Out Of Sync, but Who?

spiritfly · October 6, 2016, 9:40pm

Hi syncthingers!

I currently have setup a master server with synthing and the files are being synced to like 10 other syncthing machines. Every once in a while I get a “Out of Sync” message on the master server. It’s not very common, but it’s happening. First time it started happening I sorted it out by checking each of the client machines and found one that had some other software running that was writing files in the same directory which was synced. And that’s how I sorted it out.

But now I get this every once in a while and I have no clue what is causing it. So my questions is: is there any way to see what connected syncthing machine is causing the “out of sync” ? This should be doable because syncthing should know which of the connected computers first changed the files. This will be much easier to sort the issues with out of sync character.

AudriusButkevicius · October 6, 2016, 11:09pm

At the point other non-masters accept the new file, there is no way to know who caused the change. Best I can suggest is to enable -audit -verbose options and check the logs on each of the 10 devices.

kluppy · October 7, 2016, 11:31am

If you are in control of all of the machines I suggest you make the folder read only except for a dedicated Syncthing user.

I understand this doesn’t work in all situations but syncthing relies on the user securing the data and if there is no good reason for it to change except on the master make it so your clients can’t change it.

nrm21 · October 22, 2016, 5:12am

Actually you can query the API now for disk events via CURL or somthing. This should only show local changes picked up not files/folders written because they came over the network.

github.com

syncthing/docs/blob/master/rest/events-disk-get.rst

GET /rest/events/disk
=====================

Returns local disk events that occur when the scanner detects local file system
changes (:ref:`local-change-detected`) or when files are pulled from a remote
device (:ref:`remote-change-detected`).

Optional GET parameters:
 - since (events starting after the given ID)
 - timeout (fail after given seconds if no event is available, 2s by default)
 - limit (return last x number of events)

.. code-block:: bash

    $ curl -s http://localhost:8384/rest/events/disk?limit=4 | json
   {
    "id": 4,
    "globalID": 45,
    "time": "2016-09-26T22:06:10.4734536-04:00",
    "type": "LocalChangeDetected",

This file has been truncated. show original

Of course you will still need to go to each of your machines to run the query, unless you want to expose the API across the network on all of your devices (it might be worth it if they are behind a secure network). And you’ll want to give them all the same API key as well so you don’t drive yourself crazy.