Oh, this is very interesting. It looks like the issue might be related to output.file or even output.file with the opus codec. I boosted the log level up to 5, and when ls behaves well and goes back to pulseaudio without a problem, it acts this way:
https://paste.xogium.me/P_.txt If I do it again however and it decides to bug, ls goes crazy like this in an endless loop. Also in mpv I hit a ton of audio underruns: https://paste.xogium.me/Pj.txt Have you any ideas why this happens and how to fix it? On Wed May 31, 2023 at 9:13 PM CEST, Romain Beauxis wrote: > Hi! > > Do you have some logs around the issue? Also to confirm: you're running > 2.2.x, correct? > > Le mer. 31 mai 2023 à 04:02, Tyna Pelletier-Bilodeau <ecole.t...@gmail.com> > a écrit : > > > Yeah, alright there's definitely something fishy somewhere here, I just > > don't know what that might be. After a while of using this insert_track > > and such, I will get spammed by source_fail on a regular basis, over and > > over, and no going back from the live stream to pulseaudio is possible. > > Really strange issue. > > > > On Sat May 27, 2023 at 4:45 PM CEST, Romain Beauxis wrote: > > > I think that this is because your input is pulseaudio so it doesn't have > > > tracks. In this case, you also need to insert a track mark to give the > > > append operator a chance to insert the blank. > > > > > > Something like this: > > > > > > binaural = insert_metadata(binaural) > > > insert_track = fun () -> binaural.insert_metadata(new_track=true, []) > > > > > > Then, later: > > > insert_track() > > > silence_source.skip() > > > > > > I just tested it locally and it worked! > > > > > > -- Romain > > > > > > Le mar. 23 mai 2023 à 02:31, Tyna Pelletier-Bilodeau < > > ecole.t...@gmail.com> > > > a écrit : > > > > > > > Of course, I will share everything needed here, even the effects I > > > > applied to the live stream in case they are useful in finding what's > > > > wrong. > > > > > > > > binaural = input.pulseaudio(device = > > > > "alsa_input.usb-ZOOM_Corporation_F3-00.analog-stereo") > > > > > > > > live = > > > > > > input.harbor("harbor",port=8005,user="username",password="password",icy=true) > > > > > > > > def multiband_band(live, from, _to, attack, release, ratio, threshold, > > > > gain) > > > > high_live = filter.iir.eq.high(live, frequency = from,) > > > > live_filtered = filter.iir.eq.low(high_live, frequency = _to,) > > > > live_compressed = ladspa.tap_limiter(limit_level = -0.5, > > > > ladspa.sc4(rms_peak = 0.5, attack_time = attack, release_time = > > release, > > > > threshold_level = threshold, ratio=ratio, makeup_gain = gain, > > > > live_filtered)) > > > > live_compressed > > > > end > > > > live = ladspa.gate(live, threshold = -80.0, attack = 0.15, hold = 1.0, > > > > decay = 200.0, range = -25.0) > > > > live = normalize(gain_min = -16., gain_max = 0., target = 0.5, live) > > > > live = ladspa.sc4(rms_peak=0.3, attack_time = 0.5, release_time = 3., > > > > threshold_level = -36.0, ratio=1., makeup_gain = 6., live) > > > > live = ladspa.tap_equalizer(band_1_freq=70., band_1_gain=4., live) > > > > live = ladspa.tap_equalizer(band_1_freq=150., band_1_gain=-2., live) > > > > live = add(normalize=false, > > > > [ > > > > multiband_band(live, 0., 200., 3.5, 30., 3., -10., 2.), > > > > multiband_band(live, 200., 1000., 2.25, 40., 2., -13., 2.), > > > > multiband_band(live, 1000., 3500., 2.25, 40., 3., -9., 2.), > > > > multiband_band(live, 3500., 6500., 2.25, 60., 2., -6., 1.5), > > > > multiband_band(live, 6500., 20000., 2.25, 70., 2., -4., 1.) > > > > ]) > > > > live = ladspa.tap_limiter(limit_level = -0.5, live) > > > > > > > > # Set to `true` when we should be adding > > > > # silence > > > > should_append = ref(false) > > > > > > > > # Append 10. of silence when needed. > > > > silence_source = append( > > > > binaural, fun (_) -> > > > > if should_append() then > > > > should_append := false > > > > blank(duration=10.) > > > > else > > > > source.fail() > > > > end > > > > ) > > > > > > > > # Transition to live > > > > def to_live(binaural, live) = > > > > sequence([binaural,live]) > > > > end > > > > > > > > # Transition back to playlist > > > > def to_binaural(live, binaural) = > > > > # Ask to insert a silent track. > > > > should_append := true > > > > > > > > # Cancel current track. This will also set the playlist > > > > # to play a new track. If needed, `cancel_pending` can > > > > # be used to for a new silent track without skipping the > > > > # playlist current track. > > > > silence_source.skip() > > > > > > > > sequence([live, binaural]) > > > > end > > > > > > > > source = fallback( > > > > track_sensitive=false, > > > > transitions=[to_live, to_binaural], > > > > [live, silence_source] > > > > ) > > > > > > > > I stream to icecast in mp3.cbr with a 192 kbps bitrate, in case this > > > > matters. > > > > > > > > I also see the blanks coming on and going, and the append track being > > > > played, except it doesn't play any silence and just goes back to > > pulseaudio. > > > > > > > > Hope this helps! > > > > > > > > On Mon May 22, 2023 at 11:12 PM CEST, Romain Beauxis wrote: > > > > > No it should be the same. Could you share a little more of your > > script? > > > > > > > > > > Le lun. 22 mai 2023 à 09:20, Tyna Pelletier-Bilodeau < > > > > ecole.t...@gmail.com> > > > > > a écrit : > > > > > > > > > > > Hi, > > > > > > I tried with a git pull on the master branch. Should I switch to > > the > > > > > > 2.2.x branch? > > > > > > > > > > > > On Mon May 22, 2023 at 3:46 PM CEST, Romain Beauxis wrote: > > > > > > > Hi, > > > > > > > > > > > > > > Have you tried with the latest rolling release code: > > > > > > > > > > > > > > > > > > > https://github.com/savonet/liquidsoap/releases/tag/rolling-release-v2.2.x > > > > > > > > > > > > > > Le dim. 21 mai 2023 à 15:14, Tyna Pelletier-Bilodeau < > > > > > > ecole.t...@gmail.com> > > > > > > > a écrit : > > > > > > > > > > > > > > > After I used the example provided in the cookbook to append a > > > > silence > > > > > > > > track, I get a whole bunch of these while simulating a timeout > > > > from > > > > > > > > harbor by disconnecting the vpn while the phone was streaming, > > and > > > > > > > > there's no silence playing, it immediately switches to > > pulseaudio: > > > > > > > > > > > > > > > > 2023/05/21 22:02:16 [fail_0:3] Content type is > > {audio=pcm(stereo)}. > > > > > > > > > > > > > > > > What does that mean, exactly? Should I be doing something for > > the > > > > > > > > source.fail() portion of the code? > > > > > > > > > > > > > > > > Thanks for your work, by the way! > > > > > > > > > > > > > > > > On Sun May 21, 2023 at 6:43 PM CEST, Romain Beauxis wrote: > > > > > > > > > Le sam. 20 mai 2023 à 23:08, Romain Beauxis < > > > > > > romain.beau...@gmail.com> a > > > > > > > > > écrit : > > > > > > > > > > > > > > > > > > > > > > > > > > > > > Le sam. 20 mai 2023 à 13:02, Tyna Pelletier-Bilodeau < > > > > > > > > ecole.t...@gmail.com> > > > > > > > > > > a écrit : > > > > > > > > > > > > > > > > > > > >> Hi Romain, > > > > > > > > > >> thank you, I will try this and see how it goes. > > > > > > > > > >> > > > > > > > > > >> By the way I've been toying a bit with harbor, and > > wondered if > > > > > > there > > > > > > > > was > > > > > > > > > >> a way to have liquidsoap waiting a little in case the live > > > > drops > > > > > > out > > > > > > > > > >> for a few seconds instead of immediately swtching back to > > the > > > > main > > > > > > > > > >> source? > > > > > > > > > >> > > > > > > > > > >> I stream to the input over wireguard, which means > > sometimes > > > > I'm > > > > > > > > roaming > > > > > > > > > >> to