Discussion:
[mpd+ncmpcpp] mpd hangs when manually moving forward on track, or to nex track
(too old to reply)
Javier Vasquez via arch-general
2017-03-27 02:22:58 UTC
Permalink
Raw Message
Hi all,

If someone is familiar with mpd (user, non root) + ncmpcpp on Arch,
perhaps can help me out...

Whenever I move forward/backward on the same playing track (f/b), mpd
just hangs, usually so hard, that mpd needs to be restarted to do
anything. A workaround is to pause the track, and then move
forward/backward, and then play again, most of the time that works...

When playing a track, and then moving to next/prior track (>/<), then
mpd hangs. To get it back alive, it's enough to stop mpd (s), and
start any track (enter). The pause work around for moving
forward/backward on the same track doesn't work, mpd keeps hanging,
but stopping mpd (s), and starting any track (enter) works. Stopping
mpd (s), to move to the next track as a work around doesn't work,
given there's no current track playing, therefore mpd can find the
next...

So it's sort of weird that moving around where to play mpd, kinds of
makes it hang...

BTW, I have 2 mpd outputs, the pulse output but also a visual fifo,
however by keeping only the pulse audio doesn't help, so It's not
related to the visual fifo...

Any hint would be appreciated...

Thanks a lot,
--
Javier
Gaetan Bisson
2017-03-27 03:53:38 UTC
Permalink
Raw Message
Post by Javier Vasquez via arch-general
If someone is familiar with mpd (user, non root) + ncmpcpp on Arch,
perhaps can help me out...
Whenever I move forward/backward on the same playing track (f/b), mpd
just hangs, usually so hard, that mpd needs to be restarted to do
anything. A workaround is to pause the track, and then move
forward/backward, and then play again, most of the time that works...
I can't reproduce this with ncmpc so it's probably ncmpcpp-specific.
Perhaps a rebuild against the new libmpdclient will fix this. But
anyhow, please create a bug report on our tracker. This is the proper
way to get things fixed.

Cheers.
--
Gaetan
Gaetan Bisson
2017-03-27 03:55:21 UTC
Permalink
Raw Message
Post by Gaetan Bisson
Post by Javier Vasquez via arch-general
If someone is familiar with mpd (user, non root) + ncmpcpp on Arch,
perhaps can help me out...
Whenever I move forward/backward on the same playing track (f/b), mpd
just hangs, usually so hard, that mpd needs to be restarted to do
anything. A workaround is to pause the track, and then move
forward/backward, and then play again, most of the time that works...
I can't reproduce this with ncmpc so it's probably ncmpcpp-specific.
Perhaps a rebuild against the new libmpdclient will fix this. But
anyhow, please create a bug report on our tracker. This is the proper
way to get things fixed.
And to help us better understand your issue please tell us what specific
versions of what packages are affected. Does downgrading ncmpccc, mpd,
and/or libmpdclient fix the issue, etc.

Cheers.
--
Gaetan
David Runge
2017-03-27 10:40:35 UTC
Permalink
Raw Message
Hey Javier,
Post by Javier Vasquez via arch-general
Hi all,
If someone is familiar with mpd (user, non root) + ncmpcpp on Arch,
perhaps can help me out...
Whenever I move forward/backward on the same playing track (f/b), mpd
just hangs, usually so hard, that mpd needs to be restarted to do
anything. A workaround is to pause the track, and then move
forward/backward, and then play again, most of the time that works...
When playing a track, and then moving to next/prior track (>/<), then
mpd hangs. To get it back alive, it's enough to stop mpd (s), and
start any track (enter). The pause work around for moving
forward/backward on the same track doesn't work, mpd keeps hanging,
but stopping mpd (s), and starting any track (enter) works. Stopping
mpd (s), to move to the next track as a work around doesn't work,
given there's no current track playing, therefore mpd can find the
next...
So it's sort of weird that moving around where to play mpd, kinds of
makes it hang...
BTW, I have 2 mpd outputs, the pulse output but also a visual fifo,
however by keeping only the pulse audio doesn't help, so It's not
related to the visual fifo...
I basically have the same setup (all latest packages), but I'm not
experiencing your issues. I'm using a custom service file [1] though, as I
sometimes have several mpd instances on different machines, that I like
to be able to use with pulseaudio.

