mvordeme wrote:
> I'll have an eye on whether it always affects the same tracks.
I think I've identified a corner case that could explain the last block
of a track being skipped
LMS 7.9 on Pi 3B+ & Odroid-C2 - *SqueezeAMP!*, 5xRadio, 3xBoom, 4xDuet,
1xTouch, 1 SB3. Sonos PLAY:3, PLAY:5,
I'll have an eye on whether it always affects the same tracks.
scaleo home server 2105 & picoreplayer 6.1.0 | logitech media server
8.1.0 | server power control 20120716.103808 |
transporter & duet & touch & boom & radio | rotel rc-995 & rmb-100 |
nubert nuvero 140
mvordeme wrote:
> So far, Tidal has been playing without any problems. No stuck
> connections.
>
> The skipping of the last few seconds of a track seems unrelated. It
> occurs now and then. I have no idea how to pin it down.
So skipping still occurs ?
A possible hypothesis.
I'm assuming
So far, Tidal has been playing without any problems. No stuck
connections.
The skipping of the last few seconds of a track seems unrelated. It
occurs now and then. I have no idea how to pin it down.
scaleo home server 2105 & picoreplayer 6.1.0 | logitech media server
8.1.0 | server power
OK - i've moved to the dev release - Logitech Media Server Version:
8.1.0 - 1608064080 @ Tue Dec 15 22:13:24 CET 2020 - should that have the
changes in? Will let you know what I find.
That's the latest build. It should have all the changes committed today :-).
--
Michael
mherger wrote:
> > That's interesting. I used to find that I'd get stuttering with
> older
> > hardware players with flac, due I think to tidal flac being highly
> > compressed. So, I have a flac to flac re-encode with -1 to make it
> less
> > hard work for my boom and transporter. Does your
bpa wrote:
> mvordeme's persistence "a voice in the wilderness" and narrowing
> problem down to network was essential.It was something that annoyed me almost
> every day, but that made it
reproducible and easy to investigate. If I knew Perl, I would even had
looked at the code, but as it
That's interesting. I used to find that I'd get stuttering with older
hardware players with flac, due I think to tidal flac being highly
compressed. So, I have a flac to flac re-encode with -1 to make it less
hard work for my boom and transporter. Does your comment mean that I
don't need that
The fix is now in the 8.1 branch.
I wondered if this was going to make its way into the 8.01 branch as
well in a while. I'd rather avoid running the dev branch on my main
system, particularly at this time of year when it seems to be running
pretty much full time.
And I prefer to have
mherger wrote:
> > FLAC would play without transcoding - until we
> introduced transparent transcoding for TIDAL(?) to work around the
> stuttering reported before LMS 8.
>
> Michael
Hi Michael
That's interesting. I used to find that I'd get stuttering with older
hardware players with
psketch wrote:
> I wondered if this was going to make its way into the 8.01 branch as
> well in a while. I'd rather avoid running the dev branch on my main
> system, particularly at this time of year when it seems to be running
> pretty much full time.
I think it may need to be "tested" and
mherger wrote:
> Yep. And before LMS8 we would not transcode AAC, would we?l
LMS would transcode HTTP/AAC only for internet streams - so no EOF.
For MPEG4/AAC no standard LMS for streams. To trasncode MPEG-4
podcasts/streams probably used ffmpeg with input from $URL$ not stdin.
paul- wrote:
> The fix is now in the 8.1 branch.
I wondered if this was going to make its way into the 8.01 branch as
well in a while. I'd rather avoid running the dev branch on my main
system, particularly at this time of year when it seems to be running
pretty much full time.
Looking at the code fix and as it is so "central" - why was it not
prevalent before now and can it explain any other "bugs" with streaming
services.
I'm not sure we did a lot of (or any at all!) transcoding single tracks
from online sources in the past. AAC transcoding was added in 8.0,
mvordeme's persistence "a voice in the wilderness" and narrowing
problem down to network was essential.
Looking at the code fix and as it is so "central" - why was it not
prevalent before now and can it explain any other "bugs" with streaming
services.
The code path with EOF would only have
Thanks - I'm really glad you found it so quickly. I think LMS (and
Michael :)) need more of us knowing the code to help and I'm candidly
(and often) struggling a fair bit.
Yes indeed! I need more of you! Thank you very much bpa_44!
--
Michael
___
paul- wrote:
> BTW Phillip,Thanks for pointing me in the right direction..it
> was a fun lesson in Perl and LMS code paths.