get a signal from 4g towers. I'd like if I could stream > > > > some > > > > > > > > silence > > > > > > > > > >> instead of immediately falling back to the main source, > > which > > > > is > > > > > > > > another > > > > > > > > > >> live done with input.pulseaudio, and which is jarring > > since > > > > the > > > > > > > > > >> transition happens out of nowhere. Maybe up to say 10 > > seconds > > > > of > > > > > > delay > > > > > > > > > >> before giving up and switching to the main source, is this > > > > > > possible? > > > > > > > > > >> > > > > > > > > > > > > > > > > > > > > Yes, it's a pretty legit use-case. I'm sure it's possible > > right > > > > > > now but > > > > > > > > > > might be a little tricky to setup. I'll think about what we > > > > can do > > > > > > to > > > > > > > > make > > > > > > > > > > this easy to setup. > > > > > > > > > > > > > > > > > > > > > > > > > > > > I have a PR making a couple of tweaks to make it possible: > > > > > > > > > https://github.com/savonet/liquidsoap/pull/3098 > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > >> On Sat May 20, 2023 at 2:18 AM CEST, Romain Beauxis wrote: > > > > > > > > > >> > Hi! > > > > > > > > > >> > > > > > > > > > > >> > Le ven. 19 mai 2023 à 02:39, Tyna Pelletier-Bilodeau < > > > > > > > > > >> ecole.t...@gmail.com> > > > > > > > > > >> > a écrit : > > > > > > > > > >> > > > > > > > > > > >> > > Hi, > > > > > > > > > >> > > it looks like after a connection timeout with > > liquidsoap > > > > > > 2.1.4, it > > > > > > > > > >> can't > > > > > > > > > >> > > automatically retry to reconnect to icecast. I'm not > > sure > > > > > > weather > > > > > > > > it's > > > > > > > > > >> > > because I use a tls enabled port or not, but in any > > event > > > > this > > > > > > > > > >> shouldn't > > > > > > > > > >> > > be happening. It is perfectly able to try > > reconnecting if > > > > for > > > > > > > > example > > > > > > > > > >> > > the icecast server has a too low maximum number of > > sources > > > > > > > > allowed, > > > > > > > > > >> but > > > > > > > > > >> > > when it is already streaming, and out of nowhere > > there's a > > > > > > > > problem, it > > > > > > > > > >> > > seems totally unable to do anything for itself. > > > > > > > > > >> > > > > > > > > > > > >> > > 2023/05/13 18:06:28 [asmrpmp3:3] Connection setup was > > > > > > successful. > > > > > > > > > >> > > 2023/05/19 08:12:42 [asmrpmp3:2] Error while sending > > data: > > > > > > could > > > > > > > > not > > > > > > > > > >> > > write data to host: connection timeout! > > > > > > > > > >> > > 2023/05/19 08:12:42 [asmrpmp3:3] Closing connection... > > > > > > > > > >> > > 2023/05/19 08:12:54 [clock.pulseaudio:2] Source > > asmrpmp3 > > > > > > failed > > > > > > > > while > > > > > > > > > >> > > streaming: could not close connection: SSL > > connection() > > > > error: > > > > > > > > > >> > > error:00000000:lib(0)::reason(0)! > > > > > > > > > >> > > 2023/05/19 08:12:54 [clock.pulseaudio:2] Raised at > > > > Cry.close > > > > > > in > > > > > > > > file > > > > > > > > > >> > > "src/cry.ml", line 193, characters 11-34 > > > > > > > > > >> > > 2023/05/19 08:12:54 [clock.pulseaudio:2] Called from > > > > > > > > > >> > > Icecast2.output#icecast_stop in file "outputs/ > > icecast2.ml > > > > ", > > > > > > line > > > > > > > > 596, > > > > > > > > > >> > > characters 14-34 > > > > > > > > > >> > > 2023/05/19 08:12:54 [clock.pulseaudio:2] Called from > > > > > > > > > >> > > Icecast2.output#icecast_send in file "outputs/ > > icecast2.ml > > > > ", > > > > > > line > > > > > > > > 509, > > > > > > > > > >> > > characters 10-27 > > > > > > > > > >> > > 2023/05/19 08:12:54 [clock.pulseaudio:2] Called from > > > > > > > > > >> > > Output.encoded#send_frame.output_chunks.f in file > > > > "outputs/ > > > > > > > > output.ml > > > > > > > > > >> ", > > > > > > > > > >> > > line 240, characters 10-24 > > > > > > > > > >> > > 2023/05/19 08:12:54 [clock.pulseaudio:2] Called from > > > > > > > > > >> > > Output.output#output in file "outputs/output.ml", > > line > > > > 174, > > > > > > > > > >> characters > > > > > > > > > >> > > 10-35 > > > > > > > > > >> > > 2023/05/19 08:12:54 [clock.pulseaudio:2] Called from > > > > > > > > > >> > > Clock.clock#end_tick.