Something I could think of related to your issue might be your
harddrive and machine (obviously).
Where do the files reside, that you want to play? What kind of hardware
are you using? Is it slow (e.g. connected through usb1, etc)?
You could use tools such as iotop [2] to see whether mpd creates crazy
amounts of I/O for your system (or maybe by proxy due to a slow drive).
Also check htop [3] while you try to go forwards/backwards.
I'm quite sure you'll see something spike.

After all, I don't think it's the software, but something hardware
related.

Best,
David


[1] https://git.sleepmap.de/software/uenv.git/about/
[2] https://www.archlinux.org/packages/community/any/iotop/
[3] https://www.archlinux.org/packages/extra/x86_64/htop/
--
https://sleepmap.de
Javier Vasquez via arch-general
2017-03-28 03:05:49 UTC
Permalink
Raw Message
Post by Javier Vasquez via arch-general
...
If someone is familiar with mpd (user, non root) + ncmpcpp on Arch,
perhaps can help me out...
Whenever I move forward/backward on the same playing track (f/b), mpd
just hangs, usually so hard, that mpd needs to be restarted to do
anything. A workaround is to pause the track, and then move
forward/backward, and then play again, most of the time that works...
When playing a track, and then moving to next/prior track (>/<), then
mpd hangs. To get it back alive, it's enough to stop mpd (s), and
start any track (enter). The pause work around for moving
forward/backward on the same track doesn't work, mpd keeps hanging,
but stopping mpd (s), and starting any track (enter) works. Stopping
mpd (s), to move to the next track as a work around doesn't work,
given there's no current track playing, therefore mpd can find the
next...
So it's sort of weird that moving around where to play mpd, kinds of
makes it hang...
BTW, I have 2 mpd outputs, the pulse output but also a visual fifo,
however by keeping only the pulse audio doesn't help, so It's not
related to the visual fifo...
...
OK, for those who replied, it basically works like that in all my
boxes, most are mintbox 2 with 2 core i5 as processors, intel hda
audio and SATA SSD. One laptop is an HP EliteBook 840 G3 with 2 core
i7 processors, intel hda audio (or plantronics USB audio headset) and
PCIe SSD. One laptop is aToshiba nb505 with Atom 64 processor, intel
hda and SATA SSD.

All of them use up to date x86_64 Arch, and have the same behavior.
The related packets:

mpd 0.20.6-1
libmpdclient 2.11-1
libmpd 11.8.17-2
ncmpcpp 0.7.7-4
pulseaudio 10.0-2
linux 4.10.5-1

Notice I recompiled the ncmpcpp package getting the exact same
behavior on the EliteBook.

Regarding what I'm playing, any ogg vorbis file I have experience the
same. I don't have mp3 files. It doesn't matter if the files are in
local SSD, or network mounted drive. In any case is the same.

I didn't file a bug at 1st, hopping it was a known issue perhaps, and
people would know about a workaround. See the behavior can be
reproduced on any machine I try, all Arch x86_64. I haven't tried
ncmpc (I like ncmpcpp better), but moc doesn't have any issues, though
I prefer mpd...

I'll try ncmpc next, to see if I only get the behavior with ncmpcpp...

