Wake - Sleep Interaction -> Failed Recordings

[ACTCMS]

I can't really help with your sleep problem - I'm a 24/7 servers in the attic with (silent) NMT extenders at each TV sort of chap (100% WAF) but I'm hoping that the following is not a portent for the future...

Was hoping there was a brute force solution (a BFH).

Replacement Wife not an option at this time]

Alex

[jam_zhou]

to keep the computer awake during auto wake (like when transcoding overnight) I use a program called insomnia. Maybe you can write a batch file that will keep insomnia running for 15 minutes at wake to ensure that the recording starts and the sleep timer doesn't cut back in.

http://dlaa.me/Insomnia/

[johnsonx42]

You're misunderstanding the problem. It's not that it wakes up for the recording and then goes back to sleep before it starts, it's that it's already awake at the scheduled wakeup time, but then sleeps between that time and the time the recording actually begins. (or in rare cases the computer will be just heading to sleep when the wakeup is scheduled, and it can't wakeup until it gets fully to sleep by which point the wakeup time has passed)

If Windows weren't so %$#^%^$%-ing stupid, this wouldn't happen. First, any scheduled wakeup event should automatically reset the sleep timer if the computer was already awake. Second, before going to sleep Windows should check whether there are any scheduled wakeups in the next couple of minutes and cancel the sleep if so. But hey, sleep states are really new, Microsoft has only had 25 years or so to figure this stuff out.

Without changes to NPVR to deal with this (setting two wakes per recording), I suggest setting <WakeupPreRecodingSeconds> a small as possible (preferably 0) and using 1 or more minutes of pre-padding.

[sub]

Here is a version that schedules two wakeups per recording. The second one is 30 seconds later.

[jksmurf]

Awesome - thanks sub - can't wait to try - will try with 120 secs <WakeupPreRecordingSeconds> just to be sure.

[johnsonx42]

Awesome - thanks sub - can't wait to try - will try with 120 secs <WakeupPreRecordingSeconds> just to be sure.

Actually I wouldn't set it longer than 30 seconds, or you'll defeat the purpose of the second wakeup.

@sub - rather than setting the second wake to 30 seconds after the first, it'd be better to set it to the actual recording start time, or perhaps 5 seconds before it. The problem is the window of time between the wakeup request and the actual start of the recording when NRecord starts hitting that ES_SYSTEM_REQUIRED call. <WakeupPreRecordingSeconds> defaults to 90; adding a second wakeup 30 seconds later merely reduces the time window down to 60 seconds.

[jksmurf]

Just reporting that so far, touch wood, I have had 3 recordings WORK without the dreaded "recording service was not running" message. Awesome.

I can supply logs if needed, but if this continues to function as it has, it is a MAJOR item for me and the WAF. No longer do I live in fear of having to bin the whole NPVR experience and return to the dark days of the DVR. Liberated from the shackles of tyranny! (etc etc).... thanks sub.

[mvallevand]

@sub - rather than setting the second wake to 30 seconds after the first, it'd be better to set it to the actual recording start time, or perhaps 5 seconds before it.

+1 for zero. I don't use pre-padding and I'd rather lose the 10 seconds or so wake takes on my backend then the entire recording.

Martin

[sub]

@sub - rather than setting the second wake to 30 seconds after the first, it'd be better to set it to the actual recording start time, or perhaps 5 seconds before it. The problem is the window of time between the wakeup request and the actual start of the recording when NRecord starts hitting that ES_SYSTEM_REQUIRED call.

I'm not sure there is any remaining scenario when the machine could sleep through a recording. I think the problem was when the machine was in the process of going to sleep when machine was due to wake for the start of pre-padding.

I think if it has the second wakeup 30 seconds later, you'll start seeing "Keeping machine from entering standby due to pending recording" messages in the log, and it wont go to sleep.

[mvallevand]

I'm not sure there is any remaining scenario when the machine could sleep through a recording. I think the problem was when the machine was in the process of going to sleep when machine was due to wake for the start of pre-padding.

I think if it has the second wakeup 30 seconds later, you'll start seeing "Keeping machine from entering standby due to pending recording" messages in the log, and it wont go to sleep.

In my use-case with no pre-padding, what I regulary do is use wol to check recordings or watch a bit of live sports. When it don't pay attention if I disconnect with less then 7 minutes and with the new change 5.5 minutes before the recording sleep cuts in after 5 minutes. The sleep time is irrelevant but I hope that explains. Even if it is 1 second, I'd have to pay attention. 5m 1s before the recording.

Martin