Fastest way to initialize sync copies?

I have a laptop with about a dozen shares (Music, Projects, Notes, etc) thousands of files in each, and a total size in the hundred GB range. I just bought a new one and want my files there in sync.

What is the fastest way to do it with reasonable CPU / memory usage? I can think of several approaches:

Naive Approach: add all the shares on syncthing, and let syncthing do its thing. Come back maybe a week later after it’s all sorted out, but during that time you can’t use either laptop very well due to ridiculous CPU and IO spikes. Make sure to put a fan on both constantly as the CPU / IO tends to heat them both up.

^ I started with that but I don’t ever remember having the patience to wait for it.

Pre-synced approach: use rsync to force both laptops to sync first. Then add all shares on the second lappy and you can’t use it for the about 1 week it takes to finish indexing. But at least you can use the first lappy (not IO-ing so much) and the second lappy somewhat slowly, with full confidence you have your data.

^ I moved on to this one after giving up on the first.

Pre-synced but one at a time: same as above, but pause all shares except one so that only one share syncs at a time. Every once in a while double check the syncthing GUI to see if its done, then unpause another share. And eventually you have your data and you can somewhat reasonably use both lappy’s without the IO-monster taking your lunch.

^ I moved on to this one after giving up on the one before.

Some magic operation that allows you to send both the files and pre-calculated indexes on the first one and force the second one to be automatically “live” without extra work, CPU, or IO. This would be an “advanced” operation and might result in conflicts and inconsistencies if done incorrectly (because you’re basically tricking the syncthing db) - but only for the special case of initializing an empty node would do the trick faster than any of the above.

^ i’m really looking for this one, although as best as I can understand - it’s not there and everyone seems to advocate letting syncthing eventually settle down from maybe the third one.

Any advise?

I’d go with the second option: with a reasonably powered laptop I wouldn’t expect syncthing to take much more than a night for a few hundred gigs.

Regarding the 4th option: at least in earlier versions it was possible to recreate the environment of the “source” and then copy both data and index over to a new machine. That means having all paths and files exactly the same and then starting syncthing. I’d consult calmh before trying this with the newer versions. Haven’t heard about anyone trying this and could imagine that newer versions don’t like that anymore

This topic was automatically closed 30 days after the last reply. New replies are no longer allowed.