OK.. let me know what you find
Mel
Hey @Mel_Martin,
Thanks for your patience, the extra logging we enabled did its job. Six days of logs gave us a much clearer picture, and there’s real signal in here.
Memory on your Nucleus+ does climb steadily each day and doesn’t get released until a restart. Importantly, it’s not climbing to a dangerous level, you peaked around 3 GB of the 8 GB installed, so this isn’t putting your server at risk of running out of memory. But the growth pattern is distinctive enough that we’ve been able to trace it to a specific recurring internal task, and that’s now with our development team.
More interesting for you: we also found a recurring internal stall that lands roughly every two hours, and it gets longer the longer the server has been running, about 3.5 seconds after a few hours of uptime, growing to 7 seconds after fourteen or more. That lines up closely with what you described: fine after a reboot, worse as the day goes on. This is a much better explanation for what you were hearing than memory pressure, and it’s been logged as a separate issue.
No dropouts actually occurred during this capture, and nothing crashed. Every low-buffer moment in the logs is normal behavior at the boundary between tracks. So playback was clean during this window.
One thing worth watching: Your Eversolo DMP-A6 is the only endpoint reporting clock-sync trouble, 160 warnings over six days, plus a few connection refusals. Your Naim Uniti Atom, in the same group, logged one. If dropouts return on the grouped zone but not on the Atom alone, the Eversolo is where I’d look first, independent of the memory work above.
A favor to ask: If you can, avoid rebooting between now and Wednesday afternoon. There’s a metadata refresh cycle that only comes due every seven days, and the next one falls on September 2nd. Capturing that with the current logging in place would tell us whether that refresh releases the accumulated memory or adds to it, that’s the single most useful data point we’re still missing. If you need to reboot for any reason, no problem at all, just let me know so we know the window closed.
Thank you!
Couple of useful nuggets there and glad it got reported and investigated. A rising tide lifts all boats.
Sure.. I’ll stop rebooting… were any others reporting these issues? I can substitute another device for the Eversolo when this other stuff gets ironed out. We can see if it makes a difference. After my morning reboot tho, the system runs fine into the late night.. dont hear any glitches on the eversolo.. it is up to date on firmware. Thanks for all the effort.
Did not reboot server as directed and had a brief hang at 7:59 Mountain Time. If these Eversolos are having problems, and since you approved them for Roon, have you had any discussions with Eversolo? If you like, I have a Hifi Rose RS250 which is Roon ready I can swap with the Eversolo.
Did not reboot server again. Again had a 30 second stall loading an album at 9:23 AM MST. When I reboot, it works. When I don’t, it stalls building a playlist. Once it stalls, it usually works for the rest of the day. Very consistent bad behavior.
Hi @Mel_Martin,
To be clear, what you’re experiencing is memory pressure on the Roon Server due to the volume of underlying metadata in the database. Roon Server gets caught up trying to finish background essential metadata maintenance for the library while music is playing. Indexing the library in the background is taking up 70-80% of Roon’s available RAM and processing capacity, leaving very little for playback. The symptom might affect networked playback, but the problem is nothing to do with your network, Eversolo, or Roon’s connection to Eversolo.
That’s not to say this is expected performance. Roon Server should easily handle a library at your size regardless of its metadata complexity. There are a handful of other reports from users who find their servers in a similar condition, and senior engineering is investigating to pin down and resolve the cause with a future Roon update.
Thanks for the note… no, as you say, I don’t hear playback issues. Only a hang loading one CD in the AM when I don’t reboot the Roon server. After the hang all is well until the next day.
I did not reboot the server this morning, but I was able to load a dozen CDs onto a playlist with no pauses. Maybe a coincidence, or maybe you’ve done something to the server. FYI
Had a hang loading a CD this morning at 7:58 MST.. about a 20 second hang…FYI