Hey @Evan_Adams,
Thanks for the Aug 10/11 timestamps, and for your patience while we chased the network angle. I’ve been through your Roon Server logs and I want to redirect us, because the logs don’t support a connectivity cause: your server hasn’t restarted since Aug 9, and ARC’s connectivity checks have been returning success throughout, including across the whole window you flagged. Please leave your setup as it is now, behind the router. Nothing more to change there.
What the logs do show is on the library side, and I think it explains the pattern you’ve been describing since post 1.
Every album on your Aug 11 list is a hybrid. GUTS, Dopamine, Future Nostalgia and Access All Areas are TIDAL albums grouped with local files, and Access All Areas is assembled from tracks belonging to two different TIDAL releases alongside your local copies. Each time ARC builds a page for one of these, the server logs a warning that it’s encountered a mixed local/TIDAL source. That warning appears over 2,000 times in your logs, and it only appears on albums shaped this way.
More significantly: there are 199 albums sitting in your server’s identification retry queue that never actually get retried. Every check, every night, for the full twelve days of logs, the queue reports itself as not ready, and the count only climbs. The albums you’ve edited are among them. So when your library refreshes overnight, those albums can’t re-confirm which version they are, and they settle on whichever one is available. Re-scanning in the desktop app forces identification through, which is exactly why that keeps fixing it temporarily.
I can see the switch itself, too. Your album records get a new internal ID each time you edit one, and I can watch a single ARC album ID resolve to one record and then to a different record minutes later, same album, two different underlying versions.
One correction worth noting: there’s no library activity at all in the 10pm–11am window you gave me. The nearest event was about twenty hours earlier. That tells me the reverts are happening earlier than you’re catching them, which is why I’d like to tighten the timing.
Three things that would help a lot:
- Pick one album and check it on a schedule. Access All Areas is the best candidate. Fix it, then check it in ARC the next three mornings. Your server’s overnight maintenance runs 1am–5am, so checking right after it closes should pin the change to that window.
- On that same album, open Album Editor → Identify Album in the desktop app and screenshot what Roon offers you. With 199 albums stuck unidentified, I’d like to see what identification actually returns for one of them.
- Try ungrouping one affected album, separate the local version from the TIDAL version so it’s single-source, and leave it three nights. If that one holds while the others keep reverting, we’ve confirmed the trigger and you’ll have a workaround you can apply to the rest immediately.
And when you next catch a revert: force-quit ARC and reopen before doing anything else. If the album is still wrong, the server changed it. If it corrects itself, ARC was serving a cached page. That distinction narrows things considerably.