(fun) in file "clock.ml", line > > 300, > > > > > > > > characters > > > > > > > > > >> > > 14-22 > > > > > > > > > >> > > 2023/05/19 08:12:54 [clock.pulseaudio:2] > > > > > > > > > >> > > 2023/05/19 08:12:54 [asmrpmp3:3] Closing connection... > > > > > > > > > >> > > 2023/05/19 08:12:54 [clock:2] Error when leaving > > output > > > > > > asmrpmp3: > > > > > > > > > >> could > > > > > > > > > >> > > not close connection: SSL connection() error: > > > > > > error:0A000126:SSL > > > > > > > > > >> > > routines::unexpected eof while reading! > > > > > > > > > >> > > 2023/05/19 08:12:54 [main:3] Shutdown started! > > > > > > > > > >> > > > > > > > > > > > >> > > Any ideas what I can do to mitigate this issue? I'm > > > > thinking > > > > > > of > > > > > > > > > >> checking > > > > > > > > > >> > > out the latest 2.2.x but I don't know if that has been > > > > solved > > > > > > or > > > > > > > > not, > > > > > > > > > >> if > > > > > > > > > >> > > it was even known to begin with, but I don't know how > > I > > > > could > > > > > > try > > > > > > > > this > > > > > > > > > >> > > out if I'm using opam. > > > > > > > > > >> > > > > > > > > > > > >> > > > > > > > > > > >> > This looks like a leak in the resources initializations. > > > > > > > > > >> > > > > > > > > > > >> > I would definitely try with `2.2.x`. The SSL stack was > > > > entirely > > > > > > > > > >> rewritten > > > > > > > > > >> > there and some resource cleaning issues were fixed. And > > we > > > > also > > > > > > > > have TLS > > > > > > > > > >> > support now so you have two stacks to try if one still > > has > > > > > > issues. > > > > > > > > > >> > > > > > > > > > > >> > To use SSL/TLS in v2.2.x you can now pass an abstract > > > > transport > > > > > > > > > >> variable. > > > > > > > > > >> > This is explained here: > > > > > > > > > >> > > > > > > > https://www.liquidsoap.info/doc-dev/harbor_http.html#https-support > > > > > > > > > >> > > > > > > > > > >> > > > > > > > > > >> > > > > > > > > > >> _______________________________________________ > > > > > > > > > >> Savonet-users mailing list > > > > > > > > > >> Savonet-users@lists.sourceforge.net > > > > > > > > > >> > > https://lists.sourceforge.net/lists/listinfo/savonet-users > > > > > > > > > >> > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > _______________________________________________ > > > > > > > > Savonet-users mailing list > > > > > > > > Savonet-users@lists.sourceforge.net > > > > > > > > https://lists.sourceforge.net/lists/listinfo/savonet-users > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > _______________________________________________ > > > > > > Savonet-users mailing list > > > > > > Savonet-users@lists.sourceforge.net > > > > > > https://lists.sourceforge.net/lists/listinfo/savonet-users > > > > > > > > > > > > > > > > > > > > > > _______________________________________________ > > > > Savonet-users mailing list > > > > Savonet-users@lists.sourceforge.net > > > > https://lists.sourceforge.net/lists/listinfo/savonet-users > > > > > > > > > > > > _______________________________________________ > > Savonet-users mailing list > > Savonet-users@lists.sourceforge.net > > https://lists.sourceforge.net/lists/listinfo/savonet-users > > _______________________________________________ Savonet-users mailing list Savonet-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/savonet-users
