Official moOde 6.6.0 support thread

[TheOldPresbyope]

Very odd cos your MPD log showing "added" doesn't agree with mpc stats which queries the MPD database same as if you ran command below.

telnet localhost 6600
stats

It took ages for the 32-bit to finish indexing, but it successfully finished, and all the tracks are indexed and shown in the UI. So some odd problem in the 64-bit world…

FWIW, I had virtually the same experience with 6.6.0 indexing a large 4TB USB drive of music files. With the 64-bit kernel chosen, the indexing of 23,786 separate albums stalled out the same way, with an MPD Error message displayed. With the 32-bit kernel chosen, the indexing was successfully completed. I was then able to switch over to the 64-bit kernel, and everything's working fine. When I add more files to the USB drive, I suppose I'll first switch back over to the 32-bit kernel, get the indexing completed, then switch over once more to the 64-fit kernel.

So yes, as hestehandler noted, there does seem to be something odd going on in the 64-bit world...

Not to put too fine a point on it, the moOde System Config page still says "NOTE: The 64-bit kernel is a testing option for Pi-4, Pi-3, and Pi-2B rev 1.2 boards." 

All of moOde has been compiled to run on the 32-bit kernel so changing to the 64-bit kernel isn't introducing some new moOde capability. I haven't found any advantage to using it, and I find it strange that the situation where it seems to be causing issues for a few users involves large collections. Simplistically, this would seem to be a domain where 64-bit pointers would have a distinct advantage.

As I've reported previously, I haven't experienced personally any issue with either kernel and with either MPD version so I'm waiting for actionable data. What, for example, was the "MPD Error message displayed"?

Regards,
Kent

[Tim Curtis]

Note that even though the Linux kernel supports 32-bit or 64-bit architecture the Linux userland applications which are everything other than the kernel for example MPD, Samba, PHP, etc, run in a 32-bit architecture. This means that there could be issues when 32 bit userland interacts with 64 bit kernel.

There is a beta 64-bit userland RaspiOS but not ready for prime time just yet.
https://www.raspberrypi.org/forums/viewt...p?t=275370

My goal in offering 32 and 64 bit kernel options and 0.22~git MPD in moOde is to allow experimentation and feedback as to what works and what doesn't during the transition to the production 64 bit RaspiOS. When 64-bit RaspiOS is finally released we will have an experience base of what to look out for :-)

[Dradder]

Very odd cos your MPD log showing "added" doesn't agree with mpc stats which queries the MPD database same as if you ran command below.

telnet localhost 6600
stats

It took ages for the 32-bit to finish indexing, but it successfully finished, and all the tracks are indexed and shown in the UI. So some odd problem in the 64-bit world…

FWIW, I had virtually the same experience with 6.6.0 indexing a large 4TB USB drive of music files. With the 64-bit kernel chosen, the indexing of 23,786 separate albums stalled out the same way, with an MPD Error message displayed. With the 32-bit kernel chosen, the indexing was successfully completed. I was then able to switch over to the 64-bit kernel, and everything's working fine. When I add more files to the USB drive, I suppose I'll first switch back over to the 32-bit kernel, get the indexing completed, then switch over once more to the 64-fit kernel.

So yes, as hestehandler noted, there does seem to be something odd going on in the 64-bit world...

Not to put too fine a point on it, the moOde System Config page still says "NOTE: The 64-bit kernel is a testing option for Pi-4, Pi-3, and Pi-2B rev 1.2 boards." 

All of moOde has been compiled to run on the 32-bit kernel so changing to the 64-bit kernel isn't introducing some new moOde capability. I haven't found any advantage to using it, and I find it strange that the situation where it seems to be causing issues for a few users involves large collections. Simplistically, this would seem to be a domain where 64-bit pointers would have a distinct advantage.

As I've reported previously, I haven't experienced personally any issue with either kernel and with either MPD version so I'm waiting for actionable data. What, for example, was the "MPD Error message displayed"?

Regards,
Kent

Not to put too fine a point on it, indeed, but as recently as 6.3.0 2019-10-02, Moode with the 64-bit kernel selected was able to perfectly index the large 4TB collection that 6.6.0 is now able to index only with the 32-bit kernel selected. So this is not some "new Moode capability," but one that Moode had previously, but which has been lost in the post-6.3.0 versions; I downloaded & installed & tested them all.

The MPD Error was the same one I reported in the discussion here about Moode 6.5.x, in regard to the same indexing failure issue; I'll attach it here again.

[TheOldPresbyope]

Thanks for the error message.

