Two issues: first, since the last update, dsf files can only play after converting to PCM (I was using DoP). I have a Bryston BDA-3 DAC and found that playing native dsf files was cumbersome. Second, and possibly related, searches still take a long time; 're-booting the iOS app sometimes helps, but not always. I have a Nucleus (not plus) with 69k tracks, so I recognize I might need new hardware -- could this be the issue?
Additional information: DSP minimal (auto level, upsampling) with 2 grouped endpoints. When playing music, and concurrently doing a search, I occasionally get ‘waiting for Roon server’ errors, followed by minutes-long wait until I can re-establish a connection. I use iOS for the most part to control Roon. Closing the app in the ‘waiting’ mode does not recover more quickly – I still need to wait.
Thanks for taking the time to write in, and I’m sorry to hear you’re running into issues on your Nucleus!
I was able to review a fresh server diagnostic and see repeated traces of database corruption throughout the report. As a next step, I’d like to test to see how things perform on a completely fresh database.
I’ll post the steps below:
Create a Backup of your current Roon Database (Or, ideally, you have a prior backup saved that predates the issues you’ve been experiencing)
Rename the “RoonServer” folder to “RoonServer_old”
Restart the RoonServer in the WebUI to generate a new Database
On the Roon Remotes, press “Use another Roon Server” and connect to the new database
Do not restore from a recent back up in this case - and only connect your Qobuz account to this fresh database. Let me know how things perform for you.
Thanks for the help. I have a few questions, though, before I commit to a new db.
-Will this eliminate current playlists and settings for recordings?
-If so, is there an intermediate step possible, such as re-installing the server software?
-I assume this would not fix the dsf playback issue – any thoughts on that?
-And most important: could the diagnostics suggest the cause of the db corruption? Without fixing that, might not the db simply get corrupted again and I’m back where I started?
Since the last update (1368) things seem to be running smoother. I noticed that I lost the connection to the Roon Core/db in the AM of Feb 22. A restoration of a backup was suggested on the iOS app. First, I re-booted my router, then re-booted the Nucleus, and after a few minutes all came back, without restoring from a back up. So I wonder if there is something in the internet connectivity or the links (through 2 switches from router) to the Nucleus that might either be the cause of the corruption. (Still doesn’t address the dsf playback issue, tho.)
Reinstalling the software likely won’t help in this case, as the corruption lies within the database itself. It will eliminate everything, as if you were starting Roon from scratch.
From your Nucleus logs, I see corruption-based error traces followed by Too many dropouts (>3s dropped out in the last 30s). Killing stream traces, leaving me to believe there could be some correlation. However - you can always make a fresh backup here, so if there’s any concern with testing things on a fresh database, just know you can restore from your backup. That said, I suspect you’d soon run into additional and equally odd issues further down the road running the same database. Similar to -
–
The diagnostic report unfortunately didn’t provide specific information around the diagnosis of the corruption. I know users have used third-party applications like dbpoweramp to analyze their local library to scan for clumps of corrupted files.
Another option you have is to try to restore from your oldest backup - or one that predates the issues you’re seeing. We could analyze that restored backup to see if any traces of database corruption occur.
This AM (I submitted a new support request but it hasn’t posted yet), after an attempt to update to 1388, I got a message that the library could not be loaded and the app suggested a restore from a backup. I tried multiple times, and each time (with different backups) the library would load, and then, just as all tracks were loaded, I’d get the same message – cannot load library.
Next steps were as you outlined in the earlier thread: stop server, re-name Roon and RoonServer, re-start Server software. Each time, I do NOT get a prompt to choose another server, but instead I get the original server, playlists, etc. It’s as if the old db self-loads. I tried re-starting a new server 3 times now – same result. And each time, after a while, I get a prompt to restore from a backup. I cannot understand why a new server hasn’t been created. Any further thoughts?
