Tvheadend

All I can suggest is to upgrade to a newer version of tvheadend, and see if that works.

I've been getting very similar problems on the most up-to-date code, and haven't had a chance to isolate it yet (too busy using the machine!). 3.9.1987~gc053acd is very stable for me, and I downgraded back to it yesterday after missing some stuff last night.

And if that fails... there's always get_iplayer

What's interesting is that I have a DiBcom 7000PC adapter as well (an old Hauppauge stick - WinTV-Nova-T, something like that?), although it's only there as an emergency DVB-T stick - most of my stuff is routed through PCTV DVB-T2 and -S2 sticks. I can't immediately say that it's related to the 'time missed' issue, as I can't see how the application would make a difference, but...

Thanks for that. I already use get_iplayer but one of my reasons for running tvheadend is that its future is uncertain. The BBC are introducing a new API designed to defeat unauthorised clients and the devs are unsure whether they'll be able to work around it.

Most of the problems are with BBC4 yet it sometimes records programmes from that channel perfectly, and it occasionally fails on other channels. A pain, as much of what we want to record is on BBC4. I caught it in the act last night, trying to get it to record "The Secrets of Quantum Physics" as it was being broadcast. It failed every time with "waiting for stream".

When it fails tvheadend itself appears still to be running: it responds normally to the UI but, once failed, refuses to record anything until I reboot. The fact that a reboot is required rather than a service restart makes me suspect that the fault has temporarily broken the stick's firmware, and I had considered buying a different adaptor not using the Dibcom chipset.

Have you experienced this problem with both your sticks? Please report back if you find 3.9.1987~gc053acd fixes your problem - I'll give it a try too.

Regards, Pete

I'll monitor it. What confuses me is that - even if this stick dies - it should surely just go and find a stream from another tuner? I have three DVB-T(2) sources, plus some merged channels onto DVB-S (e.g. BBC Four and BBC FOUR). You might be right, though... maybe what's changed is that tvh used to find a spare tuner and now doesn't, so the symptom is that the recording fails when I'm not seeing the different behaviour. I'll try to do some testing.

3.9.1987 certainly works for me, and works every time, BUT subject to the comment above... it may be that it's simply ignoring the DiBcom tuner. Time for some syslog grepping, I think...

Inconclusive. I'll need to do some more detailed testing and tracing, I think.

An example from Monday evening...

I can see a recording starting and concluding on a DVB-S2 tuner (POP!), and that worked fine.

The next recording (Freeview BBC ONE HD) hits one of my DVB-T2 tuners - not the DiBcom DVB-T one as that's not HD - and that fails, 'no transponder available'.

I then manually tried to start the recording on DVB-S2 (Freesat BBC One HD) and the problem continued - no recording started, 'no transponder available'. It then continues cycling through all tuners until I downgraded to 3.9.1987 and rebooted, and then the DVB-S2 events kicked into life while the DVB-T2 ones were cancelled with 'time missed'.

Thanks for the info. I've upgraded to 3.9.2275~g066b273~trusty which is working at the moment. However as failures usually occur on BBC4 at night it will be at least a day before I can report back. As failures are intermittent it would take several days of reliable operation before I'll start being confident.

I do wonder if there is some subtle issue to do with programme metadata as the failures are associated with particular TV channels. I've noticed that our PVR fails to record programmes (very rarely indeed) which could be down to metadata. I'd forgive tvheadend if failures were equally rare and if they didn't render it inoperative until I reboot.

When you experience these failures do you need to reboot to regain any ability to record?

Regards,Pete

I'd need to check... TBH, when I do get them, I've normally re-installed tvh back to a known good version and then all is good - I don't recall if that's normally a reboot or only sometimes, it's certainly not a necessity. I do only seem to have problems when I'm venturing back towards the more up-to-date commits - so my Monday failure was on 3.9.2245~gd4fed88.

When I stop recording so much stuff for the kids, I'll try and do some structured testing...

Interesting!

I also had an issue with one of my Hauppauge tuners, below is what I currently have & had in my machine.