Function openMpdSock() has been in moOde for a very long time. Here's the code snippet from /var/www/inc/playerlib.php

Code:

// core mpd functions

// AG from Moode 3 prototype
// TC retry to improve robustness
function openMpdSock($host, $port) {
       for ($i = 0; $i < 6; $i++) {
               if (false === ($sock = @stream_socket_client('tcp://' . $host . ':' . $port, $errorno, $errorstr, 30))) {
                       debugLog('openMpdSocket(): connection failed (' . ($i + 1) . ')');
                       debugLog('openMpdSocket(): errorno: ' . $errorno . ', ' . $errorstr);
               }
               else {
                       $resp = readMpdResp($sock);
                       break;
               }

               usleep(500000); // .5 secs
       }

So, one thing to be done is to turn on moOde debugging and then examine the moOde log (and MPD log too) after the error occurs in order to get more diagnostic information. But be sure to include more of the log than just the "openMpdSocket()..." line.

I just read through the earlier r6.5 thread and I see the related discussion came to an resolved end because no more diagnostic information was provided. Without it, people who are genuinely interested in fixing issues are hamstrung.

Regards,
Kent

[SoundAdikt]

It's another way to push you to get a new phone I suppose.
Are you not able to get round the issue by using a different browser or do they use the same javascript engine?  Sorry I don't use Apple devices so am just curious.

You are almost certainly correct

I've tried using different browsers from the Apple Appstore: 

Most (for example Google Chrome, Firefox, Opera & DuckDuckGo) are incompatible with iOS 9.3.5, and won't even install.

Others (like MS Edge (old version) & Dolphin) will install, but have the same issues as Safari, in that MoOde will start up, the library loads, but the Library (Radio, Album, Tag etc) & Playlist (also Appearance) screens are "all grey".

Probably time to pension off the iPad 2, unless Swizzle can fix it easily - for sure, it's not worth compromising the new build of MoOde, just for the sake for the older iPads.

I agree, although it may be possible to detect and use the older Javascript/CSS that worked on older devices.  For me at least, I use Kazoo as the controller and this runs fine on iOS 9.3.5 (for the moment at least).

[grasshopper]

It's another way to push you to get a new phone I suppose.
Are you not able to get round the issue by using a different browser or do they use the same javascript engine?  Sorry I don't use Apple devices so am just curious.

You are almost certainly correct

I've tried using different browsers from the Apple Appstore: 

Most (for example Google Chrome, Firefox, Opera & DuckDuckGo) are incompatible with iOS 9.3.5, and won't even install.

Others (like MS Edge (old version) & Dolphin) will install, but have the same issues as Safari, in that MoOde will start up, the library loads, but the Library (Radio, Album, Tag etc) & Playlist (also Appearance) screens are "all grey".

Probably time to pension off the iPad 2, unless Swizzle can fix it easily - for sure, it's not worth compromising the new build of MoOde, just for the sake for the older iPads.

I agree, although it may be possible to detect and use the older Javascript/CSS that worked on older devices.  For me at least, I use Kazoo as the controller and this runs fine on iOS 9.3.5 (for the moment at least).

What about using the Squeezelite renderer?  I realise it won't use MPD, which is not necessarily a bad thing, last time I used LMS which was several years ago admittedly, it was very good.
There may be a control point that works on the iPad & still give the benefit of using MoOde.
May be worth spending a couple of hours investigating is you have a spare Pi laying around to put LMS on?

[kit1cat]

I have used my system with LMS on my pi4 (NAS) and moode on my pi3B+ (Player) running the Squeezelite renderer with no problems, controlled via my ipads.

[happymax]

Hi everyone,
I have the same problem as other users: my USB DAC (Khadas tone card) is no longer recognized with Moode 6.6.0 with a new installation on Pi4B with 4 GB of RAM; with version 6.5.2 everything works fine.

I noticed that if I restart without connecting the USB dac I can connect to Moode but in the Modify MPD configuration, Device type does not show "USB audio device"; if I restart with USB dac connected I cannot connect to Moode and the device is not visible on the network either in wireless or cable mode.

Thanks for all this beautiful and great work !!!!!

[TheOldPresbyope]

Folks,

let's not keep repeating the same issue about enumeration of USB DACs. It's known, the cause is known (changes introduced with RPiOS did us in), and a fix with 'new and improved features' is being developed and tested. I'm sure an update will be winging its way to the moOde server when it's ready.

Regards,
Kent

[Tim Curtis]

Right, the presence of an HDMI connected display now causes the ALSA card numbers to dynamically change. This was introduced in the new RaspiOS 10.4 and Linux kernel 5.4.y branch. It's to provide better support for the twin HDMI ports on the Pi-4.

moOde has up to this point relied on card0 being assigned to the Pi Headphone jack or HDMI port, or an I2S audio device and card 1 being assigned to a USB Audio device.

The new dynamic ALSA card numbering however can result in card0 = HDMI, card1 = Headphone jack and card2 = USB Audio device.

Until the fix is released, unplug HDMI display, reboot and USB device should again be recognized on card1.