Same thing here - at first I thought Docker restarted the container - or the process crashed. Nope - massive GC. On my chart it kind of starts off with small sawtooth patterns, before finally just going to the moon with memory, and then you get the big GC dump.
While I am having better luck (so far) running containerised, as at least a short pause in playback is better than a full Roon server crash, its far from optimal.
Given that we are “tinkerers” and dare I say most of us probably have or have had some professional kind of IT career and exposure to how all of this works, and we are also (if I may speak for everyone) on the audiophile spectrum, we are poised to troubleshoot these things on our own. It is annoying but for the most part we are self sufficient.
However - not everyone is necessarily technically minded like us, and I gather than Roon doesn’t just market this product to people who are IT nerds! Logically, and possibly even necessarily, this must be happening to other users. Even ones that are running “supported.” Even those who are nowhere near our level of expertise in IT systems - but share the same audiophile spectrum and music desire all the same. What do they do? Turn it off and on again all the time? That, seems outrageous.
1 Like
mjw
(Here I am with a brain the size of a planet and they ask me to pick up a piece of paper. Call that job satisfaction? I don't.)
22
I, and another, reported errors with ARC because we used Graphene OS, and this provided additional data points on an issue (which was widespread at the time.)
That thread was moved to Tinkering, and closed, so I doubt Roon is interested in listening to these reports.
I, and another, highted issues with Qobuz syncing a year before it caused widespread problems, but no action was taken. I lost seven years play history etc., and as a consequence, now use Roon solely as a player–everything else is handled outside Roon.
Since closing down the Alpha channel, Roon has taken a different approach to customer feedback, and only takes action when issues affect many users.I guess this is a consequence of their success and growth.
Unfortunately there are some – usually self-declared “IT professionals” – who surface here occassionally that don’t know their trade, and this probably reflects poorly on those who do.
But I still think there is mileage here, and I will continue to monitor. Roon has performed better for me under a container than ROCK or a dedicated Linux box.
Sadly this is quite often the case - a few people can see the red flags before it happens, but it falls on deaf ears. Its not even quite good enough in the end for an “I told you so”.
Anyway, same here, my systems are constantly monitored so I will have many data points if required. So far as mentioned, the containerised instance is better - so fingers crossed.
I didn’t know you had this history. Thanks for sharing it I’ve had my share of issues, too, including database corruption which had been included in backups for weeks.
I’ve run Roon on a variety of linux distros, on Rock, as a Synology app, and in a container. A few years ago, I wrote this walkthrough on how to stand up a container on Synology (prior to them supporting compose files) just about 4 years ago: Walkthrough : Roon in a Docker container on Synology DSM 7. I’ve been running Roon in a container since before I wrote that. It’s been the most manageable, predictable, and stable runtime environment for me.
Something changed in May. It might have been a change to Roon code, an updated dependency. or a configuration change. Or a combination of multiple things that are interacting in an unexpected way.
What frustrates me about this isn’t the regression, it’s my perceived lack of accountability and customer engagement on issues. Bugs come and go - people make mistakes, the combinatorial complexity of code, config, dependencies is hard to manage. It’s the sense that there’s nobody that actually notices or cares that’s the problem. The issues that a number of us have been discussing since May should be causing alarms on Roon’s dashboards, not just mine, to be firing.
In any case, I hope they do find and fix this. There seems to be some indication that they’ve found something.
The EA build released yesterday is described as fixing a memory leak.
I don’t typically run Early Access builds because of the requirement that all clients be updated. In this case, I made an exception for testing and created an EA-based container. It takes days to see the full picture but the early results look identical to the current production build.
It stayed flat for a while, but then started its current pattern of growth of over 100MB / hour.
If anyone wants to play with my EA image, I’ll make it available over direct message. It runs Photon OS. I can also share the dockerfile if you want to build it yourself.
mjw
(Here I am with a brain the size of a planet and they ask me to pick up a piece of paper. Call that job satisfaction? I don't.)
Split this topic
26
Hey, @scidoner - there’s hope on the horizon for the memory leak(s) we were hitting. In case you’re not following, you can check out the recent threads here: Early Access - Roon Labs Community
Builds 1547 and later seem to completely fix the large memory leak I was hitting. If there are any leaks left, they’re much smaller. With these releases, memory use looks far, far better.
Letting you know in case you want to play with it.
I generally don’t like to play with beta products, but Roon is nearly unusable for me in its current state… I can’t go 2 days without it consuming all of my memory (unless I’m not playing any music… then I can go longer) which makes this tempting. I’m not even running Roon Appliance in a VM - it’s on bare metal. Hopefully this fix makes it into the release product shortly as this is driving me nuts.
2 Likes
mjw
(Here I am with a brain the size of a planet and they ask me to pick up a piece of paper. Call that job satisfaction? I don't.)
30
You may want to hold fire as the last production release was about 20 days ago, and releases are roughly every month.
2 Likes
mjw
(Here I am with a brain the size of a planet and they ask me to pick up a piece of paper. Call that job satisfaction? I don't.)
31
Here’s data from my [production] container. Not much data presently, but clearly memory is increasing periodically. Note, the activity around 29/05 was caused by a hardware change, and a few reboots.