Devices:
  Current:
    - DVBSky T9850 (DVB S2 + T2) PCIE
    - Tevii S471 (DVB S2) PCIE
    - PCTV Nanostick 290e (USB)
    - Hauppauge NOVA-TD (USB)

  Removed:
    - Happauge NOVA-T (USB) 

Setup:
  - HP Microserver N36L
  - 3 GB RAM
  - Debian Wheezy x64
  - Tvheadend 3.9.2257~g948de59
  - Vanilla 3.18.0 Kernel

Whenever the NOVA-T was connected with the other usb tuners it would constantly fail within 24 hours reporting I2C read/write errors in dmesg unfortunately the bloody thing would also take out the others (usb only), Tvheadend then gobbles memory like a pig before finally conking out (hangs), the crazy memory consumption only started happening recently though.

I'd also tried the NOVA-T without the TD connected and although it lasted longer than before it still failed after a few days, yet I'd connected the NOVA-T to a Pi on its own, randomly making recordings and scanning muxes.. It didn't fail once, I'd left it running for a week as well. :s

The tuners in the current list have been running without issues for a while now without the NOVA-T connected.

I'm honestly not sure what to make of it myself, it's very strange.
---
The above may not help at all, but I wanted to let you guys know you're not the only ones suffering!

It does seem as if there may be two different issues here, with you (Prof Yaffle and Mark Clarkstone) having multiple tuners connected whereas I have a single USB device, albeit with a dual tuner internally. In my testing I've not encountered any problems with tvheadend using the dual tuners: the fault can occur on either tuner, and can also occasionally occur on the first recording after a reboot. Your problems appear to be around tvheadend arbitrating between multiple USB devices.

I recorded a lot of programmes last night without error so it's just possible that using the "unstable" 3.9.2275~g066b273~trusty may have fixed my problem. I'm normally wary of installing bleeding edge code. In any event it'll be several days before I can be at all confident the problem is fixed.

I do like the improvements to the already excellent web UI! I think tvheadend is brilliant software if only it delivered reliability. I plan to bung them a donation if it works for a month.

Thanks both for the feedback. I'll report back if it fails or (after a few days) if it doesn't.

Regards, Pete

You may well be right - I've got a trace running on my system now, trying to find what's giving me grief (Sod's law, 3.9.2280~ga67c6bb is completely, utterly and unshakeably stable at the moment, despite having recorded an entire afternoon or random crud while trying to break it).

See how you get on, I guess. I know that's not really helpful, but if we get to a stage that a problem is repeatable then we can debug it and see... and if it's not apparent any more, then it's clearly not really a problem!

Feedback/suggestions from Jaroslav/perexg based on my log file:

1. Disable idle scanning on the tuners - it's not necessary, since a service scan happens every time you tune to a mux anyway.

2. Enable power saving on the tuners, so they're kicked into life when needed.

3. Enable 'restart on error', so the tuners don't just sit there, sulking, when there's a problem.

I've been stable now for 12+ hours, including a good few hours of overnight quiescence. I'll see how it runs for the rest of the day, but it's looking promising for the moment.

I still can't say what changed between my stable 1987 version and the later revisions, but hey - that's life on the edge. I'll report back as and when I've something to say.

Oh, and a note to all: ~18 hours of trace +all ran to 69GB and over 1,000,000,000 lines. Be prepared with some serious split and sed-fu if you ever go down that road...

This may be totally unrelated but I got the same "time missed" messages for BBC3 HD in the last 3 weeks. When I look for reasons I find "Missing in PAT" for the BBC THREE HD and BBC NEWS services. No such messages for BBC ONE HD or even for BBC 3 (SD).
Rescanning the muxes does not progress, the scan queue length for 28.2E stays at 91 muxes... hm.
Live viewing and recording of mus channels (except BBC3HD) works. 3.9.2217 and upgraded to 2300 this morning.

Prof Yaffle wrote:

Feedback/suggestions from Jaroslav/perexg based on my log file:

1. Disable idle scanning on the tuners - it's not necessary, since a service scan happens every time you tune to a mux anyway.

2. Enable power saving on the tuners, so they're kicked into life when needed.

3. Enable 'restart on error', so the tuners don't just sit there, sulking, when there's a problem.

I've been stable now for 12+ hours, including a good few hours of overnight quiescence. I'll see how it runs for the rest of the day, but it's looking promising for the moment.

