I have 2 “always on, receive-only sync nodes” servers (Device A, Device B). Recently they underwent some maintenance and the drives holding the ST DB have been replaced. I let it fully re-scan everything and rebuild the DB - it took about a week. After few more days, everything is seem to be settled.
“Device A” reports itself as fully “up to date” - all its folders and devices are synced and happy.
“Device B”, however, reports that “Device A” is outdated, with “1142 items out of sync”.
The weird part is that clicking on the “out of sync items” reveals an empty list, as seen in this screenshot:
The devices are in this state for a few days now, so it’s not like something is actively syncing there. Then 1.18.2 upgrade was applied, with a full index exchange enabled on Device A. Nothing changed.
Regarding ignores - the .stignore files haven’t been modified since before this story, so I don’t think that there is any problem there (and when I had asymetric ignores issues in the distant past, I had the out of date items actually listed, and not “empty” as I do have now.)
The empty list make is seem like some kind of a bug.
Do you have any suggestions on how to go about debugging it? Any logs/debug facilities that I can enable? (the standard log output doesn’t reveal anything out of the ordinary)
No, it’s definitely not a UI issue, when I actually add new files to one of the devices, and the sync is actually in progress - I do can see the list of out of sync items properly.
I noticed though that some folders report different “global state” on different devices:
Device A (receive-only node), Folder X: Global State: 64 files, 2 dirs
Device B (receive-only node), Folder X: Global State: 130 files, 2 dirs
Device C (send-receive PC), Folder X: Global State: 64 files, 2 dirs
Device D (send-receive PC), Folder X: Global State: 64 files, 2 dirs
So, device B (the one that thinks that A is out of sync) thinks that the global state of a folder is larger that the actual global state.
Edit: note - Device B doesn’t have “revert local changes” button either - it thinks that all it’s local files are correct.
If you can reliably reproduce the issue, then the standard procedure would be to enable db and model debug logging first, then record the logs and upload them here.
Personally, I would also pause all unrelated devices and folders in advance just to reduce the noise and make debugging easier for the maintainers.
