Sync problem if not all devices know each other?

Hello there. Lets say I’ve shared a folder between 3 devices. 1 is a “server” and always on, but it’s only supposted to be used directly by clientA - so only clientA and server know each other in the remote devices list. Client B and Client A know each other. All systems share the same folders. So my understanding was if B modifies a file it gets synced to A (if it’s running) and A can sync it to the server. But this seems not to happen. I have plenty of files “out of sync” when I look on the “server” listing and these are all changes from Client B. So did I understand the setup wrong? I thought my client A can “relay” these changes to the server. I don’t want to setup all the other clients to have a connection with server. Please - I don’t want to discuss why this setup has been choosen - it is at it is :slight_smile: I just want to get syncthing running with a “man-in-the-middle”. Thanks for any hints.

This sounds fine and should work. The reasons it doesn’t will be something unrelated to the topology. See the reason for the sync failures in the GUI for why the files can’t sync.

thank you for your quick reply. Where can I find these? On Client A (the one known by server and vice versa) the folders are all up2date and in the remote device I see only the large number of “out of sync items” under the server entry. On the servers gui it says everything is up2date - the only errors I see there are related to inotify (which I can’t fix because it’s a shared host I don’t have the rights to change sysctl).

image

Out of sync items:

Expanding the folder in the GUI, on the receiving side.

there looks everything ok. this is the servers view

image

and thats clientA (the hub) image

may the problem be that i dont use reduce patterns on the (encrypted) server?

Ignore patterns can’t be applied on the encrypted side, because it doesn’t know the real filenames. Possibly this is the cause, if the other client doesn’t have the same ignore patterns.

the clients themselves use the same reducepattern. I always put a file into my share and include this, so I can propagate changes from a central place. Ok…then I just ignore the situation? I’m just afraid that if the hub dies, files on the server are not up2date and I lost all the changes from other machines.

I’d assume the problem with “Out of Sync Items” stuck in sync is probably https://github.com/syncthing/syncthing/issues/7474. Can you post screenshots of the Web GUI from all devices, so that it’s clearer what the whole situation looks like?

this could be. one of the clients went away (died physically) and I still see “pushes” on my hub - I don’t know how to get rid of them. Any Ideas? Should I just remove the device from my remote-devices list?

On a random Client B from my other shared folder I see this (looking all good to me): image

I also double checked that it IS using the same reduce patterns and a lot of the out of syncs are actually part of them! should I prefix them with (?d)? or should I temporarily remove them from the reduce patterns?

If you’re affected by that issue, there is no workaround, unfortunately. You either need to live with it or connect all devices with one another. The synchronisation itself should still work fine regardless of the “out of sync items”.