I still can't say what changed between my stable 1987 version and the later revisions, but hey - that's life on the edge. I'll report back as and when I've something to say.

Oh, and a note to all: ~18 hours of trace +all ran to 69GB and over 1,000,000,000 lines. Be prepared with some serious split and sed-fu if you ever go down that road...

I've tried these options before (enabling / disabling each one) and non seemed to have helped at all, the tuner lasted longer but it still died in the end.

On an unrelated note I have tried to use my spare RTL2838 stick but for some unknown reason it insists on being top-dog and ignoring any adapter_nr settings screwing my config up.

Did you use modinfo to see if the module supports adapter_nr, or is it just ignoring it?

I seem to remember having to flounder around before - on the em28xx driver, for example, the adapter_nr applies to em28xx_dvb and not the em28xx module itself. I think em28xx is a wrapper for a couple of modules, one for the DVB tuner and one for the remote control, something like that...

tvh is still behaving itself for the moment, though.

The "unstable" build I mentioned above has performed faultlessly since I installed it. Thanks for suggesting it. I hope you guys find fixes for your problems.

Two days stable here now... changing the tuner parameters as above seems to have cured my problem for the moment, at least...

In my case it was a permission problem.
I'm using HTS Tvheadend 3.9.1901~g43e805d~trusty
To fix the problem I added a new Profile Name in Configuration/Recording.
In the profile I defined a Recording System Path to a new folder in /home/hts and changed permission on that folder, to allow others to Create and delete files.

mega bump! - just to prove I AM searching for answers before asking

I too am suffering with this 'time missed' error on recording SO will make the changes suggested above.

However, I suspect the problem is file permissions. I have changed the location of the recording directory to a mounted network drive (to allow more space). I had successfully (partly) recorded something before changing the directory location but that failed when it ran out of space, I've only an 8GB SD card in the TVH Pi and didn't realise that recordings would be quite so large - hence (long story short) file permissions are a primary suspect.

I'll let you know as I'm sure you're all interested

Why not attached a decent capacity USB stick to the RPI and just supply the correct path to USB. Once you attach the USB it will be auto mounted so no concerns about permissions.

Jonas Lang wrote:

Why not attached a decent capacity USB stick to the RPI and just supply the correct path to USB. Once you attach the USB it will be auto mounted so no concerns about permissions.

I would if I had one. I'm averse to buying more 'stuff' in truth, particularly when I'm skint

Andrew N wrote:

Jonas Lang wrote:

Why not attached a decent capacity USB stick to the RPI and just supply the correct path to USB. Once you attach the USB it will be auto mounted so no concerns about permissions.

I would if I had one. I'm averse to buying more 'stuff' in truth, particularly when I'm skint

Afraid I can’t help you on the lack of funds front. USB sticks are as cheap as chips these days. You could try reading this as an alternative though

https://tvheadend.org/boards/5/topics/46516

The majority of issues in respect of recording destinations relate to incorrectly defined paths to destination.

Jonas Lang wrote:

Andrew N wrote:

Jonas Lang wrote:

Why not attached a decent capacity USB stick to the RPI and just supply the correct path to USB. Once you attach the USB it will be auto mounted so no concerns about permissions.

I would if I had one. I'm averse to buying more 'stuff' in truth, particularly when I'm skint

Afraid I can’t help you on the lack of funds front. USB sticks are as cheap as chips these days. You could try reading this as an alternative though https://tvheadend.org/boards/5/topics/46516

The majority of issues in respect of recording destinations relate to incorrectly defined paths to destination.

Jonas Lang wrote:

The majority of issues in respect of recording destinations relate to incorrectly defined paths to destination.

Thanks for the link. It ought help with what I'm doing and trying to achieve.

I've got the path right (stats on dashboard reflect it) and I saw tvheadend was running as hts (though can be changed to root) so figured the issue was permissions, so I've been playing with specifying uid & gid etc. I've reverted back to the default recording dir for now but that link gives me more things to try

Right now, I'm getting a "No input source available for subscription" error and I've switched the debug log on to see what that's about. I'll get there...

Edit: you're right about the price of USB sticks though. Just had a look and blimey they're way cheaper than I remember a few years back (even here in France) You can get a bucket full of them from China for tuppence!!