Discover does not work, some edits do not get saved

Core Machine (Operating system/System info/Roon build number)

Intel NUC (i7, with SSDs for OS and music), OS: ROCK, Version 1.0 (build 174) stable, Server Version 1.6 (build 401) stable

Network Details (Including networking gear model/manufacturer and if on WiFi/Ethernet)

Fritzbox Router (4 LAN-Outputs), ROCK and Lumin D2 directly via Ethernet, remotes via WiFi

Audio Devices (Specify what device you’re using and its connection type - USB/HDMI/etc.)

Several: Lumin D2, ca. 3 Squeezeboxes etc. Audio works fine on all of them

Description Of Issue
The problems began yesterday, when I added an album by a relatively unknown musician that Roon did not recognize. I was unable to have it identified: The process of identifying just went on and on for minutes but did not produce any results. I eventually canceled it.
Afterwards a few more problems came up: Connections between artists and their albums were missing in some places, edits were not saved and always reverted to the previous version and, most notably discover stopped working completely.
What I mean is this: On the main page where I have the choice between Discover, Recent, Tags, Genre, etc. it is impossible to select Discover. Actually I can select it, but for an indefinite time (I tried up to more than 10 minutes) the Roon “patience symbol” keeps animating, but no content shows up. This happens on all remotes (Windows 10 PC, Android phones with recent OS, and older Android Tablets).
Recent, Tags, and the rest work normally.
I tried restarting and reinstalling the server software which brought no changes.
Any ideas what I could try now?
My main concern would be not to risk my database since it contains a huge amount of effort …

Go into the router setting and find “Software Packet Acceleration”. Disable it, and save this setting.

Shutdown the Roon Core, power off everything including NUC, audio devices and remotes. Power cycle the router, wait for 5 minutes, then restart everything else.

Thanks, Peter!
I just tried that (had to switch off hardware packet acceleration as well, though), but nothing seems to have changed: Discover still does not find anything.

Try removing the problematic album temporarily, then invoke “Clean Up Library” function - whether this will risk your effort, I don’t know. Or you can wait for advices from support.

I tried that as a first attempt to solving the problem - just forgot to mention it in my initial post.
Guess I will have to wait for support, then …
BTW: “Clean up library” found 22 files, but cleaning did not actually remove them. When I open the cleanup dialog again, these 22 files are still there.

Hi @Elmar_Heimes,

Just to verify, is this new edits that you were making or were prior edits being reverted too?

I only noticed it on new edits - the older ones seem unchanged, but I cannot say for sure 100 percent.

Thanks, @Elmar_Heimes.

I think the next step here is to enable some diagnostics on your account so our technical staff can get some more insight into what’s going on here.

However, before I enable this feature, I’d like to ask for your help ensuring we gather the right information.

First, can you please reproduce the issue once more (try to open Discover) and note the time at which the you do so Let this try to load for a minute or so. Then respond here with that time, and I’ll make sure we review the diagnostics related to that timestamp.

I opened Discover on my laptop just a few seconds ago.

And I am switchig to Recent now - at 10:22 pm Central European Time (Berlin Zone)

Sorry: Switching to Overview, not Recent.
To make communication easier I just switched my language to English.

Thanks, @Elmar_Heimes!

Now that I have the timestamps, diagnostics have been enabled on your account. The next time your Core is active a diagnostics report will automatically be generated and uploaded directly to our servers

Once that’s been received, I’ll be sure to update this thread and pass the diagnostics over to the team for further analysis.

Thanks @dylan !
Is there anything I can do to make it easier for your team?
I cannot stay in front of the computer for very much longer now (have to get up early tomorrow), but I can certainly keep ROCK running and have some client machine doing whatever makes sense to you.

