I would like to setup a small system based on CentOS6
“power outage”-save as possible. The hardware will be switch off by pulling the plug.
To accomplishing this goal, I would mounting some fs parts readonly (e.g. /usr) and thinking about tmpfs for volatile parts (e.g. lock, run under var). Additionally “optimize”
some vm.dirty_* kernel- and fs/ext4 parameters. /persistent would be used with jffs2 on a CF card. So far the theory.
Does anyone have some experience with such type of systems?
Any pointer to pitfalls are welcome.