In my network infrastructure, my Unifi XG-8 is the router/firewall but the way it works is that each of the 8 10G ports are separate LANs. I have a single LAN setup going to a distribution switch - a Unifi XG-16 - that everything hangs off of either directly or indirectly (through other switches). If I hook the Nucleus+ to the router I’d have to create another LAN and then bridge the 2 LANs and I think that adds more variables to the mix.
I think the best next step would be to hook the Nucleus+ to the XG-16 and see what happens. Do you agree? I can try this tomorrow.
Yes, I agree that this is probably the best next step here given the complexity of the network. This will definitely bypass some of the other variables and it’ll be good to see how things perform in this configuration. Please let us know how this goes.
Thanks for your cooperation and patience here, Craig — It’s much appreciated!
Got around to getting my Nucleus+ out of the rack and connecting it to my Unifi XG-16, my main distribution switch which has a single connection to my XG-8 router/firewall and there was no change. Roon Server continues to reboot 1-2 times per minute. For good measure also connected it up to the switch my Mac Pro is connected to that has my second copy of Roon Core running properly with the same configuration as the Nucleus+ and it also fails there the same way. Don’t think it’s the network…
Given this, I wonder if I should reinstall the OS, clear out the DB and reinstall from backup and see what that does. What is the next step you guys think I should do?
Well, my Mac Pro’s Disk that had all my media on it died today. I have it backed up so it isn’t a problem but until I get it replaced my backup roon core is now dead too.
I need to get my nucleus+ running. Where are we going now? Need some direction on where we’re going to get this fixed as I’m at almost 2 weeks dead in the water on it…
I have a meeting scheduled with the technical team tomorrow to discuss our next steps here, but as you no longer have a backup Core to use, I’m hoping that we can try using a fresh database on the Nucleus+ in the meantime so we can see if that yields any changes. Please see the steps to do so below:
Ok, did that and Roon is up and running and Roon Server is not crashing, but of course there is no data. Assume I restore the old database? Or do a brand no loading of all my music?
Let’s try restoring the database and seeing how that goes as a start. Since things are stable without media, it could be that there is a media-related issue here. If things return to the old behavior with the backup, then you can go ahead and try this again but importing the music newly for the time being and see if that has any change.
Can you confirm how many tracks total you have in your collection?
Ok. It was about 60-65K tracks in my own library and a few thousand from tidal and qobuz. I hope I don’t have to go through a reload and re-analyze again. And I will note that my Mac Pro Roon Core was running fine off a restore, but we’ll see what happens…
Well, restore complete and it is up and running and seemingly stable. I still have to move it back into the rack and we’ll see if that changes anything but so far so good. Assuming no issues, then it appears the roon server dB got corrupted? Have you seen this before? And I assume I can delete RoonServer_old?
Ha. Spoke to soon. Just crashed and is back in a crash loop. Never got it back in the rack. Fail on this approach I guess.
There is one difference between my nucleus+ and the Mac Pro versions. My music is on a ssd inside the nucleus+ but on a attached hard disk for my Mac Pro. The music files should be identical though as my Mac Pro disk is my master and then they are synced to the nucleus.
Should I do a reload from ssd or go so far as to wipe the ssd, recopy the files to it and then reload into roon? Or do a wipe and recopy and then a dB restore (which would work if for some reason the files on the ssd we corrupted…? I suppose the nucleus+ itself could also be borked in some fashion also… which direction do you suggest I head?
Thanks for giving this a try as well. So, the same backup that worked for the Mac failed for the Nucleus+, and the media on the Nucleus is exactly the same as the Mac, correct?
This is definitely some good data that I’ll discuss with the technical team tomorrow so we can determine some further next steps.
In the meantime, I’d suggest starting fresh once more on the Nucleus and re-importing the media from the SSD. Let us know how this goes and I’ll follow up as soon as I’ve discussed with the team.
It was exactly the same backup and as best I know it was the same music files even though stored in two different locations (internal ssd for the nucleus and attached ssd for the Mac Pro) as the locations are synced.
As I mentioned, I’ll follow up ASAP once I’ve met with the team. Thanks again for you patience and cooperation during the troubleshooting here, it’s very much appreciated.
The next step is going to be a long process so I’m going to start completely fresh —wipe the ssd and reinitialize, blow away the roon server dB, recopy the music files and reload into roon. If that doesn’t work it’s the nucleus+.
Is there any chance that you have a Windows machine that you could temporarily use as a test here? If you restore the backup there, and are able to temporarily connect the USB drive to it containing the media from the Mac, does the same crash occur?
And in any case it was a thunderbolt 3 disk that carried my music files on my mac pro not a usb drive and in fact that is the one that died today and took my mac pro down… so out of luck on all fronts.
I spoke with the team about this today and they’ve looked into things once more. Upon further investigation, along with the details that we’ve learned from the troubleshooting steps you’ve taken, the team believes that there may be some media here that is causing Roon to crash as it is extracting data.
Mac, Linux, and Windows all use different drivers when extracting data from media, and so things working on Mac but not Linux makes sense in this case — It’s possible that the issue only affects Linux drivers.
Moving forward, we are hoping that you can remove and media that was added in the last month from your watched folder on the Nucleus. Please move it somewhere that is not being watched by Roon on the Nucleus. Once you’ve done so, can you confirm if this behavior persists?
I don’t think I can do this. As I said yesterday I was going to start from scratch. So I’ve reinstalled the Nucleus+ OS, reformatted the internal SSD and are about 80% done transferring 75K tracks from my backup NAS.
As I also said, the external disk on my Mac Pro which was running as my backup Roon Core crashed. That was the disk that held the media for that Roon Core. I will have to restore the 75K tracks from my NAS to that system next but then it will be in the same state as the Nucleus+.
Given all this, will there be any way to determine what was added in the last month???
Normally you could look at the file creation dates, but if the media that you have now won’t have dates that match this, we can try splitting your library up into chunks to try to track down where it’s occurring.
If you split your library into two watched folders on the Nucleus, if you enable one watched folder at a time, does one half cause the crash while the other works?