Can you recommend some devices? Most of the ones i saw had good prices, but not performance relative to power usage. The N100 with its 4 efficiency cores is actually quite good for the price and power usage. Unfortunately most mini pcs with it have limited ports.
I also think, that 2 ssds might be sufficient for the beginning. I'm even thinking of just adding 2 external ssd's and call it a day for the beginning (one as backup), but that does not scale well.
The jbod idea sounds good to explore further, as it tha home server and storage would be separated. However it would add an additional device to the power bill.
However i don't need the full amount of all disks at all times. If i'd want to unplug via shell script, i'd need to plug it manually in person back in for storing things. I actually do not need it running all the time, as the home server ssd can cache most of what i need recently in access. The jbod is then more an archive.
i'm mainly looking for a way to power down the inexpensive hdd's. I could use the raspberry pi as the jbod controller, but it does not properly support wake on lan, so thats also not an option
I don't know how the code is currently working, but I like this feature idea and would suggest to start very simple and proceed from there.
For example you could: a) Make a list of communities that are siblings with their id and instance b) add a toggle to view sister community posts yes /no c) query all communities, list the last x posts from each with time constraints, e.g. not older than 1 day or hour depending on the community post frequency d) list them sorted by time of x , depending on what was chosen
The biggest issue I see with this simple approach, besides others, is that different communities are different in terms of activity / post frequency. So ideally the better, but more effort, way would be to let each community instance communicate their posts themselves via a query with activity metric parameters. Basically the amount of returned posts would depend on common parameters set by the most active instance.
It's not yet thought out, but just getting an mvp started and test the waters would probably be better than having it perfect right away while working on it for months