Rescan on node connecting


First of truly awesome work on syncthing. It works like a charm! Syncing 100 gigs of valuable data on a beagle bone black is easy as pie.

My apologies if this has been asked before.

The structure of my backup network hinges around a central node that does the back-up for other nodes ( A <-> BeagleBone <-> C). This node is always online and has an external hard-disk connected. The drive is spun down when there is no activity (sometimes days). However rescanning the local content requires a disk spinup, most of the times only to find no changes.

As far as i understand sync thing (I’m not very good with go) it only syncs when boths ends have differences in recorded state (detected by scanning). A long time to rescan time thus has the possibility that sync windows are missed? Rescanning before a node connects then enables fast and efficient synchronization of data, only it requires disk spinups.

Would it be a smart tradeoff to have a feature that rescans only happens when a node connects (useful for a central node)? The start to sync time is slightly longer but the disk can sleep when no node connects. Furthermore this would mean that a fresh ‘status localis’ is available on every reconnect, ensuring optimal sync coverage by the central node. Of course with a window that prevents new rescans when nodes connect just seconds after the last scan.

Thanks again for the awesome tool!


You can disable the rescan interval and use syncthing-inotify.

1 Like