I just watched RAM use by RoonAppliance in it’s latest jump from 3.5 to 4GB
It appears that in my case, on this new dev PC each jump in RAM use is about 400-500MB.
And each jump in RAM happens at the same time as a large single-core spike (looks like one core maxed) lasting maybe around 30 seconds.
If I do nothing at all with Roon then this still happens at intervals of a few hours.
FYI, since moving the server to my development PC this problem has not really gone away.
If anything, the problem is worse on this PC because the RAM issue seems to accumulate here much faster.
At the moment I am working around this by closing and relaunching Roon Server at least once a day. This being my development PC, I am not happy to just sit and watch a large pool of memory being eaten away until it crashes. Re-launching it once or twice a day prevents that happening but it’s not a great solution and I am not very comfortable with it.
I have still not tried a Linux install (I do have one available but I’m not sure I wat to take the effort to move the server again without a likely solution) and from from comments on these forums it sounds like Roon has the same problems over on Linux too so I doubt I would gain much by trying. In any case I have better tools and familiarity on Windows to assess issues with the server.
I am still desperate for a fix so please let me know when you have something new to try and I will give things another go.
Hi @Andrew_Beveridge ,
I am looking over your thread and I noticed you mentioned that the issue still happens with Qobuz disabled, which is strange. We had a few similar reports we were investigating but they all seemed to be tied to the Qobuz usage, which sounds like it’s not the same in your case. Do you by any chance use a duel IPv4/IPv6 network on these PCs? Is there any change if you try to temporarily disable IPv6?
I am new to Roon and noticed similar memory problems. I have Roon server on a dedicated windows pc. Roon client runs on macOS Sonoma. When I start Roon client it immediately uses about 800MB. As soon as I stream an album from Qobuz, the memory usage slowly increases to many GBs. As soon as I stop streaming, the memory usage does not increase further. If I disable the Qobuz service and only use music files, the memory usage does not increase. The only exception I have noticed so far is when I tag albums (file or Qobuz). For every tag I add or remove from an album, the memory usage increases by about 40MB.
Eventually Roon takes up so much memory that I have to shut down Roon. This happens several times when I add albums to my library.
Hi noris
I do not (knowingly) use IPv6 on my LAN. It is disabled on several boxes here BUT since it tends to default to on it is probably enabled on one or two. It is enabled on this development PC (where Roon server is currently running) and it was enabled on the old dedicated server.
you mentioned that the issue still happens with Qobuz disabled
yes, it was definitely happening on the old server with Qobuz disabled. This led me to believe it was not to do with Qobuz so when I setup the new server I re-enabled Qobuz (because I have a fair few playlists with Qobuz tracks in them).
I have just disabled both IPv6 and Qobuz service on this new server. I will close it and relaunch now and we’ll see what happens.
I will report back in 24-48 hours. Thanks.
Roon eats memory. Fact. On all platforms. I find the Roon staff responses in this thread a little condescending, TBH. Please man up and take the reports seriously, As bad as it is, I can’t imagine no one at “Roon Staff” hasn’t noticed it. Look at this, for a “remote”, running a day!!
It’s so bad I’m using Tidal Connect to my devices and bypassing Roon
OK, so in the last 24 hrs with IPv6 disabled and Qobuz disabled this server has been fairly stable at 1.5GB memory footprint.
So, to free my development PC I have de-authorized this one and re-enabled the RoonServer on my dedicated i5-NUC. And then I disabled IPv6 on there (Qobuz was already disabled there - I checked so can confirm, Qobuz was definitely disabled when I was seeing that big memory leak and CLR stack overflows).
The NUC is hovering up and down about 1GB memory use.
BUT it is now also trying to add 28 items to my library. So far it has taken about 2 hours to do this and the spinny thing in top right is still going. If I click on that a window pops up informing me, “Adding music to library: Of 28 tracks, 28 added, 26 identified”. Two hours ago it had only done two of those so maybe it will finish eventually? The purple progress bar gives the impression it is maybe 75% done.
