I want to use SyncThing 1.1.0 on an ARM board, a NanoPi NEO2 (with a paltry 512MB RAM, and a 500GB SSD hooked up with the “NAS kit” accessory board), running Armbian Stretch. This OS runs from a nice new MicroSD card, which is a Sandisk Ultra, with “A1”. The CPU is quad core, 900MHz, H5 CPU.
As I look around at expectations for RAM usage, it seems the actual syncing takes a sensible amount of RAM, say 150MB-ish for a repo, but the GUI itself, if just left to sit and run for a long time, can eventually hog more RAM (at least, that was a sore spot in the past). I get the sense my 512MB could potentially run out on me, if I’m not careful.
I plan to have 1 repo, starting with 61GB of data (this is on a nice new SSD, where reading and writing are at 20-40MB/sec, sustained, through that NAS add-on-board connection). This 61GB is 40,542 items at present, growing and changing pretty slowly over time.
So how does this sound?
At first I’ll run syncthing on the command line (on the ARM board, over SSH) with “-no-browser”. Then a GUI waits on localhost port 8384. I’ll use an SSH tunnel (using “-L”) to actually get to that GUI, as this ARM board is headless.
Once I’ve used the GUI to “Add Remote Device” to a remote machine (also running ST), then I “Add Folder”, etc. such that a sync actually starts, then I’ll “drop” the GUI, by closing the browser window, but leave syncing to continue on the console.
Is this all I really need to do, to keep my RAM usage tight on this ARM board? Just close the browser GUI whenever I don’t need it?