Thanks - I'm really glad you found it so quickly. I think LMS (and
Michael :)) need more of us knowing the code to help and I'm candidly
(and often) struggling a
BTW Phillip,Thanks for pointing me in the right direction..it
was a fun lesson in Perl and LMS code paths.
piCorePlayer a small player for the Raspberry Pi in RAM.
Homepage: https://www.picoreplayer.org
Please 'donate'
philippe_44 wrote:
> Yes, you found it. I don't know why this delete is here. It does not
> make much sense as there is a close() called later when the pipeline
> ends. And delete() does not call the destructor, it will only be if
> there is no more reference to that $sock and there might be
The fix is now in the 8.1 branch.
piCorePlayer a small player for the Raspberry Pi in RAM.
Homepage: https://www.picoreplayer.org
Please 'donate'
(https://www.paypal.com/cgi-bin/webscr?cmd=_donations=U7JHY5WYHCNRU=GB_code=USD=PP%2dDonationsBF%3abtn_donateCC_LG%2egif%3aNonHosted)
if you like
paul- wrote:
> Yup when a pipeline is in use..Opened issue
> https://github.com/Logitech/slimserver/issues/475
Yes, you found it. I don't know why this delete is here. It does not
make much sense as there is a close() called later when the pipeline
ends. And delete() does not call the
Yup when a pipeline is in use..Opened issue
https://github.com/Logitech/slimserver/issues/475
piCorePlayer a small player for the Raspberry Pi in RAM.
Homepage: https://www.picoreplayer.org
Please 'donate'
paul- wrote:
> I think I found the problemgoing to take this to git to discuss.
Per my other post, seems it's only when transcoding, the socket is not
closed by the local endpoint when it is always closed when there is no
transcoding. I've reproduced that 10+ times
LMS 7.9 on Pi 3B+ &
I think I found the problemgoing to take this to git to discuss.
piCorePlayer a small player for the Raspberry Pi in RAM.
Homepage: https://www.picoreplayer.org
Please 'donate'
mvordeme wrote:
> Doesn't it have to send something, or else the O/S wouldn't report the
> connection as CLOSE_WAIT?
I agree, hence I'm confused and need to take more logs
LMS 7.9 on Pi 3B+ & Odroid-C2 - *SqueezeAMP!*, 5xRadio, 3xBoom, 4xDuet,
1xTouch, 1 SB3. Sonos PLAY:3, PLAY:5, Marantz
paul- wrote:
> So it's in the Async HTTP. I see no difference in the logs whether it
> is streaming direct vs streaming to a transcoder helper. I made a few
> changes to the logging.
>
> 1. It's not tracking content received vs content length. Even though
> Content-Length is present.
> 2.
So it's in the Async HTTP. I see no difference in the logs whether it
is streaming direct vs streaming to a transcoder helper. I made a few
changes to the logging.
1. It's not tracking content received vs content length. Even though
Content-Length is present.
2. It's requesting to close the
paul- wrote:
> Is there an external process involved? There doesnt appear to be
> one.
No - MPEG-4 AAC to ADTS AAC is done in Slim::Formats::Movie.pm
ADTS-AAC is then fed into transcode chains (i.e. faad) via stdin.
> The port in CLOSE_WAIT status is a local problem. How is the socket
>
philippe_44 wrote:
>
> - Remember that streaming Tidal is not direct (in AAC), it is proxied
> through LMS (I mode) as we need to re-wrap the MP4-AAC into ADTS frames
> (this is a change I made last summer)
>
I dont see this when playing Tidal to a squeezelite player.it goes
directly.
philippe_44 wrote:
> - When transcoding, the Tidal server does not send the close sequence
> (at least I don't see it in wireshark)
Doesn't it have to send something, or else the O/S wouldn't report the
connection as CLOSE_WAIT?
scaleo home server 2105 & picoreplayer 6.1.0 | logitech media
psketch wrote:
> Not sure if it's transcoding, or just flac. I saw the problem yesterday
> playing a tidal album that was flac, played on pcp with native. When I
> use transporter, I do have a custom re-encode set as the older hardware
> can't handle flac files which are really heavily
It happens when LMS is using an external helper (at least some of them).
It might be tied to only certain plugins too.
I tested bit faad->flac from Tidal, And sox converting mp3 to flac from
Pandora.
piCorePlayer a small player for the Raspberry Pi in RAM.
