Author Topic: FPP 1.9 Longevity Test  (Read 419 times)

Offline patdelaney

  • Administrator
  • *****
  • Join Date: Jun 2014
  • Location: Madison, WI
  • Posts: 1,028
  • Kudos: 9
FPP 1.9 Longevity Test
« on: April 16, 2017, 06:23:16 AM »
I think there may be a bug in 1.9 that stops a running playlist when it crosses over midnight on Saturday nights. But I can't prove it yet.


A couple of weeks ago I decided to upgrade a few of my 2x2 P10 matrixes I have. I was running some test and left them on over night. The next day I checked and they were not running. So I checked to see I had some sort of schedule to stop them still remaining. I didn't see any active schedules.


So last Sunday I started 3 of my 2x2 matrix playing a loop. They ran all week with no problem. This morning I want to check on them and 2 of them were stopped.


BB1 - Beaglebone running 1.5 Master  was still running
BB4 - Beaglebone running 1.9 was stopped
FPPv3-1 - Raspberry Pi running 1.9 was stopped.


fppd was still running according to the UI


I don;t see anything in the logs that show why it would have stopped.


Thoughts?


Pat





Offline CaptainMurdoch

  • Administrator
  • *****
  • Join Date: Sep 2013
  • Location: Washington
  • Posts: 7,851
  • Kudos: 139
Re: FPP 1.9 Longevity Test
« Reply #1 on: April 17, 2017, 01:45:59 PM »
Were these scheduled or manually started?  You can't have a schedule cross a midnight boundary.  If they were manual, I haven't tested having a manual playlist cross a weekend boundary, but believe it should work.
-
Chris

Offline patdelaney

  • Administrator
  • *****
  • Join Date: Jun 2014
  • Location: Madison, WI
  • Posts: 1,028
  • Kudos: 9
Re: FPP 1.9 Longevity Test
« Reply #2 on: April 17, 2017, 02:33:16 PM »
These were all started manually. I started all of them Sunday Morning so I'll let them run again through the week. I'm sure I could turn of NTP and change the date, but I'm in no hurry.  I'll let you know Sunday Morning.


Pat

Offline patdelaney

  • Administrator
  • *****
  • Join Date: Jun 2014
  • Location: Madison, WI
  • Posts: 1,028
  • Kudos: 9
Re: FPP 1.9 Longevity Test
« Reply #3 on: April 23, 2017, 09:28:45 AM »
Longevity Test update for 1.9....


The 2 units running 1.9 were no running this morning wne I wnt do to the basement to check them. The BB1 running (
Version:
[/size][/color][/size]
  Host:
[/size][/color]) was still running. So we have a crossing over Saturday bug.


Pat

Offline CaptainMurdoch

  • Administrator
  • *****
  • Join Date: Sep 2013
  • Location: Washington
  • Posts: 7,851
  • Kudos: 139
Re: FPP 1.9 Longevity Test
« Reply #4 on: May 04, 2017, 10:16:11 AM »
The playlist is stopping in the released versions of the code because the 'weekly seconds' is rolling over back to zero.  There is an internal structure which is used to track the currently scheduled item including its end time.  This structure is initialized to all zeroes at program start.  The scheduler constantly checks this to see if the end time matches the current 'weekly seconds'.  When the weekly seconds rolls over to 0 at midnight Saturday into Sunday morning, then the scheduler thinks that it needs to stop a scheduled playlist and it stops it.  I think we could initialize the structure differently, but would have to think about it a little.  If you want to test a change, I can probably give you a line or two of code to insert to see if it fixes it.

The master code is probably working because there are parts of it that aren't hooked up yet.  You are using the new playlist code to play the playlist, but something is missing between the player and the scheduler.  I haven't made any changes yet to the scheduler to handle the weekend rollover.

 

Back to top