· Something else
· None of the above
· Other
This is a placeholder post
Wired Ethernet between my Nucleus +, MSB Renderer as Roon endpoint and the Internet
After the recent Roon updates, performance of my Roon system, which had been depressing (1 or 2 minutes waiting to load an album/track, or to edit anything, or to show the list of artists…), improved remarkably.
Everywhere?
Not entirely
One area of my library continues to be plagued by performance issues, as well as some bizarre malfunctions. It precisely covers Mozart’s “The New Complete Edition”, a CD collection comprising 200 units. I uploaded each of these CDs a couple of years ago as a separate entity to Roon and heavily edited them).
⁃ Very bad performance: These albums take between 20 and 45 seconds to load (it was twice as long before the Roon updates, by the way). Tracks may take between 20 and 30 seconds to start playing, sometimes more. Restarting a track by pressing the back arrows may take between 20 and 30 seconds to have effect (the track continues playing for that long before jumping back to the beginning, or just stops), compositions take around 30 seconds to load, and so on. Editing is painful, with waiting times of more than 30 secs to show an artist list…Sometimes, Roon freezes on me, and I have to restart/reboot. Strangely, as I said, the rest of my Roon library does not appear to show this behaviour: Albums seem to load instantly or, at worst, after 3 or 4 seconds, as well as tracks)
⁃ Immortal tag: Thinking that perhaps one tag with the collection name that I had applied to each of the 200 albums was causing the problem, I deleted it a while ago. This had no effect on performance. However, an image of the tag has “remained” in the first 50 albums, as a ghost. I can see it in the album page, but when I click on it, I just receive the alert “This tag was not found in your library”. The tag does not show in the tag list either. If I create a new tag with identical name, it appears twice in the albums with the immortal tag.
⁃ Another peculiar behaviour: In one album of the collection, part of the track list shows the wrong titles, which even are different to the titles showing below in the Now playing area. Changing metadata preferences or rescanning the album has no effect. Deleting and reimporting that album solves the issue for a few seconds, then the track list reverts to the wrong one.
Nothing has worked.
A pity, right now I am listening to all of Mozart’s compositions following the Köchel catalogue. With each successive K number, the same frustrating story. Now at K 185
Sorry to hear about your Mozart issues! If you could, please reproduce some of the performance stalls and share the specific date, time, and name of track played.
Can you test out removing the entire complication and re-analyze it within Roon?
We’d be happy to investigate the complication specifically if you’d like to send it via our Media Uploader. Let me know, thanks!
Hi, Benjamin, thank you for dropping by.
Excuse me, when you say “send the complication” via de media uploader, do you mean that I upload the whole Roon database? Or something else?
Loading album Mozart 225 - CD 024 - A4: 41 seconds : 21 June, 23:41h
Starting track String quartet no 4 . Allegro: 44 seconds : 21 June 23:45h
Loading album Mozart 225 - CD 025 - A4: 18 seconds 21 June 23, 23:46h
Starting Playing string quartet No 11: 48 seconds, 21 June 23:47h
Pressing Pause and back to restart above track: 54 seconds, 21 June 23:49h
I was not seeing most of the albums in Roon, only grey squares. Just restarted the App on my iPad. (23:54)
Still waiting for the albums to appear
Sluggishness seems to have expanded to all of my collection now.
After 1 min 21 seconds of grey squares, my albums became again visible (23:56)
A non-Mozart album (Callas sings Verdi arias) took 1 minute 01 secs to load. (23:58). I’d rather reboot my Nucleus plus now.
One more test: Album: Gemini and Aix concerti grossi: 7 seconds to load (June 22, 00:01)
Album: Boulez conducts Ligeti: 30 seconds to load.
Back to Mozart before rebooting
Loading album Mozart 225 - CD 050 - Orchestral: 22 seconds
I have a string quartet in pause. I’m resuming the playback now
Resuming playback after pause, string quartet in E flat: 12 seconds (not too bad, usually it is 2 or 3 times that long. June 22, 00:08h
Still many albums in my whole collection appear as grey squares. Rebooting Nucleus plus now (00:10)
AFTER REBOOT
Resuming playback after pause, string quartet in E flat: 4 seconds, not bad
Loading album Mozart 225 - CD 050 - Orchestral: serenades, divertimenti, cassations: 21 seconds (22 June, 00:14)
Cassation in B Flat, K 99 I Marche: Starting track: 55 seconds (00:16).
Cassation in B Flat, K 99 I Marche: : Pause and resume playing 34 seconds (00:18)
Cassation in B Flat, K 99 II Allegro molto: press back to restart track: 38 seconds (00:22)
Fairly bad.