Homepage:
paul- wrote:
> So I signed up for a tidal account..got a month free to play. Here
> in the US, tidal is using AWS Cloudfront.
>
> I started testing with no transcoding in the LMS server. pCP players
> handle AAC natively.
>
> When a track is played from tidal, it sends to complete
You can look at Slim::Networking::Async::HTTP.pm
Code:
if ( !defined $result || $result == 0 || (defined
$self->response->headers->header('Content-Length') &&
length($self->response->content) ==
$self->response->headers->header('Content-Length')) ) {
paul- wrote:
> So I signed up for a tidal account..got a month free to play. Here
> in the US, tidal is using AWS Cloudfront.
>
> I started testing with no transcoding in the LMS server. pCP players
> handle AAC natively.
>
> When a track is played from tidal, it sends to complete track
So I signed up for a tidal account..got a month free to play. Here
in the US, tidal is using AWS Cloudfront.
I started testing with no transcoding in the LMS server. pCP players
handle AAC natively.
When a track is played from tidal, it sends to complete track as fast as
it can. All of
mvordeme wrote:
> One thing I found out just now: Whenever I skip the currently playing
> track, the connection is closed properly. The new connection is using a
> different port, so I don't think it is the old connection reused. Only
> when a track is played to the end does the connection hang
One thing I found out just now: Whenever I skip the currently playing
track, the connection is closed properly. The new connection is using a
different port, so I don't think it is the old connection reused. Only
when a track is played to the end does the connection hang around. I
could have a
mvordeme wrote:
> I don't know about Perl, but in Java, not closing a resource explicitly
> in the "finally" block but relying on garbage collection is a sin, and
> code inspectors will flag occurrences as a critical bug. Can I check the
> Perl code myself?
Garbage collection is normal in
bpa wrote:
> I'm not sure of the Tidal and new AAC support details, but within Perl,
> normally a TCP connection is an "object".
> TCP conection is closed when an explicit "close" is performed or when
> TCP connection object is destroyed (i.e. no variable has a references to
> it).
> Object
Is this another user with the same problem
https://forums.slimdevices.com/showthread.php?112300-Attention-TIDAL-users-changes-ahead!=998883=1#post998883
bpa's Profile: http://forums.slimdevices.com/member.php?userid=1806
mvordeme wrote:
> Restarting the LMS moved the counters a little.
>
> >
Code:
> > tc@piCoreServer:~$ head -n2 /proc/net/netstat | cut -d' ' -f 51-55
> TCPAbortOnData TCPAbortOnClose TCPAbortOnMemory TCPAbortOnTimeout
TCPAbortOnLinger
> 191 3667 46 503 0
mvordeme wrote:
> It does, but the output doesn't look interesting to me.
Pity because "lsof -i" can show something like below. Note the
"CLOSE_WAIT" on chrome.
Code:
firefox 2229 testuser 387u IPv4 8306345 0t0 TCP
Restarting the LMS moved the counters a little.
Code:
tc@piCoreServer:~$ head -n2 /proc/net/netstat | cut -d' ' -f 51-55
TCPAbortOnData TCPAbortOnClose TCPAbortOnMemory TCPAbortOnTimeout
TCPAbortOnLinger
191 3667 46 503 0
scaleo home server
Players are currently re-buffering. Time to check the stats.
Code:
tc@piCoreServer:~$ netstat -tn 2> /dev/null | grep CLOSE_WAIT | wc -l
110
bpa wrote:
> To see if there is some odd network activity at the same time as the
> playback issues,
mvordeme wrote:
> Alright, I decided to change my mindset and try to get to the bottom of
> this myself. This looks very much like a connection leak.
>
> LMS running for half a day:
> >
Code:
> > tc@piCoreServer:~$ netstat -tnp
> netstat: can't scan /proc - are you
If Fastly is "regional" CDN it might explain why other users do not
"see" the problem
The basic principle of CDNs is that they have distributed datacenters to
be close to the user. So yes, this could very much be a regional issue.
--
Michael
___
mherger wrote:
> > Who is 151.101.114.142 ? - it seems to be in Germany. Is it
> > mysqueezebox.com or Tidal ?
>
> It's Fastly - a CDN provider. Most likely what TIDAL is using (we're
> not)
Do you think this is a Europe wide CDN for Tidal or Germany/W.Europe
specific ?
