On Thu, 8 Mar 2007, Kejun (Kevin) Dong wrote:
> Hi Adam,
>
> Thank you for your answer. The memory of our ce node uses to be eaten in
> each two weeks and I think it is mostly because of mysql process. Please
> see the result of ps auxww|sort -rn +3|head below.
> Next time, when the memory is run out of again, I will stop the mysql
> process to see if the memory will be released.
>
> [root@ce-lcg root]# ps auxww|sort -rn +3|head
> root 5429 0.0 0.3 54320 27360 ? S Mar07 0:42 /usr/sbin/maui
> mysql 5488 0.0 0.2 56976 19280 ? S Mar07 0:00 /usr/sbin/mysqld
> --basedir=/ --datadir=/var/lib/mysql --user=mysql
> --pid-file=/var/lib/mysql/ce-lcg.sdg.ac.cn.pid --skip-locking
> mysql 5487 0.0 0.2 56976 19280 ? S Mar07 0:00 /usr/sbin/mysqld
> --basedir=/ --datadir=/var/lib/mysql --user=mysql
> --pid-file=/var/lib/mysql/ce-lcg.sdg.ac.cn.pid --skip-locking
> mysql 5486 0.0 0.2 56976 19280 ? S Mar07 0:59 /usr/sbin/mysqld
> [...]
Note that mysqld is a multi-threaded program and "ps" counts the
shared memory for each thread, so killing the mysqld will _not_
make a big difference.
Those sizes are interesting, though.
On the CEs at CERN we typically see this:
mysql 17798 0.0 0.4 31836 10008 ? S Feb10 0:00
/usr/sbin/mysqld --basedir=/ --datadir=/var/lib/mysql --user=mysql
--pid-file=/var/lib/mysql/ce101.cern.ch.pid --skip-locking
mysql 18302 0.0 0.4 31836 10008 ? S Feb10 0:17
/usr/sbin/mysqld --basedir=/ --datadir=/var/lib/mysql --user=mysql
--pid-file=/var/lib/mysql/ce101.cern.ch.pid --skip-locking
[...]
Why would your mysqld be 25 MB bigger? Is your CE also running
other services that may use MySQL?
|