Ok, let’s test other non-Mozart 225 albums
Loading album Corelli - Torelli: Christmas concertos: 3 seconds (00:25h0
Starting track Concerto Grosso op 6 no 8: 2 seconds
Restart same track: 3 seconds (00:28h)
Loading album Prokofiev: Pedro y el lobo: 4 seconds
Loading album Ravel Bolero: 4 seconds (00:30h)
Start track Bolero: 7 seconds (00:33h)
Restart same track: 6 seconds
Back to Mozart-225 collection
Loading Mozart 225 - CD 128,129,30 - Opera Figaro: 10 seconds (00:34)
Start Sinfonia 9 seconds (00:37)
Loading Die zauberflotte Mozart 225 - CD 137,138 (5 seconds)
Starting Overture: 25 seconds
Restarting same track 18 seconds (00:42h)
Jump to Bizet: Carmen we 31 act II: 40 seconds (00:44)
Loading Mozart 225 - CD 078 - Orchestral - Concertos: 38 seconds (00:44h)
Starting track 1, Concerto in C for flute, harp,and orchestra: 1. allegro: 49 seconds (00:47h)
Jump to track 2:(2: Andantino): 35 seconds (00:49h).
Jump back to restart track 2: 35 seconds (00:50h)
OK, now out of the Mozart collection again, to finish
Load album Ponchielli: la gioconda: 5 seconds (00:53h)
Play track 1 Preludio: 40 seconds (00:54h)
Jump,to track 2: 11 seconds (00:55h)
Jump back to beginning of track : 21 seconds (00:57h)
Loading Handel Water Music: 9 seconds
Starting track 1 Overture 13 seconds (00:59h)
I think the above reflects rather well my everyday experience with Roon
I am referring to the Mozart files causing you issue - I’ve shared our media uploader so our team can test further.
Does this issue happen no matter the remote device you use? Both your Mac and your iPad?
Yes, it happens in all remotes, iPad, iPhone, Mac. Also in Roon ARC, although in a slightly different way. But performance is fairly dismal.
[quote=“benjamin, post:8, topic:275675”]
I am referring to the Mozart files causing you issue - I’ve shared our media uploader so our team can test further.[/quote]
But this is about 176 albums, or 68 GB, the full Mozart 225 collection. Upload limit is only 5GB. Do you want me to upload them all or just a few?
Expanding / downloading the above image to the original size will give you an idea what I’m speaking about
Thanks for the update - yes just a small subset should work in this case.
After a fresh diagnostic scan of your Nucleus, we’re seeing repeating errors around Roon not being able to connect to your local network/ router. Here’s an example:
Trace: [rnet/RnetJsonClient] no data received for >10000ms. Killing connection.
Info: [remoting/serverconnectionv2] Client disconnected: 192.168.0.144:60339
Warn: [rnet/RnetJsonClient] failed to connect Connection refused
With RAATServer going down as well:
Trace: [raatserver] [RaatServer localhost @ 192.168.0.192:9200] lost client connection. Retrying
Trace: [raatserver] [RaatServer localhost @ 192.168.0.192:9200] connecting (attempt 1)
Is your Nucleus hardwired directly to your primary router? Does the same issue happen if you only use the system output from your phone or iPad for playback versus the MSB?
Lastly, if you set both Background audio analysis and on-demand analysis to ‘Off’ is there any change in behavior?
The Nucleus is connected to a switch that connects to the main network switch that connects to the router. I connected the Nucleus directly to the main switch to see what happened, and performed some measures. Again, more than 20 seconds to load an album,
Up to 50 seconds. Between 35 and 45 seconds to start a track, with isolated cases at 6 or 7 seconds. So, about the same as before.
Please note that I have this issue only in this particular collection of albums. Loading any other albums may take between 1 and 5 seconds.
It is possible that a faulty connection is making things a bit worse, but again, I do not have that problem in the rest of my library, that I’m aware of.
Yes, indeed. The problem appears to be the same with any endpoint and any remote, iPad, Mac, iPhone, Naim loudspeaker, all of them
Ok, I’m uploading 137 files, 2.7 GB
Background analysis was already off (I had already turned it off to see if things improved) I now turned off in demand analysis.
No change
I have a zipped file with the contents of my database. Would you like me to upload it, too?
Only just seen this , in MusicBrainz the Mozart edition is split into volumes of bite size pieces the biggest around 20 CD it cleared up the original sluggish performance of the big box. It was highlighted a while back as one of the breaking points of the search development
Simply any box above a nominal 50 CD is not worth running in Roon , I keep my JRiver legacy running purely for BIG BOXES
Roon has indicated a dev effort specifically for box sets , hopefully soon
You can also try rebooting the server that helps , but IDing the set as the subsets is the way to go
You will see the Karajan and Abbado sets are all “volumed” into 30 is CD sets
Hi, Mike, thanks for jumping in.
I was fearing myself that creating a huge set with 200 CDs would bring trouble to the Roon library. Thus, I imported all CDs in the Box one by one, as separate albums, except for operas, which could include between 1 and 3 CDs. The result was about 170 albums. I added a tag to all of them “New Mozart 225 Collection”.
