MySQL Binary Logs Fill Disk on OSGeo2

[warmerdam]

This directory appears to be full of binary logs on osgeo2, similar to the issues on osgeo1 mentioned in #126.

[root@osgeo2 etc]# du /var/lib/mysql/.a
177892908       /var/lib/mysql/.a

[warmerdam]

I'll attempt to address this.

[warmerdam]

I attempted a "RESET MASTER" but it did not appear to have any effect.

[root@osgeo2 .a]# mysql 
Welcome to the MySQL monitor.  Commands end with ; or \g.
Your MySQL connection id is 212 to server version: 5.0.27-standard-log

Type 'help;' or '\h' for help. Type '\c' to clear the buffer.

mysql> reset master
    -> ;
Query OK, 0 rows affected (0.01 sec)

The /var/lib/mysql/.a directory is unchanged.

I'm starting to wonder if the directory is called .a for some particular reason. Perhaps by someones action?

[warmerdam]

Martin,

I don't see you in IRC. Would you like me to take care of cleaning up the mysql logs problem? I'm just worried about us colliding on this issue. I'd also kind of like to restore the stuff you put in the .a directory so I can clean this up in the mysql intended way. But I don't know exactly what you did.

[warmerdam]

Martin writes:

For clarification: I've been moving all the mysql-bin. files into a (hand-made) subdir .a to see, if MySQL will resume running afterwards. In other words: To see if MySQL really still needs these files or if it can live without. I simply didn't know better than just trying it out.

If you came to the conclusion that MySQL does _not_ need these old files (at least this is my conclusion) then simply remove the whole .a/ subdir and you'll be fine for a while. I have to leave urgently (having a concert this evening: Brahms Requiem), therefore I don't have the time to show up on IRC,

I have:

• deleted the contents of /var/lib/mysql/.a
• disabled binary logs using "SET SQL_LOG_BIN=0"
• disabled binary logs on startup by commenting out the "log-bin=/var/lib/mysql/mysql-bin.log" in /etc/my.cnf added originally my Shawn.

I believe this should permanently correct this problem. Based on the review in #126 we established that binary logs were overkill, and we weren't monitoring them in a useful fashion.