MySQL maximum memory usage

Viewed 278742

I would like to know how it is possible to set an upper limit on the amount of memory MySQL uses on a Linux server.

Right now, MySQL will keep taking up memory with every new query requested so that it eventually runs out of memory. Is there a way to place a limit so that no more than that amount is used by MySQL?

7 Answers

Database memory usage is a complex topic. The MySQL Performance Blog does a good job of covering your question, and lists many reasons why it's hugely impractical to "reserve" memory.

If you really want to impose a hard limit, you could do so, but you'd have to do it at the OS level as there is no built-in setting. In linux, you could utilize ulimit, but you'd likely have to modify the way MySQL starts in order to impose this.


The best solution is to tune your server down, so that a combination of the usual MySQL memory settings will result in generally lower memory usage by your MySQL installation. This will of course have a negative impact on the performance of your database, but some of the settings you can tweak in my.ini are:

key_buffer_size
query_cache_size
query_cache_limit
table_cache
max_connections
tmp_table_size
innodb_buffer_pool_size

I'd start there and see if you can get the results you want. There are many articles out there about adjusting MySQL memory settings.


Edit:

Note that some variable names have changed in the newer 5.1.x releases of MySQL.

For example:

table_cache

Is now:

table_open_cache

If you are looking for optimizing your docker mysql container then the below command may help. I was able to run mysql docker container from a default 480mb to mere 100 mbs

docker run -d -p 3306:3306 -e MYSQL_DATABASE=test -e MYSQL_ROOT_PASSWORD=tooor -e MYSQL_USER=test -e MYSQL_PASSWORD=test -v /mysql:/var/lib/mysql --name mysqldb mysql --table_definition_cache=100 --performance_schema=0 --default-authentication-plugin=mysql_native_password

Since I do not have enough reputation points to upvote a previous answer, I concur that the "table_definition_cache = 400" answer worked on my old Centos server.

Related