Mass Files Out of Sync - Large Blocks Inconsistency Between Versions?

[Moisie]

Hello:

I’ve got a setup with a work Mac, two NASes and a laptop Mac in a cluster.

If I remember the setup details correctly, I initially set up the work Mac and one NAS on Syncthing, then added the NAS and the laptop. The data in question here was cloned from the work Mac to the laptop before the laptop was added to the cluster.

Now I’m seeing a large number of out of sync files on the laptop - but the files look present and correct.

Looking at the API for one of the files, I see this:

{
  "availability": [
    {
      "id": "GKRQ2C4-*******-*******-*******-*******-*******-*******-*******",
      "fromTemporary": false
    },
    {
      "id": "LE3IMUG-*******-*******-*******-*******-*******-*******-*******",
      "fromTemporary": false
    }
  ],
  "global": {
    "deleted": false,
    "ignored": false,
    "invalid": false,
    "localFlags": 0,
    "modified": "2011-11-15T18:39:36Z",
    "modifiedBy": "GKRQ2C4",
    "mustRescan": false,
    "name": "<file/path/here>",
    "noPermissions": false,
    "numBlocks": 15869,
    "permissions": "0775",
    "sequence": 35952,
    "size": 2079909726,
    "type": "FILE",
    "version": [
      "GKRQ2C4:1"
    ]
  },
  "local": {
    "deleted": false,
    "ignored": false,
    "invalid": false,
    "localFlags": 0,
    "modified": "2011-11-15T18:39:36Z",
    "modifiedBy": "RWCKRI7",
    "mustRescan": false,
    "name": "<file/path/here>",
    "noPermissions": true,
    "numBlocks": 1984,
    "sequence": 36668,
    "size": 2079909726,
    "type": "FILE",
    "version": [
      "RWCKRI7:1"
    ]
  }
}

I’m thinking that the work and laptop systems might have been setup either side of the move to large blocks - in which case, what’s the advised route to resolving this?

I’ve seen mention about nuking the database - that’s just deleting the DB folder whilst Syncthing isn’t running, isn’t it? And presumably any systems which are reporting the larger number of blocks from the API call?

Thanks!

[imsodin (Simon)]

The API output shows that it is are out of sync (note the different versions), even though the data is the same. When it does sync, it will notice that the file is actually the same and just update the version without transferring or writing to disk anything. So from the information given I don’t see anything wrong, i.e. from information about the actual problem is needed.

[Moisie]

Hi Simon:

Thanks for the feedback.

The problem is that the systems are ~1TB out of sync in this folder - and there’s only a few GB spare on the drive hosting it - so I’m trying to find a way for Syncthing to agree that it’s all the same without attempting to re-transfer it…

(And, for the record, the file MD5s the same on each system.)

[imsodin (Simon)]

As I said, if all those files look like this (or most), it won’t retransfer anything. “Sync” does not mean “transfer”, it means “make sure we have the same thing as everyone else”. So really the only thing you can do is stop doing whatever you do to prevent the devices from syncing.

[Moisie]

Thanks Simon - I’ll see what I can persuade of it and check on that!

[Moisie]

Ok - for other reasons, I removed the config for the folder in question on all devices in the cluster, and re-added it afresh. It’s now happily back in sync (which wouldn’t have happened already if there was 1TB data to transfer) - so all is good again. Thanks for your reassurance!