Syncthing downloaded only half of files in a folder and claims it's up to date

Syncthing 1.18.6-2 on Arch linux

I’ve shared a 400k file folder from a VPS via syncthing with my fresh local machine. The client downloaded only ~200k files and claims it’s “up to date”

On the server:


On the client:


It’s the second time Syncthing has done this with that folder. How can I force the client to download all files?

Are both sides running the same versions of Arch Linux and Syncthing? If possible, please post screenshots showing the full Web GUI on both sides. You can blur folder/file names and such, but please leave the rest visible, including specific Syncthing version information.

Also, if this is your first time trying Syncthing out, I’d suggest to experiment with test or at least fully backed up data first, just to be safe.

1 Like

I guess I’d very that web ui agrees with what the GTK ui is showing, so yes, screenshots of the web ui would be nice.

The server is running Debian (Yunohost) and Syncthing v1.18.5, and the client is running Arch linux and Syncthing v1.18.6

Here’s the server:

And the client:

It managed to sync some more files since my original post, but still nothing. I’ve left the PC turned on for the whole night in hopes something would “click”

Ignore patterns are not the same on both side. Please do check Carbon.

Carbon has empty ignore patterns list. If I understand correctly, that should make sense if carbon had more files in its state than the server, but it’s the other way around :thinking: Or am I getting this wrong?

Nyuno is aware of more files than Carbon is. Although, both try to send files to the other. Furthermore all other devices in the cluster are disconnected. Maybe both active devices are waiting for files from disconnected ones. Maybe Carbon was down when a now-disconnected device updated the global status. Just a idea, I don’t think anything is broken. Just click the blue links in “Other devices” list on both Nyuno & Carbon to see which files are pending upload and look in respective File Systems if these files are really there.

Also try to make other devices in the cluster active again.

And again, can you have the ignore patterns lists to be the same on each side ?

Are all device in the cluster linked with all other devices?

Is Carbon intended to be something like a server/central node?

Ignore patterns work locally only, i.e. the device that hosts this list will ignore files FROM other devices AND ignore its own FILES (from others and own that match patterns). Local file number says how many files our device will try to send to others (success depends on remote ignore patterns list). Each device tells all others how many file she tries to send. The bigger is the winner for the global number which is displayed on all devices… this may depend if all devices are linked to all others (not sure of that, i.e I don’t know what happens when you have some parts of the cluster topology that is like a daisy-chain, i.e. some devices don’t know some others, don’t have direct link to all others or are in a cul-de-sac(dead end). When the last-before-others has custom filters, then possibly the deadend may only see the local number of the last-befo… and compare this one to its own to decide the global size, then this one maybe lower than in all other ones in the cluster - breathe :wink: ).

Oooh, that makes sense. There are some machines that probably had different ignore patterns and that are now offline. That would explain it. The UI led me to believe something else was the issue. Thank you for the explanation! <3

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