I developed a web-based app for a client to store huge amounts of PDF data - we setup a 1.2TB drive for the purpose. Recently, the drive became full. Prematurely.
Database statistics showed that the PDF table was only 400GB in size - so where was the rest of our data going? As with most tech issues, the problem was only solved when the heat was on - unable to gracefully restart the database (because we upgraded the Gentoo init script), we had to kill the process.
Resulting in corruption.
On a 400GB database.
Which couldn’t repair itself, since the myisamcheck functions require free space to make a backup copy, before making any changes. Aaarg!
A bit of research brought the answer.. we had the “log-bin” entry in “my.cnf”. This told MySQL:
“Create a binary log of every insert we’ve ever done. Since we haven’t specified a filename, use the server’s hostname, followed by “-bin” and a number”
I zapped the 600GB of 10GB-sized data-bin files, and restarted myisamcheck. Hopefully it’ll be finished before lunch!
No Comments on "Harassed by MySQL’s binary logs"