I analysed them one by one, as if I had purchased them separately. Each has its own title: often it follows the sequence in the collection. Sometimes, when importing, I selected the metadata of the original CDs outside of the Mozart 225 box, as the box is made of material previously released as independent CDs. If you click on the image I uploaded in an earlier post, with all my Mozart 225 covers in Roon, you will get an idea. I edited the metadata to have better consistency and accuracy. So I tend to believe that all these albums are treated independently from one another inside Roon…
Now that I think of this, it suddenly occurs to me a possible cause of the whole issue:
When looking at the metadata after the import, I discovered that most of the albums showed more than 1,200 (yes) credits each. To each and every album analysed as part of the Mozart 225 box, Roon had seemingly attributed a credit for every single person or group with a role in any of the CDs of the full Mozart 225 set.
Could it be possible that Roon is reading the 1,200 credits everytime I load an album or start a track? Imagine, all the conductors, orchestras, soloists, engineers, featuring in any of the thousands of tracks in the Mozart 225 set, be processed by Roon whenever I play any track in the set, looking for links and the like. So, even if I imported the CDs as separate entities, the way Roon attributed its metadata in the first place would make the library treat them still as a 200 CD set, at least as credits are concerned. Performance could suffer horribly. At the same time, this could be consistent with the improvement I noticed after some updates a couple months ago (loading waiting time was reduced almost in half).
When I saw those metadata, I went to clean them up. I deleted the wrong ones, leaving only the essential ones, on an album by album basis. Now I will not find orchestras and choirs playing piano sonatas.
However, the credits I deleted have not entirely dissappeared from Roon. Roon has not “forgotten” them. They remain in the library, not visible if I look for them in the credit search, but still appearing as “removed” in the album, when I open the edit window for it.
See below_
If you have the time, make a complete copy and recreate the folder structure above, you will find the disc details in MusicBrainz, then remove the big set , do a database clean up (VERY IMPORTANT otherwise you will get the old metadata rather than forcing a rescan) then add the new folders to your Roon library. I think you may be surprised.
As far as I know Mozart 225 individual discs are not previous releases, eg the sonatas are a mix of Uchida , Brendel etc, so Roonwill not ID them. However the track timings will match the discs in Mozart 225 , hence what you are seeing. As they are seen as a part of Mozart 225 they will accrue the credits of Mozart 225 , exactly what you are seeing.
The new DG Bach and Beethoven were the same.
The old Mozart edition (1989?) was a collection of previously issued box sets and identity as such .
I wouldn’t be surprised if all the “pseudo albums” you created are unidentified which would add to the performance issues.
The golden rule is leave a box as its box unless it has identified sub volumes, otherwise you will get a mess. Some boxes are exceptions like the Original Jackets series where they are a collection of previously released CDs. They are a rarity.
I have found that any boxes >50 CD will show performance issues, 10 boxes such as Beethoven sonatas are a breeze.
We are reliably informed that a major development of box sets is in the wings so …
Actually, they are all identified by Roon (hence the issue with the credits)
Hi, Mike, thanks for the reply
In my view, the folder structure I currently have isn’t likely to be the cause of the problem, as there is not any big set in it. Each album is inaide its own folder, which is in the corresponding artist folder, as the normal Roon practice. You can see that in the screenshots below. Note that an artist folder (Christopher Hogwood_Academy of Ancient Music) includes, among many other albums, a Mozart 225 album (in this case, No 58). The same with the Claudio Abbado folder, and so on
As I said, I had imported them as separate albums. Thus, I do not see how grouping the files in a different file structure would make things change.
I have the feeling now, after reflecting about your input, that this may be related to how the Roon database handles the metadata in the collection, rather than to the location of the music files in the Nucleus + disc. I thank you for that, particularly if this discussion helps Roon developers to better address the issue of CD sets. The case is extreme with 200 CD and, precisely for this reason, can be enlightning
If you rename a folder , Roon will treat it as a separate album. I am not sure how Roon’s id Algorithm works but I think it’s based on track times.
If you take one of your separated albums and manually ID it I suspect you will get an ID On Mozart 225. Think a combination of track timings and track numbers and of a album , it’s likely to be unique or very close to . How many albums have identical track times and track count to the nearest second ?
This is how MucicBrainz does it
https://musicbrainz.org/doc/Disc_ID_Calculation
For a box set the best advice is a master folder a Mozart 225 , then each CD as CD1 … CD200, with no additional text . Also set the Album to Mozart: The new Complete Edition and the Album Artist = Various Artists.
I hate to say it but I think at the moment you are making your own problems
Roon will get the best chance of ID.
Splitting the disc all over the place will not help especially if you have altered the Album name as well
I think you assumption is correct , every split album will carry the credits for the entire box.
My advice would be to either follow the volume splits i showed above or bulk the entire 200 together and then use the assorted tools , composition, filter and focus to drill into the album, even as the 200 CD album it’s unlikely to be worse than you are currently experiencing.
EDIT
I just tested the theory, I added a random CD , CD39 to my library, it was identifies as CD39 of he Mozart 225 set as below
with a complete set of credits for the whole box set
