Restarting after a system rebuild

I have two large desktop machines and a possibly laptop that I want to keep in sync w/ each other. I had this working for a while with all the machines on Debian Buster… I’ve recently been upgrading the desktops by replacing the spinning sda w/ a 1TB SSD, and updating the OS to Debian Bullseye…

On one of the desktops this has not been a smooth process in any way, and I eventually ended up doing a fresh install of the O/S and restoring all of the non-hidden files in ~/home from backup…

This wiped out the ST install on that machine. I have reinstalled ST and gotten it to the latest version, but haven’t tried reconnecting the machines… What is the best / safest way to do this without risking data loss?

What I’ve found in the docs seems mostly aimed at doing a new setup from scratch, not in fixing a broken one.

Thanks ex-Gooserider

Since you haven’t copied any of Syncthing’s configuration or identity, it’s basically like adding a new device into the cluster from scratch. Check that you have a different device ID in the Web GUI of the freshly installed device, not matching the one listed for the old device in the others’ GUIs.

Then you should be safe just adding the new device ID on the other machines and sharing the folder(s) with it.

It may get tricky if you have the data restored from backup already in place before connecting the shared folder. In that case it’s safer to configure the other devices (presumably with the latest versions of all files) with send-only folder type temporarily. In case the reinstalled device tries to push out an older version from its backup to the others, you can then see the difference and override the changes on the other devices.

Having a proper, up-to-date backup is always best when going through such a procedure. The easy way is starting with an empty folder and letting the data sync back from the other devices.

1 Like

Well I have started trying to do this, and have set all the folders on ‘Bill-box’ (the system that seems OK so far) to send only, and telling ‘Fancy-box’ to sync with them…

However I now have a bunch of messages on Bill-box like:

Blockquote

“fancy-box wants to share folder “Downloads” (d5fai-t5c7q). Add new folder?”

(The other messages have the same text, just different directories and ID’s)

If I try to add it, I get a warning:

BlockquotePath to the folder on the local computer. Will be created if it does not exist. The tilde character (~) can be used as a shortcut for /home/atorrey. Warning, this path is a subdirectory of an existing folder “Downloads” (ey7ee-wksuk).

The Folder ID says:

Blockquote Folder ID d5fai-t5c7q

BlockquoteRequired identifier for the folder. Must be the same on all cluster devices. When adding a new folder, keep in mind that the Folder ID is used to tie folders together between devices. They are case sensitive and must match exactly between all devices.

But I can’t edit the folder id in that box to match the one in the warning?

What should I do?

Looking back at what I restored from backup on Fancy-Box, I didn’t restore any of the .hidden files and directories in my home directory, but probably did restore them in any of the subdirectories, like ~/Downloads, which would have included the .stfolder files from the old ST install that got wiped out when I did the reinstall. Could these be the problem? If so, how do I fix?

ex-Gooserider

If I understand correctly, you reinstalled on fancy-box and added the folders there again, then selected sharing with bill-box? And on bill-box they were already configured because that one you didn’t reinstall?

You probably should go the other way. Remove the folder on fancy-box again. Only from Syncthing, not the actual data on disk. Then select sharing to fancy-box for the existing folder in the bill-box Syncthing GUI. This will get you a prompt to accept it on fancy-box, with the correct folder ID.

Thanks, it looks like that did the trick. I removed the fancy-box folders and told bill box to share them to fancy box as you suggested. I then got the message to accept on Fancy-box, which I did.

Fancy-box eventually showed as synced, but most of the folders on bill-box (which I had set to send-only) showed as out of sync.

I am now slowly switching the folders on Bill-box back to send & receive, which appears to be getting everything back into sync.

I did notice the upload rate from Bill-box is very slow, a few hundred B/s - is this normal? I have left both the upload and download rate limits in the settings at the default of 0 - assuming this meant no limit. Should I change this?

(I know it takes me a few days to respond, a lot of this is because the machines are in different meatspace locations, and I don’t often use both in the same day…)

ex-Gooserider

You don’t need to change the limits. Slow transfer speeds can have a number of reasons. Some possible measures are mentioned in the FAQ section of the documentation. It might also be limited not by the transfer, but because Syncthing is busy with scanning the files if it’s a large dataset.

I’m noticing a lot of out of sync items, currently over 8,000 files / 266GB, although the number is slowly decreasing. Spot checking it appears that they are scattered across a great many of my directories. Mostly they are old files that should have all been in sync before I started the rebuild process, so presumably should still be unchanged on both machines…

Is this normal? Is there anything I should do about it to either speed the process or tell ST that everything is OK? Or should I just let it sit and slowly deal with it at it’s own pace?

The other thing I’ve noticed that seems a little strange is that if I have the screen showing the info for both boxes, both show the same rates for upload and download speeds - which does not make sense to me, shouldn’t the numbers be ‘flipped’ so that the upload speed on one machine is the same as the download speed on the other?

This probably doesn’t matter from a functional standpoint, but it seems odd, or is it possibly a minor bug?

ex-Gooserider

To me it sounds like the two machines are busy with scanning and hashing the (rather large) dataset. During that process, each sends its result (index data) to the other, which results in both doing uploads to the other. The numbers are not very accurate unless you have a continuous data transfer. Since the actual data is already present on both sides, no content should be transferred, so you don’t see traffic from that.

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