CPU use is mostly steady at <1% with blips of ~2% to 4%. I have no idea what it is doing.
And it’s still going. Seems to be stuck at this…
Also, very wierdly the Windows client here is telling me that it is using 16 to 18% of my GPU !!! 
What on earth are you doing to use 16% of my GPU. Are you by any chance redrawing EVERYTHING continually, flat out in a loop? I cannot think what else would account for that. That is just wrong. Please don’t do that. You don’t need to be redrawing anything except the little spinny-thing at this point. There isn’t even anything playing.
A restart/reboot of the server usually fixes that.
Yeah, it’s a bit disappointing that the most common fix is, “have you tried turning it off and on again?”.
I really want Roon to work well. I like what Roon promises and for a while it seemed to do all of that fairly well. Nothing else seemed to come close to the real solution.
But lately its like Roon doesn’t deliver what it promises any more.
Fixes everything (short term), we’re all slaves to software!
You mean “bad software”.
It is not supposed to be like that.
OK. I restarted the server.
Several minutes later it is using 25% (ie. all of 1 core) and has 1.5GB of memory, and is still growing.
IPv6 is still off, as is Qobuz. We’ll see where it settles tomorrow and whether it is then stable.
And today I keep seeing this. Just trying to play a playlist from the Roon Home screen
Not only that but the Roon client UI is very sluggish. Pause/play button seems to have a lag of several seconds. Several tracks in this playlist have skipped after just a few seconds of playback… without me touching anything it has advanced to track 7, one track at a time, in about 30 seconds. It just stops playing suddenly, displays that message, and the resumes but playing the next track.
IPv6 is still disabled. Qobuz is still disabled.
My LAN is fine. My broadband is fine as far as I know… connection has not gone down and there is zero other traffic at this time and no sign of any unexpected network activity.
However I have not seen this problem before.
And Roon Server, with IPv6 and Qobuz both disabled, does not appear to be hogging memory at the moment. I will try re-enabling each separately to see what happens. Doesn’t Roon have some sort of automated pre-release test-suite for all this kind of thing? That would surely be easier (and quicker and more reliable) than relying on me to manually reconfig and test all the time.
Weird! I seem to have difficulty playing anything much from Tidal…
Now this…
But Tidal plays just fine using the Tidal desktop client.
(still with IPv6 and Qobuz disabled)
Going to try reboot server and this client, and all target devices now.
OK, so everything rebooted (properly rebooted, not just close and relaunch server).
No Tidal, just playing an album from my library, then a few seconds in the audio stops and I see this…
Followed by the same “Too many failures. Stopping playback.” message again.
(IPv6 still disabled, Qobuz still disabled, Podcasts folder still disabled).
Router shows almost zero network traffic. The Roon server PC is also showing no significant network activity except for RoonAppliance process which was showing about 3Mbps network activity before the problem then intermittent 12 Mbps and 0.1Mbps
Update: OK, Have tried repeatedly and it seems that since I re-authorised this server nothing plays more than a few seconds any more. I just tried logging out and then back in again on this client PC (Windows desktop)… and it just sat busy at the jellyfish.
Then whilst it was doing that RoonAppliance crashed on my server. RoonServer restarted. roon client now just does nothing… blank window. I’m going to close it, and the server, and start them both again, server first.
Checking the NUC Windows logs I see two CLR Stack overflows. One at 11:22 today, and the last one at 15:01 today (ie. just now).
Have restarted everything… but it’s just the same. Nothing plays for more than a few seconds any more :-/
Update2: Nothing playing here but RoonAppliance just crashed again. Windows event logs show another CLR stack overflow at 15:37
That is the fourth crash today. And Roon is just not working any more. And I don’t have time for this any more.
I have just uninstalled Roon completely (inc database) from the server.
Then I rebooted the server.
Then I checked the disk (SSD) for errors.
Then I installed a fresh download of RoonInstaller64.exe earlyaccess.
I did not restore any backup.
