Yesterday afternoon, after making changes to the binary logging (as reported in the previous post), the Blat database went down again. So, last night I spent about three hours looking at various settings on the MySQL database. I tweaked a few settings, and was running a process monitor to try and figure out exactly what was causing the database to keep falling over. I saw nothing untoward.
6 a.m. This morning and the database was dead again! Damn!
It has taken me three hours this morning to get control of the Blat server and reboot it. When the database goes stupid, the server runs out of memory and starts shuffling data in and out of virtual memory - Virtual memory is disk based memory, which makes it incredibly slow, which, in turn, makes the server unresponsive - Hence the three hours.
I’ll be working on the server today, doing my best to get to the bottom of this problem.