Thanks,
--
Javier
Javier Vasquez via arch-general
2017-03-28 03:09:03 UTC
Permalink
Raw Message
Post by Javier Vasquez via arch-general
Post by Javier Vasquez via arch-general
...
If someone is familiar with mpd (user, non root) + ncmpcpp on Arch,
perhaps can help me out...
Whenever I move forward/backward on the same playing track (f/b), mpd
just hangs, usually so hard, that mpd needs to be restarted to do
anything. A workaround is to pause the track, and then move
forward/backward, and then play again, most of the time that works...
When playing a track, and then moving to next/prior track (>/<), then
mpd hangs. To get it back alive, it's enough to stop mpd (s), and
start any track (enter). The pause work around for moving
forward/backward on the same track doesn't work, mpd keeps hanging,
but stopping mpd (s), and starting any track (enter) works. Stopping
mpd (s), to move to the next track as a work around doesn't work,
given there's no current track playing, therefore mpd can find the
next...
So it's sort of weird that moving around where to play mpd, kinds of
makes it hang...
BTW, I have 2 mpd outputs, the pulse output but also a visual fifo,
however by keeping only the pulse audio doesn't help, so It's not
related to the visual fifo...
...
OK, for those who replied, it basically works like that in all my
boxes, most are mintbox 2 with 2 core i5 as processors, intel hda
audio and SATA SSD. One laptop is an HP EliteBook 840 G3 with 2 core
i7 processors, intel hda audio (or plantronics USB audio headset) and
PCIe SSD. One laptop is aToshiba nb505 with Atom 64 processor, intel
hda and SATA SSD.
All of them use up to date x86_64 Arch, and have the same behavior.
mpd 0.20.6-1
libmpdclient 2.11-1
libmpd 11.8.17-2
ncmpcpp 0.7.7-4
pulseaudio 10.0-2
linux 4.10.5-1
Notice I recompiled the ncmpcpp package getting the exact same
behavior on the EliteBook.
Regarding what I'm playing, any ogg vorbis file I have experience the
same. I don't have mp3 files. It doesn't matter if the files are in
local SSD, or network mounted drive. In any case is the same.
I didn't file a bug at 1st, hopping it was a known issue perhaps, and
people would know about a workaround. See the behavior can be
reproduced on any machine I try, all Arch x86_64. I haven't tried
ncmpc (I like ncmpcpp better), but moc doesn't have any issues, though
I prefer mpd...
I'll try ncmpc next, to see if I only get the behavior with ncmpcpp...
...
I installed ncmpc:

ncmpc 0.27-1

And I experience exactly the same. sounds more like a mpd thing then?
--
Javier
Javier Vasquez via arch-general
2017-03-28 03:31:09 UTC
Permalink
Raw Message
Post by Javier Vasquez via arch-general
Post by Javier Vasquez via arch-general
Post by Javier Vasquez via arch-general
...
If someone is familiar with mpd (user, non root) + ncmpcpp on Arch,
perhaps can help me out...
Whenever I move forward/backward on the same playing track (f/b), mpd
just hangs, usually so hard, that mpd needs to be restarted to do
anything. A workaround is to pause the track, and then move
forward/backward, and then play again, most of the time that works...
When playing a track, and then moving to next/prior track (>/<), then
mpd hangs. To get it back alive, it's enough to stop mpd (s), and
start any track (enter). The pause work around for moving
forward/backward on the same track doesn't work, mpd keeps hanging,
but stopping mpd (s), and starting any track (enter) works. Stopping
mpd (s), to move to the next track as a work around doesn't work,
given there's no current track playing, therefore mpd can find the
next...
So it's sort of weird that moving around where to play mpd, kinds of
makes it hang...
BTW, I have 2 mpd outputs, the pulse output but also a visual fifo,
however by keeping only the pulse audio doesn't help, so It's not
related to the visual fifo...
...
OK, for those who replied, it basically works like that in all my
boxes, most are mintbox 2 with 2 core i5 as processors, intel hda
audio and SATA SSD. One laptop is an HP EliteBook 840 G3 with 2 core
i7 processors, intel hda audio (or plantronics USB audio headset) and
PCIe SSD. One laptop is aToshiba nb505 with Atom 64 processor, intel
hda and SATA SSD.
All of them use up to date x86_64 Arch, and have the same behavior.
mpd 0.20.6-1
libmpdclient 2.11-1
libmpd 11.8.17-2
ncmpcpp 0.7.7-4
pulseaudio 10.0-2
linux 4.10.5-1
Notice I recompiled the ncmpcpp package getting the exact same
behavior on the EliteBook.
Regarding what I'm playing, any ogg vorbis file I have experience the
same. I don't have mp3 files. It doesn't matter if the files are in
local SSD, or network mounted drive. In any case is the same.
I didn't file a bug at 1st, hopping it was a known issue perhaps, and
people would know about a workaround. See the behavior can be
reproduced on any machine I try, all Arch x86_64. I haven't tried
ncmpc (I like ncmpcpp better), but moc doesn't have any issues, though
I prefer mpd...
I'll try ncmpc next, to see if I only get the behavior with ncmpcpp...
...
ncmpc 0.27-1
And I experience exactly the same. sounds more like a mpd thing then?
Filed the following bug:

