> because of how the root filesystem is built, any changes made to the filesystem will reside in the upper layer of the union [fs] and will be lost on reboot. This is good, because it means that all the state you have on the device is expendable, and you can likely rebuild it after a reboot. If you rely on the state of the device, you will have to figure out a state to persist it outside of cattlepi.
This seems like a showstopper for me because I want to run a webapp with user registration, logs, etc. I'd certainly want to keep these things around after next reboot.
Can the user-writes be persisted on the device? Maybe on another partition?
thank you for the feedback.
there are a few things to consider here: the first one is that the main failure scenario for something like a pi running for a long time (that I've seen) is the SD card failing (long time here means months to years depending of your usage pattern); the second thing is that I made this choice for the first iteration - there isn't any technical reason why you wouldn't be able to have a 2nd partition that you mount RW. to achieve this you could hack it directly into a image you would build or you can leverage the bootcode and/or usercode hooks in the config. (see https://github.com/cattlepi/cattlepi/blob/a5ad6b57dc28313819... and https://github.com/cattlepi/cattlepi/blob/35eccb61029f46076c...) Ping me directly or comment on this thread if you want to give this a shot and you have trouble getting it to work.
I guess I could see this, but personally I've built this to "scratch an itch": I wanted to make playing/experimenting with Raspeberry Pis in a repeatable manner as easy as possible. Resin look like a "batteries included" approach and there is a monetization aspect there. CattlePi is more like a building block.
Only technically so. The parts they send you are open source because of the GPL. A substantial portion of the parts in the cloud (which are 100% required to use the platform in any manner) are not open source. They've been suggesting that they will open source enough of the backend to actually run resin.io on your own (without several features) for 3 years now, most recent target release was 2017.
While you are correct that we have not completed our open sourcing promise on time, and that is our fault, we have (and continue) to release as open source a lot more of our software than we "have to" because of the GPL. If you look at our github orgs, there are hundreds of projects, including the entirety of our operating system, and we did not have to release any of that.
We are taking a long time to release the open source version though, it's hard to deny. If it helps our case at all, this is because we're trying to do this in a way that allows us to make the eventual open source offering a first-class citizen of the ecosystem, receiving updates at the same time as our online service, out of the same repos, which is a non-trivial development workflow issue.
The latest version of resinOS even allows devices to de-provision from our cloud and re-provision onto an open source server (and vice versa). But of course the use of that won't be apparent until an open source server exists.
> because of how the root filesystem is built, any changes made to the filesystem will reside in the upper layer of the union [fs] and will be lost on reboot. This is good, because it means that all the state you have on the device is expendable, and you can likely rebuild it after a reboot. If you rely on the state of the device, you will have to figure out a state to persist it outside of cattlepi.
This seems like a showstopper for me because I want to run a webapp with user registration, logs, etc. I'd certainly want to keep these things around after next reboot.
Can the user-writes be persisted on the device? Maybe on another partition?