Max2Play Home › Forums › Max2Play as Squeezebox (Player / Server) › Library too big for a RPi3?
- This topic has 15 replies, 3 voices, and was last updated 6 years, 9 months ago by
Heiner premium.
-
2. Mai 2017 at 0:27 #28643
I’m trying to use M2P/LMS on an RPi3 and a 4TB USB drive to host a nearly 100,000 track audio library. I use the Squeezer app on a Nexus 7 (2013) tablet to control a Transporter, but when I scroll through the Music Folders list by flinging it, Squeezer shows lines of „Loading“ for a long time after the scrolling stops before finally showing the folder names. Sometimes during the wait the currently playing music even stops. Flinging the Song or Album lists is much much worse, sometimes even causing the Transporter to temporarily lose connection with LMS. I’ve tried the various Database Memory Config performance options in LMS but they didn’t seem to help.
Is this large of a library just too much for the RPi3? Could the Squeezer app be at fault? Or maybe M2P/LMS isn’t making efficient use of the quad-core processor? If I fling the Folders list several times, „top“ shows that one of the cores in the RPi3 CPU is at 100% until the folders are finally displayed. (The other 3 cores seem to be sitting idle for some reason.) None of this happened before switching from an ancient Intel P4 based system to the RPi3.
Any suggestions? Thanks!!
2. Mai 2017 at 9:57 #28646Hello bumpy7,
Please note that the maximum partition size is 2TB. If using larger disks you need to divide the disk space into partitions smaller than 2TB. And for more information, please check the following link:
https://www.max2play.com/en/wiki/creating-a-sambashare/
Regards,
Mohammad Mbydeen2. Mai 2017 at 16:31 #28671Thanks for the reply Mbydeen, but when I asked in December (2TB partition limit – Fact or Fiction?), Christoph said “It may now be possible to mount more than 2TB with the Raspberry Pi and Raspbian. But we haven’t a 4TB hard drive to test it.” User sfraser131 mentioned that he was also using a 4TB NTFS partition.
Do you have specific knowledge of why using an Advanced Format drive larger than 2TB with M2P would be a problem, or are you just repeating the (possibly outdated) limitation statement that appears on the M2P Filesystem Mount page?
In LMS I have increased the Priority of the Server to Above Normal, the Scanner to Normal and the Database Memory Config to High, and that seems to have helped quite a bit when scrolling by Music Folder. Scrolling by Songs, Albums and Artists is still a problem though, to the point where those browse options are unusable. Seems to me more like a large database issue, a transfer speed bottleneck due to the database being on the SD card or a processing bottleneck due to inefficient use of the multiple cores of the RPi3.
If nothing can be done to M2P or LMS to improve the performance of its large database handling, I can live with it. Or I suppose I could try an Odroid instead, but I rarely browse by anything other than Music Folder anyway.
Does anyone else have experience with this large of a library?
17. Mai 2017 at 16:01 #28888I’m still hoping that either an M2P or LMS developer will answer this question: Is M2P/LMS multi-threaded? If so, then why is it that when I try to fling/scroll a long list of songs (~100,000) in Squeezer, one of the RPi3 cores goes to 100% and stays there for a long time while the other 3 cores sit idle, until the list finally finishes displaying? Sometimes the currently streaming music even stops during this wait, so that must be on the same thread too, right? Is there any plan to more effectively utilize the multiple cores of the RPi3? Thanks!
18. Mai 2017 at 12:07 #28908Hi bumpy7,
How have you connected the external storage/Hard-disk ? Does this storage take the power from the RPI ? We recommend you to use separate power supply for the storage, it could be that the storage does not get enough power supply, which could affect the performance.
Best Regards,
Mohammad Mbydeen18. Mai 2017 at 15:32 #28927The HD has external power and is connected via USB to the RPi3. When you’re scrolling through the song list it seems like it’d be accessing the database anyway, which is on the SD card not the HD. (Yes, it’s a Class 10 card).
Still can’t/won’t answer the „Is M2P/LMS multi-threaded?“ question?