Addendum - ca. 10 minutes later: Since your team probably has been listening to my core, I have performed a few of the actions that do not work:

  • Tried to add artist picture (Torsten Zwingenberger) from URL - did not work
  • Tried to add artist picture ((Axel Zwingenberger) from file - did not work
  • Edited the artist “Axel” from a Jools Holland CD to “Axel Zwingenberger” - seemed to work, but this specific edit has reverted itself several times before
  • Keep the page “Discover” open on my Windows 10 Laptop client
  • I let a long playlist simply run through the night. This has not been the source of problems, though. Audio works just fine on all clients.

I will now leave everything in that state and check back tomorrow.

Update - 10 hours later: Discover still showed just an empty page.
I have to stop Roon on that client now, because I need my laptop for work.
If needed, though, I could let some other client run …
Just send me a note.

I just kept on using Roon as if nothing was wrong and have just now discovered another problem:
I have added tow albums to the watched folder and they do not get processed - it is as if I had not changed anything.
They also do not show up when I search for the artist, Roon just gives me the album that was there before.

Hi @Elmar_Heimes,

Thank you again for your patience while our technical team has been looking into this issue for you.

I spoke with our team today regarding their evaluation of the logs from your install. Sometimes, issues like the one you saw can be spurious, but unfortunately in this case they are seeing signs of low level corruption in your database.

This type of error is extremely rare for us and not one that we see often. Our database infrastructure is designed specifically to prevent this type of corruption, and we don’t take this class of issue lightly.

We’ve traced a few reports like this in the past to hard drive integrity issues but generally speaking, this means that Roon is reading information from your hard drive that is different from what was originally written, and the database is now unable to load properly.

What’s Next?

If you’ve been making regular backups, my advice would be to install Roon fresh on your Core machine, and roll back to one of your backups. If the database loads properly, your edits, playlists, tags, etc should be intact, and we can confirm everything is performing properly once it’s been restored.

If you do not have any backups, unfortunately you will need to start with a fresh database.

Again, this class of issue is extremely rare for us and you have our apologies for the trouble here.

Thanks Dylan.
BUT: I have seen this text about the corrupted database before in my thread from about half a year ago.

So this means, that this “very rare occurrence” happened to me twice in a rather short period of time. OK, the first time I was rather adventurous concerning changes to my files and I can see a correlation to the corruption (even though I would expect Roon to handle that gracefully), but this time I have not done anything apart from normal edits within Roon.

Could there be something wrong in my setup which causes problems for the database? What can be done to prevent that?

I really like Roon (as a lifetime subscriber), but if the database remains so fragile it becomes almost unusable for me - I need a system I can trust and that does not try to delete my edits (and there were many!!) on a regular basis. This problem now makes me think hard about returning to my LMS-based setup again. It lacked a lot of features, but it never caused any major problems for me …

OK, for now I will try to salvage one of my backups. What procedure would you recommend for that?
Reset the database from ROCK’s website and reinstall the OS from the same website? Or is that not enough and I should rather format the SSD and start from a fresh USB-Installer?

I suggest running this for no less than 24 hours:

You need to boot the NUC from a USB flash drive containing this program written as a bootable image as instructed.

Hi @Elmar_Heimes,

Is the Core machine that you’re using the same as the previous Core machine? This is definitely a rare occurrence, doubly so to happen twice that close together.

This type of issue occurs when because we’ve written something to the drive that, when we try to read it back, is different. In the past we’ve traced issues like this to faulting drives or bad RAM on the Core machine. Since this issue has occurred for you multiple times, it would definitely be worth looking into the possibility of faulting hardware.

The link that Peter shared above is one such way to test for this.

Thanks for your patience and understanding here, Elmar, and apologies for the troubles.

Yes, the Core machine is the same as before. I will try running the test that Peter recommended (thanks a lot!) and see what it brings.

It may be a while, though, before I can do that since I will be going on vacation tomorrow night and there will probably not be enough time for checking this before I leave.

Anyway: Thanks for your help @dylan and @wklie !
I will come back to this thread after my return.