Hello .Throughout last year I bought couple hundreds of Neo and Noe2 units for our project of kind of data collector.
/MySql, PHP, Apache, FTP, our app C++, software watchdog/. The project is really growing, now we suffering from scale effect.
Due to this effect found that units randomly by itself (from my perspective) turn to readonly FS with limited functionality (eg. no Apache nor FTP) or even cannot boot Linux after reboot or shudown -r or even reboot -f command.
So I was thinking about how to prevent from such behavior which is caused, in my opinion, by power connected micro issues like rapidly limited power condition for milliseconds etc. (board can hang forever on transient state - until next power recycle).
I have two ideas: first is simple - hardware watchdog, to recycle power when transient state detected. But board from time to time is no booting since some files on sd card are broken. I started to thinking about read only-only RFS to prevent /boot and /root from being damage. My idea is to create additional partition p03 (-RW) for MySQL storage and turn the rest to RO. To be clear, my goal is not data backup (it's important but not crucial) but increase reliability, by being able to power recycle to my own "factory reset" (our data collector is constantly transmitting data to upper level server software). All my data sources: fresh config and app will be kept on RO /root partition and as soon as p03 partition data had been broken just will send power-recycle signal and rebuild the whole p03 config from fresh config from /root partition.
Could You give me a clue how to deal with my problem, and achieve the goal?
Was trying to deal with "overlayroot" from: https://unix.stackexchange.com/question ... able-layer and https://spin.atomicobject.com/2015/03/1 ... ilesystem/ but for some reason it dosen't work on NanoPI ubuntu nor debian distro.