come across that allows
mysql to
Matthew has already responded. But I will also add, you must rethink
how your application is coded. I can not possibly imagine a situation
where a user session needs 200meg insert. No matter what database you
use you will have a terrible time trying to scale this application. If
you give more
Hi, We are running mysql version 4.1.7 with innodb on i686 running
2.4.26 linux kernal with 8G memory. Mysql crashes consistently
during heavy usage with fatal innodb error. We are running a high
volume front end application which inserts 50M data to a row in
innodb table often. I don't
You definitely should increase memory sizes in your my.cnf file. The
settings that you have are for a very smal setup. You also need to
allocate more space for innodb table extender. So instead of 10m have
something like 50m or try and see which settings is better. The reason
is under heavy load