I fired up a Windows desktop client on a neighboring PC to finish config in settings, adding the two storage locations.
I waited whilst Roon scanned everything.
Then tried playing things. A few times it has paused and the server disappeared and reappeared later. In at least a couple of those times I saw the RoonAppliance process disappear in Task Manager and reappear a bit later with heavy CPU use and a growing memory footprint.
I checked the Windows logs at the server and I see that RoonAppliance has crashed (and restarted itself) 4 times. In each case it is a similar CLR Stack overflow.
Hi @Andrew_Beveridge,
Since this thread has gone one so long without resolution I think we need to take a step back and analyze this without preconceptions about what the problem is. Let’s start by just stating the bad user experience you are having. To my understanding the problem is you are getting slow loading and this happened on both of the computers that you tried. Can you supply some screenshots of the sections of the Roon app where you are seeing the jellyfish for a while?
Since we’ve exhaustively checked into memory usage fruitlessly let’s look at network. In your original post you say you have a 60 Mbps connection on your router and you say you have a 1 Gb LAN. Can you clarify what is meant by that? 60Mbps is very slow.
As a side note the swing back and forth between more memory use and less memory use is likely because garbage collection (i.e. the thing that frees up memory) is triggered in intervals not exactly when a memory block is no longer in use. You can prove this by checking RoonServer_log when you see high memory usage in task manager. That log will have a bunch of lines like
09/06 12:37:20 Info: [stats] 2103331mb Virtual, 292mb Physical, 143mb Managed, 1083 Handles, 28 Threads
That’s how you know how much memory is currently being actively used on Roon.
Can you clarify what is meant by that? 60Mbps is very slow.
That is the broadband speed. Just emphasizing plenty of bandwidth for Tidal/Qobuz.
Can you supply some screenshots of the sections of the Roon app where you are seeing the jellyfish for a while?
One of the most noticeable was the home screen, right below the Recent Activity panel. I have a whole bunch of screenshots from when things were happening but I also now have various changes all the way through as we tried things… disabling things, switching servers etc.so I’m not sure how helpful they are going back to them.
My current setup is with Roon Server back on the AMD dev PC. This things is v.fast and has 32GB. The problems show less on here and it certainly does the CLR stack overflow crash less often (but still does happen).
Audio analysis in setup is still turned off.
Qobuz is still disabled.
I also noticed some things that are hard to 100% confirm (since just anecdotal) but I THINK…
Memory problems were reduced by turning off IPv6.
I suspect issues are more common when using a WiFi endpoint.
Two ropieee endopints had to revert to Pi on-board WiFi because ropieee support for RTL8011 chipsets failed in a recent update. The on-board WiFi gives higher WiFi error rates because of antenna. In these circumstances both servers piled up the memory more quickly and both gave stack overflows more often… especially with the server on the older i5 NUC.
I know what you are saying about CLR garbage collection. That may account for the memory and CPU blips I see now (though it implies you still have a LOT of churn) but it does not account for what I was seeing before. If it causes stack overflows then you are mis-managing something. 4MB is a lot of stack.
And if you are losing stack frame on events like packet loss then you are probably losing allocated objects from stackframe pointers at the same time… those things will not get garbage collected and you will not get the lost stack back. That kind of problem will accumulate. Garbage collection will only free memory that you hand back (and destruct objects for classes with allocated RAM. It won’t pick up things you just leave lying around.
Does you regular testing include cases with differing amounts of packet loss?
I have just today switched one of my previous WiFi endpoints to wired. And the other one has it’s fast WiFi with antenna back and is now much nearer the AP. Together with the IP6 disabling (unless I am imagining the impact of that) it all seems much more stable now.
If/when I see any more problems I will check RoonServer_log (that’s handy - thanks for that).
If all remains well on this server I may re-install on my i5 NUC to see how that goes since that showed the problem much more.
Note these are all wired. It is only those two endpoints that were WiFi. Now only one (and that is 5G WiFi about 1.5m from the AP).