If Fastly is
Who is 151.101.114.142 ? - it seems to be in Germany. Is it
mysqueezebox.com or Tidal ?
It's Fastly - a CDN provider. Most likely what TIDAL is using (we're not).
--
Michael
___
unix mailing list
unix@lists.slimdevices.com
Does pcp have a lsof command
Code:
lsof -i
bpa's Profile: http://forums.slimdevices.com/member.php?userid=1806
View this thread:
mvordeme wrote:
> What about all those open connections? Today, I was on the phone a lot,
> so I didn't pay much music, but there are already 97 connections waiting
> to be released
I think CLOSE_WAIT is where the local end (i.e. LMS) has asked to CLOSE
the connection but the far end (i.e.
The columns seem to be shifted by one.
Code:
tc@piCoreServer:~$ head -n2 /proc/net/netstat | cut -d' ' -f 51-55
TCPAbortOnData TCPAbortOnClose TCPAbortOnMemory TCPAbortOnTimeout
TCPAbortOnLinger
137 2918 24 275 0
What about all those open
OK. Normally TCP connections are reliable but when a connection has to
aborted it usually means an odd network issue or perhaps one end of the
connection has power off, broken,crashed.
To see if there is some odd network activity at the same time as the
playback issues, you could have a shell
Code:
tc@piCoreServer:~$ cat /proc/net/netstat | cut -d' ' -f 1,35,41
TcpExt: TCPLostRetransmit TCPTimeouts
TcpExt: 6863 13430
IpExt:
IpExt:
scaleo home server 2105 & picoreplayer 6.1.0 | logitech media server
8.0.1 | server power control
mvordeme wrote:
> si
> >
Code:
> > TcpExt: SyncookiesSent SyncookiesRecv SyncookiesFailed EmbryonicRsts
PruneCalled RcvPruned OfoPruned OutOfWindowIcmps LockDroppedIcmps ArpFilter TW
TWRecycled TWKilled PAWSActive PAWSEstab DelayedACKs DelayedACKLocked
si
Code:
TcpExt: SyncookiesSent SyncookiesRecv SyncookiesFailed EmbryonicRsts
PruneCalled RcvPruned OfoPruned OutOfWindowIcmps LockDroppedIcmps ArpFilter TW
TWRecycled TWKilled PAWSActive PAWSEstab DelayedACKs DelayedACKLocked
DelayedACKLost ListenOverflows
mvordeme wrote:
> unfortunately, not
The "-s" just make /proc/net/netstat look pretty. Does Pcp have
/proc/net/netstat
e.g.
cat /proc/net/netstat
bpa's Profile: http://forums.slimdevices.com/member.php?userid=1806
unfortunately, not:
Code:
tc@piCoreServer:~$ netstat -ts
netstat: invalid option -- 's'
BusyBox v1.30.1 (2019-06-20 23:16:54 EDT) multi-call binary.
Usage: netstat [-ral] [-tuwx] [-enWp]
Display networking information
-r Routing table
-a
mvordeme wrote:
> Devices in the 192.168.178.20# address range are players,
> 192.168.178.104 is my Windows PC. And yes, I need to sudo it, next time.
> Is there anything else I can do to find out more?
After ssh'ing into Touch, you could check on the interface statistics
with ifconfig or
Playing the third track after restart:
Code:
tc@piCoreServer:~$ sudo netstat -tnp
Active Internet connections (w/o servers)
Proto Recv-Q Send-Q Local Address Foreign Address State
PID/Program name
tcp0 0 192.168.178.101:3483
Alright, I decided to change my mindset and try to get to the bottom of
this myself. This looks very much like a connection leak.
LMS running for half a day:
Code:
tc@piCoreServer:~$ netstat -tnp
netstat: can't scan /proc - are you root?
Active Internet connections
Can this discussion be moved?
scaleo home server 2105 & picoreplayer 6.1.0 | logitech media server
8.0.1 | server power control 20120716.103808 |
transporter & duet & touch & boom & radio | rotel rc-995 & rmb-100 |
nubert nuvero 140
Might have to make a post with Tidal in the subject.
piCorePlayer a small player for the Raspberry Pi in RAM.