23. Mai 2017 at 12:23 #29029Hi bumpy7,
Max2Play is developed based on Raspbian OS.Therefore, Max2Play supports multi-threading. You can improve the performance of LMS, by clicking on settings–> Advanced–> Performance. Please have a look at the below pic:
and you can try to configure those settings as follows:
Database Memory Config —> High (recommended for machines with 1+ GB RAM)
Trigger Scan on Preference Changes —> Prompt to scan, but dont trigger it automatically
Artwork Pre-caching —> Pre-cache album, photo and video artwork
Best Regards,
Mohammad Mbydeen23. Mai 2017 at 20:11 #29040Thanks for the reply, but as I stated previously these performance options haven’t helped much.
You stated that M2P is multi-threaded but didn’t answer the same question about LMS. Is it? And if so, then why is it that when I try to fling/scroll a long list of songs (~100,000) in Squeezer, one of the RPi3 cores goes to 100% and stays there for a long time while the other 3 cores sit idle, until the list finally finishes displaying? Sometimes the currently streaming music even stops during this wait, so that must be on the same thread too, right?
7. Juni 2017 at 12:14 #29326Hi bumpy,
Sorry about the delayed response.
The Squeezebox Server is an open-source software whose active development in the slimdevices forums benefits our OS and every user of Max2Play. If you have specific questions on the technology behind the Logitech Media Server, please consult these forums: forums.slimdevices.com/
You can also look at these tips to optimize your sd card speed: https://www.raspberrypi.org/forums/viewtopic.php?f=63&t=144493
Nonetheless, a search task is most likely never split between cores and your list of 100.000 songs might be the critical mass for the Pi. You can consult the slimdevices forums on devices with more powerful single cores.
7. Juni 2017 at 20:28 #29351Hi Heiner,
I’ll try asking over at the slimdevices forum, but I wasn’t surprised about the search task not being split between cores, but rather that evidently the search task and the audio streaming aren’t on separate cores. When scrolling/flinging through the song list the currently streaming audio will sometimes stop, which is simply not acceptable behavior.
10. Juli 2017 at 17:55 #29877Hi bumpy,
No problem, I’ll take it out of the header. It was just meant to indicate that this topic’s subject has been thoroughly discussed and the ways in which to solve it were laid out in detail. If you still have a specific question or feedback, please let me know.
12. Juli 2017 at 20:34 #29896Hi bumpy,
If you still have unanswered questions or unaddressed concerns, please voice them here. I’d be happy to help.
Regarding your setup, you have several critical issues that cannot be solved by a software adjustment.
As mentioned before:
The Squeezebox Server is an open-source software whose active development in the slimdevices forums benefits our OS and every user of Max2Play. If you have specific questions on the technology behind the Logitech Media Server, please consult these forums: forums.slimdevices.com/You can also look at these tips to optimize your sd card speed: https://www.raspberrypi.org/forums/viewtopic.php?f=63&t=144493
Nonetheless, a search task is most likely never split between cores and your list of 100.000 songs might be the critical mass for the Pi. You can consult the slimdevices forums on devices with more powerful single cores.
12. Juli 2017 at 23:50 #29907Hi again Heiner, thanks for continuing to respond.
I plan to try a faster SD card and enable sd_overclock in /boot/config.txt and will let you know if it helps. Not sure when I’ll get the chance to try it though as other aspects of my life are taking up most of my time.
I’ll also try asking about LMS performance over at the Slim Devices forum, but to clarify once again; I never expected a search task to be split between cores, but I would think that search tasks and music streaming should be on separate cores. The worst aspect of the problem I’m seeing is when music playback stops for several seconds just because I „flung“ the song list in Squeezer.
19. Juli 2017 at 13:44 #29959Hi bumpy,
Thanks for the follow-up. And please keep us in the loop on what you find out, it might help us optimize our system in the future.
The Squeezer issue might also be related to the app itself and its connection to the LMS. Can you create the same issue on the LMS web interface or other apps?
-
You must be logged in to reply to this topic.