Roon stops playing mid track

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

AudioLinux (Archlinux) running on Ryzen 8 core 1700X with 16GB ram. Roon 1.6 build 401. Remotes up to date as well, MacOS, iPhone and iPad.

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

Core 1GB Ethernet. Cisco 2960G switch, unmanaged mode. 100Mbs internet cable modem. Remotes are wireless but they aren’t related to the Core crashes.

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

RoonCore I2S direct to PS Audio DirectStream DAC

Description Of Issue

Music playing normally for hours, some local some Tidal some Roon Radio chosen. Frequently using Roon Radio for the past few weeks, love it. Music will simply stop, Core CPU will spike and then drop to near zero, RoonServer processes die or cease to consume normal 4-15% cpu time.

Roon remotes all drop connection and then present the Login screen to authenticate/validate license. Attempting to authenticate times out.

Sometimes I can restart the RoonServer daemon and resume, but more often than not I have to restart the OS to resume normal play.

Post restart, all client auto connect, except for those I attempted to authenticate while Core was hung. For those, I can simply select the Core machine reporting Ready and connect normally.

This occurs 1-2 in 3-4 days. My system is active and playing nearly all waking hours. So it’s not terribly frequent. This same system was completely, 100% stable for several months running 1.5. No OS or Core restarts in months but that is not the case with 1.6.

Thanks.

Diagnostics may still be enabled on my account for some Radio/Tidal issues earlier this past week or week before last.

I think my issues may be related to some minor filesystem corruption. I’ve booted to single-user mode and run fsck. Repairs were made. I’ll let the system run and update…

Unfortunately the minor filesystem issues were not the cause. After ~3 hours of play (Roon Radio seeded with Tidal album), all clients drop back to the login/authorization screen. Authentication doesn’t timeout for over 4 mins, I quit timing.

If I click anywhere on the auth screen, it quits without error, returning to the ‘Connect’ to core screen. I can resubmit credentials but the same occurs.

RoonServer is still running per ‘htop’ but not consuming much cpu. Normal is 90-100% across 1-7 cores while starting up and just 4-15% on 1-3 cores when playing normally.

Restarting RoonServer results in the processes starting up and consuming about normal CPU % across many cores but dropping back to 0.1% after just a few seconds. I believe it is unable to read/load the database however the underlying filesystem is normally r/w via ssh.

My database resides on SSD internal sata connected to main board controller. It has its own power supply, also internal.

Drive is normally accessible, as is the boot OS drive, which is also SSD internal sata connected with its own PS as well. Roon uses 26GB for the database, cache, everything it creates in its database folder. 5 months ago I moved the database from the OS drive to my music storage data drive because of its size and the system has run fine until today.

No upsampling, only headroom clipping indicator, 0dB into HQPe running convolution filters and -5dB headroom.

Only recovery is to reboot server reloading all processes.

Now for the changes I’ve made in the past 3 days. I under-clocked the CPU and disabled SMT (AMD version of Hyper Treading). System ran for 2 listening sessions of 8 hours each without issue other than the intermittent Tidal latency. Tidal albums would still play no problem but the cover art load times were long, and Roon Radio would state ‘error limiting to local library’ but not miss a beat.

I’ll undo my proc changes and report.

Screen Shot 2019-03-30 at 5.44.38 PM
Screen Shot 2019-03-30 at 5.39.26 PM
Screen Shot 2019-03-30 at 5.35.18 PM

Is 26GB appropriate for these Overview stats?

After rebuilding my DB from scratch, all images downloaded, analysis not yet complete, ~20% done, DB is 5.7GB. Analysis is compete, DB is 5.8GB.

Perhaps my problems are related to a bloated, possibly corrupt, DB?

Any hints on how to avoid this kind of database growth in the future? It’s about 18 months old and I slowly add album over time but not more than 10 or so a month, mostly from Tidal. It’s a bummer to lose all of my metadata but worth it if the system remains stable.

System has remained stable since the DB rebuild, 12 hours so far.

I have returned to my former bios settings without issue. At this point, it appears the DB size/corruption was the cause of the problems recently, and likely in the past few months such as Tidal slowness, etc.

Hello @Larry_Post,

Thanks for letting me know that after the DB rebuilt, things are working as expected again.

This kind of database growth is very rare, we’ve only heard of similar reports 2-3 times since we released and we take something like this very seriously.

I just discussed your symptoms with QA and they noted that it is very likely that the low-level database corruption triggered the issue, but without log files to confirm the exact error state, it will prove difficult to document the specifics.

I’m hoping things will remain stable for your setup moving forward, but if this does occur again at any point in the future, please make a copy of the Logs folder from the Roon Database Location before wiping the database and starting fresh.

Since you have rebuilt the database already, the logs have unfortunately been overwritten in Roon and we would need these to do a deeper dive to pinpoint where exactly things were going wrong.

Thanks,
Noris

Will do. So far so good. Initially, I renamed the folder so I had a way back, but after 24 hours or so I deleted knowing I would not want to go back - other than potentially restore from back, which I still have.

Thanks, hope it doesn’t occur again.

1 Like

This topic was automatically closed 36 hours after the last reply. New replies are no longer allowed.