Hey @Frederic_MUNCH,
Thanks for the follow-up and for doing that testing, that was genuinely useful, and you’re right to push back on the NAS theory. Powering the NAS down, plugging in the USB SSD, and still reproducing the behaviour the moment Qobuz comes back online effectively rules out the drive spin-down explanation. Nicely isolated.
We’ve now gone back through the fresh diagnostics alongside the system logs, and here’s where things stand.
First, the good news on the hardware: your NUC itself is healthy. It has 32 GB of RAM (so this isn’t a memory issue), the CPU was sitting around 30 °C at boot with no thermal throttling, and the internal boot SSD is fine. The heat you were seeing really was just the NAS being kept busy, not the server struggling.
The key finding is that what looks like “continuous rescanning” is actually Roon Server being restarted on a roughly 15–25 minute cycle. Each time it restarts, it re-mounts storage and re-runs library identification from scratch, which is the rescanning you see and what keeps any connected drive active. So the restart is the cause, and the rescan is the symptom.
As for what’s driving the restarts, the logs line up with your Qobuz observation. After the library import, there’s a very large metadata-update backlog still working through the system, and on every startup Roon also runs a full Qobuz sync, plus ongoing album identification.
We can also see your Qobuz playlist requests coming back with an authorization error (HTTP 401), which is likely adding to the churn.
Disabling Qobuz removes that startup sync and the failed requests, which is why things settle down, so the mechanism is really the interaction between the freshly imported library and the Qobuz sync, rather than Qobuz playback itself.
Two things we’d like you to try, which will help us confirm the root cause:
-
Re-authorize Qobuz. Go to Settings → Services → Qobuz, sign out, and sign back in. That should clear the 401 authorization errors we’re seeing on the playlist requests.
-
With Qobuz active again, leave the server running undisturbed overnight (through roughly 1:00–5:00 AM your time). Roon does its heavier metadata housekeeping in that window, so if the import backlog clears overnight and the restart cycle stops the following day, that points us squarely at the post-import processing rather than anything ongoing with Qobuz.
If you can let it run through one full night and then let us know the next morning whether the rescanning has settled, that’ll tell us a lot. We’ll keep diagnostics enabled on our side so we can review the next restart cycle in detail.
Thanks again for your patience and for the thorough testing, it’s making this much easier to narrow down.