https://bugs.archlinux.org/task/53477

Can you reproduce with similar configs to mine (see bug for configs)?
I use non root user systemd service...

Thanks,
--
Javier
Javier Vasquez via arch-general
2017-03-28 03:46:54 UTC
Permalink
Raw Message
Post by Javier Vasquez via arch-general
Post by Javier Vasquez via arch-general
Post by Javier Vasquez via arch-general
Post by Javier Vasquez via arch-general
...
If someone is familiar with mpd (user, non root) + ncmpcpp on Arch,
perhaps can help me out...
Whenever I move forward/backward on the same playing track (f/b), mpd
just hangs, usually so hard, that mpd needs to be restarted to do
anything. A workaround is to pause the track, and then move
forward/backward, and then play again, most of the time that works...
When playing a track, and then moving to next/prior track (>/<), then
mpd hangs. To get it back alive, it's enough to stop mpd (s), and
start any track (enter). The pause work around for moving
forward/backward on the same track doesn't work, mpd keeps hanging,
but stopping mpd (s), and starting any track (enter) works. Stopping
mpd (s), to move to the next track as a work around doesn't work,
given there's no current track playing, therefore mpd can find the
next...
So it's sort of weird that moving around where to play mpd, kinds of
makes it hang...
BTW, I have 2 mpd outputs, the pulse output but also a visual fifo,
however by keeping only the pulse audio doesn't help, so It's not
related to the visual fifo...
...
OK, for those who replied, it basically works like that in all my
boxes, most are mintbox 2 with 2 core i5 as processors, intel hda
audio and SATA SSD. One laptop is an HP EliteBook 840 G3 with 2 core
i7 processors, intel hda audio (or plantronics USB audio headset) and
PCIe SSD. One laptop is aToshiba nb505 with Atom 64 processor, intel
hda and SATA SSD.
All of them use up to date x86_64 Arch, and have the same behavior.
mpd 0.20.6-1
libmpdclient 2.11-1
libmpd 11.8.17-2
ncmpcpp 0.7.7-4
pulseaudio 10.0-2
linux 4.10.5-1
Notice I recompiled the ncmpcpp package getting the exact same
behavior on the EliteBook.
Regarding what I'm playing, any ogg vorbis file I have experience the
same. I don't have mp3 files. It doesn't matter if the files are in
local SSD, or network mounted drive. In any case is the same.
I didn't file a bug at 1st, hopping it was a known issue perhaps, and
people would know about a workaround. See the behavior can be
reproduced on any machine I try, all Arch x86_64. I haven't tried
ncmpc (I like ncmpcpp better), but moc doesn't have any issues, though
I prefer mpd...
I'll try ncmpc next, to see if I only get the behavior with ncmpcpp...
...
ncmpc 0.27-1
And I experience exactly the same. sounds more like a mpd thing then?
https://bugs.archlinux.org/task/53477
Can you reproduce with similar configs to mine (see bug for configs)?
I use non root user systemd service...
Following up on htop and iotop... I don't see anything abnormal on
total memory used, 13%. No swapping at all. The load average
indicated is 0.34 0.34 0.35. The mpd CPU% load keeps between 0 and 5,
while the ncmpcpp keeps between 0 and 1. The mpd Mem% keeps steady in
0.3, and the ncmpcpp one in 0.1. The i/o indicates NO access to disk
at all while hanging, not for mpd neither for ncmpcpp...