Homepage: https://www.picoreplayer.org
Please 'donate'
I re-wired mains and network to the pi and, while I was at it, updated
the LMS to Logitech Media Server Version: 8.0.1 - 1607087403 @ Fri Dec 4
14:26:56 CET 2020. It is still showing the same behaviour. I noticed one
more thing. A few tracks before Tidal streaming breaks and the pCP
interface
The same thing again today. The moment Tidal gets stuck, the pCP web
interface does not load properly any more. And I noticed another thing:
I configured the Transporter to power down the audio section when off.
It was sitting there, clicking about once per minute. It seems that the
entire
Tidal got stuck again, today, and I tried a few more things.
- On the Receiver, the MPEG-4 podcast did not play, either. It showed
the same behaviour as tracks from Tidal, i.e. re-buffering and now and
then playing a few seconds.
- On squeezelite, Tidal tracks would still play, but
Next time the LMS goes into denial, I can just try to play it. Do I
simply need the Podcasts plug-in?
scaleo home server 2105 & picoreplayer 6.1.0 | logitech media server
8.0.0 | server power control 20120716.103808 |
transporter & duet & touch & boom & radio | rotel rc-995 & rmb-100 |
nubert
mvordeme wrote:
> No, I only pay for compressed audio. 320 kBit AAC is what the Controller
> shows.
The problem could be Tidal AAC specific (which are in MPEG4) or streamed
MPEG-4 in general.
Below is an example link for France Musique MP4 jazz podcast (there are
many) - which is streamed
No, I only pay for compressed audio. 320 kBit AAC is what the Controller
shows.
scaleo home server 2105 & picoreplayer 6.1.0 | logitech media server
8.0.0 | server power control 20120716.103808 |
transporter & duet & touch & boom & radio | rotel rc-995 & rmb-100 |
nubert nuvero 140
Can you confirm which Tidal format is associated with the problem -
Tidal MPEG4 or Tidal Flac ? Since you are referencing faad - I assumed
you are *NOT* using Tidal HiFi Flac streams.
bpa's Profile:
I have not updated for a week, now, because I didn't want to introduce
new uncertainties. The version is Logitech Media Server Version: 8.0.1 -
1606059084 @ Sun Nov 22 16:49:21 CET 2020.
It is buffering. The Controller will show rebuffering ... and only now
and then a short snippet will play.
So appears to be a tidal problem.Which version of LMS are you using?
Is it really buffering, or rather stuttering? The latter is a known
issue with some TIDAL tracks.
--
Michael
___
unix mailing list
unix@lists.slimdevices.com
So appears to be a tidal problem.Which version of LMS are you using?
piCorePlayer a small player for the Raspberry Pi in RAM.
Homepage: https://www.picoreplayer.org
Please 'donate'
A few results. When Tidal streaming got stuck again, today, I checked a
few things.
- The faad and flac processes had the same memory footprint as usual.
- The LMS would still spawn new faad and flac processes when skipping
tracks.
- Local FLAC files (16 bit / 44.1 kHz) would play fine
Currently, my FLAC library is on the other server, but I'll copy a few
albums and test.
Right now, I only have a Touch which used to be able to play Tidal
natively. There have been no problems with that, as far as I know.
"Seeking" within a track didn't work any more, last time I checked, but
Unless you see Free memory go to zero(or close to it), its not out of
memory. You still have well over 1GB of free memory. As for virtual
memory, there is compressed swap in ram, which could easily account for
that.
Anyway, let me ask to verify.
Playing local FLAC music to the Boom, will
What I don't understand about the memory: The file system looks like
there are 2.6 G of virtual file system in memory, which is not possible,
and which would not leave 1.9 G of memory for the processes, so in some
way, the numbers need interpretation. I wrote "out of memory" because
that is my
What are the connected clients? Are you trying to play them
synchronized? What is the network connectivity of the server and the
clients?
piCorePlayer a small player for the Raspberry Pi in RAM.
Homepage: https://www.picoreplayer.org
Please 'donate'
The handling of streaming MP4 (i.e. AAC in MPEG-4) is a new chunk of
code in LMS. When processing MPEG-4, the audio data atom could be one
chunk (if not in streamed format) which has to be broken up into smaller
ADTS frames which are then passed to player or faad.
I'm not sure how the Tidal
For a few days, now, I have been running a piCorePlayer 6.1.0 with LMS
on a Pi 4 B with 2G of memory. I mainly use the Pi to transcode AAC
streams from Tidal. There are no additional plug-ins installed.
I observe the following behaviour: After the LMS is started, it will run
without complaints
101 - 180 of 180 matches
Mail list logo