Bob
PS I just noticed: with the last creation of a new server, only the directory ‘Roon’ was re-created, not ‘RoonServer.’ Before, both directories were present, and re-created. Could this have been why db corruption was taking place? And is it correct to have only ‘Roon’ and not ‘RoonServer’ directories?
This unfortunately points at existing corruption within your database. From our last conversation, did you test out your oldest backup?
With a more recent diagnostic review, we are also now seeing hardware failure with the m.2 SSD. This is also likely causing you the issues you describe above.
You have two options for replacing the SSD : you can replace yourself manually, or you can send it into Roon for repair in house. Whats the purchase date of the device?
Hi Benjamin,
Thanks for the speedy reply. The Nucleus was purchased in 2020, so I’m a bit surprised that the m.2 ssd is bad, but it certainly explains the behavior. The oldest backup I tried was from December.
Oddly, after my 4th attempt at starting a new db, and being again prompted to restore, I used the newest backup and it’s been running for 10 hours.
Do you have a brand and capacity suggestion for the m.2 drive — is it the same as in the knowledge base? I can probably install myself, as I installed the 8 TB Samsung ssd for my library. I will need instructions on the proper procedure to follow.
It’s odd indeed and not something we see often.
Note that you’ll need a USB drive for this process that you can connect to a computer, and then the Nucleus.
Here is a link to one SSD option we recommend to many users:
After that, you’ll need to format the SSD. We have a good help article you can find here for next steps. Start from step #3 :
Essentially, you’ll be flashing the ROCK OS to a USB drive, and using it to boot and format the new SSD in your Nucleus.
Once you’ve successfully flashed ROCK onto the new SSD, keep your Nucleus online and let me know. From there, we can update you to the Nucleus OS on our end.
I’ll be on standby for any questions you have in the meantime!
I’ve flashed to a USB drive. But:
I don’t have a separate monitor or keyboard for the Nucleus to attach to. I was going to use HDMI to a Macbook – will this work the same way? The instructions beyond step 4 seemed to be for a NUC (hence, PC-based). And when the instruction says to unplug any USB drives, I am assuming that I would NOT unplug the SSD that holds my library (internal to the Nucleus) – or should I? Finally, as this is a Nucleus, not a NUC (per se), do I still need to install codecs?
Thanks (again) to you both. I found a way to use an iPad as a monitor (HDMI Orion app, and a video dongle with UBS-C connector). I await delivery of the m.2 drive and the dongle – likely install after the weekend.
I have installed the new m.2 drive (256 GB, Transcend) and have booted with the flash drive installed. I get a message to pick the boot drive (the OS identifies the USB drive as a boot drive), and when I hit Enter, I get a message to reboot and select the boot drive or insert boot media and hit a key. If I hit a key, I get the same message; if I re-boot with the F10 key I again get the ‘select boot drive’ screen; if I boot without the F10 key, I get the same message to re-boot and pick the boot drive. I even tried a second flash drive with Etcher (which verified the flash for each drive tested).
I am clearly not getting the instructions to install the OS. Any advice appreciated.
Perhaps the USB drive needs to be formatted differently for the bootable file to be read? Format as Mac OS, FAT,or MS-DOS? And if Mac, does it need to be a Master Boot Record, or some other format? I have erased the USB drive and formatted as Mac – still will not boot from this, even though Nucleus sees the USB drive and I can identify it (on a Mac USB keyboard) as the boot drive. From there I go into the endless loop of being instructed to re-boot and insert boot media…
This may be an irrelevant question, but you sent me to the instructions for a boot drive for a NUC, and I have a nucleus. Is it a different boot file for a Nucleus?
Thanks. I have already tried EXFAT, so unless Etcher is malfunctioning (and it verifies the flash), this still fails. I have tried both USB slots on the Nucleus also. When the device boots, I get a list of bootable drives: it shows the (new) Transcend drive, which of course is not bootable, and the USB drive – no numbers. I use the up/down arrows, as suggested on the screen to highlight the USB drive, and then get – no matter what combination of formats on the USB stick – the Re-boot and choose bootable media prompt.