I see nothing weird about memory neither cpu consumption, and the i/o
shows what's going on, nothing, mpd just hanging...
--
Javier
Javier Vasquez via arch-general
2017-03-28 04:09:06 UTC
Permalink
Raw Message
Post by Javier Vasquez via arch-general
Post by Javier Vasquez via arch-general
Post by Javier Vasquez via arch-general
Post by Javier Vasquez via arch-general
Post by Javier Vasquez via arch-general
...
If someone is familiar with mpd (user, non root) + ncmpcpp on Arch,
perhaps can help me out...
Whenever I move forward/backward on the same playing track (f/b), mpd
just hangs, usually so hard, that mpd needs to be restarted to do
anything. A workaround is to pause the track, and then move
forward/backward, and then play again, most of the time that works...
When playing a track, and then moving to next/prior track (>/<), then
mpd hangs. To get it back alive, it's enough to stop mpd (s), and
start any track (enter). The pause work around for moving
forward/backward on the same track doesn't work, mpd keeps hanging,
but stopping mpd (s), and starting any track (enter) works. Stopping
mpd (s), to move to the next track as a work around doesn't work,
given there's no current track playing, therefore mpd can find the
next...
So it's sort of weird that moving around where to play mpd, kinds of
makes it hang...
BTW, I have 2 mpd outputs, the pulse output but also a visual fifo,
however by keeping only the pulse audio doesn't help, so It's not
related to the visual fifo...
...
OK, for those who replied, it basically works like that in all my
boxes, most are mintbox 2 with 2 core i5 as processors, intel hda
audio and SATA SSD. One laptop is an HP EliteBook 840 G3 with 2 core
i7 processors, intel hda audio (or plantronics USB audio headset) and
PCIe SSD. One laptop is aToshiba nb505 with Atom 64 processor, intel
hda and SATA SSD.
All of them use up to date x86_64 Arch, and have the same behavior.
mpd 0.20.6-1
libmpdclient 2.11-1
libmpd 11.8.17-2
ncmpcpp 0.7.7-4
pulseaudio 10.0-2
linux 4.10.5-1
Notice I recompiled the ncmpcpp package getting the exact same
behavior on the EliteBook.
Regarding what I'm playing, any ogg vorbis file I have experience the
same. I don't have mp3 files. It doesn't matter if the files are in
local SSD, or network mounted drive. In any case is the same.
I didn't file a bug at 1st, hopping it was a known issue perhaps, and
people would know about a workaround. See the behavior can be
reproduced on any machine I try, all Arch x86_64. I haven't tried
ncmpc (I like ncmpcpp better), but moc doesn't have any issues, though
I prefer mpd...
I'll try ncmpc next, to see if I only get the behavior with ncmpcpp...
...
ncmpc 0.27-1
And I experience exactly the same. sounds more like a mpd thing then?
https://bugs.archlinux.org/task/53477
Can you reproduce with similar configs to mine (see bug for configs)?
I use non root user systemd service...
Following up on htop and iotop... I don't see anything abnormal on
total memory used, 13%. No swapping at all. The load average
indicated is 0.34 0.34 0.35. The mpd CPU% load keeps between 0 and 5,
while the ncmpcpp keeps between 0 and 1. The mpd Mem% keeps steady in
0.3, and the ncmpcpp one in 0.1. The i/o indicates NO access to disk
at all while hanging, not for mpd neither for ncmpcpp...
I see nothing weird about memory neither cpu consumption, and the i/o
shows what's going on, nothing, mpd just hanging...
BTW, regarding HW, I've also tried SATA HD on the mintbox 2 machines...
--
Javier
Loading...