Second monitor goes black after machine reboot

1

1

We have some 4th generation Intel NUC machines that we've placed an SSD inside of and attached two of these planar TVs. Each machine is running Windows 8.1.

The problem we're running into is that when our software triggers a controlled reboot (the PowerShell Restart-Computer cmdlet), occasionally one of the displays comes up black. This always seems to be the secondary monitor. If we keep rebooting the machine, the TV does not come back up. Power cycling the TV seems to do nothing.

Right now we have 8 of the planar displays running against 4 of the Intel NUCs. There's no pattern as to when this will happen. On some days it happens, others it does not. It also is not correlated with any specific NUC or Planar TV.

We've tried swapping which display is plugged in for the HDMI / DisplayPort inputs and rebooting the machine. It comes back up fine but sure enough, some time later the display will go black again after a reboot. Swapping out the HDMI and/or DisplayPort cable for a new one does nothing to fix the problem.

If we try VNCing into the machine that's showing a black screen, we can see both displays within VNC, yet nothing on the physical display. If we log out of the currently logged in user and log back in, the display will start showing whatever content was supposed to be there.

To summarize:

  • The secondary monitor goes black, showing no output after some non-deterministic number of reboots.
  • Some machines have gone weeks without exhibiting the behavior, then show it every few days, and then stop doing it for several weeks.
  • After going black, no number of reboots fixes it
  • VNC shows both displays through it's virtualized display adapter
  • Logging out of the currently logged in user and logging in appears to fix it
  • There is no correlation between the failure rate and which TV is plugged in.
  • There is no correlation between the failure rate and whether the TV is plugged into HDMI or DisplayPort
  • In our software, we detect the number of displays and can actually see two monitors plugged in.

Any thoughts?

Event Log Analysis

I ran an event log query against the machine and I saw literally dozens of events saying that the driver crashed.

Here's the script:

Get-WinEvent `
    -ComputerName "Frobnitz" `
    -Credential "localhost\Administrator" `
    -LogName "System" `
    -FilterXPath "*[System[Provider[@Name='Display']]]" `
    | ? { 
        ($_.TimeCreated.TimeOfDay.Hours -ge 3) -and `
        ($_.TimeCreated.TimeOfDay.Hours -le 9)
    } | Select TimeCreated,Message | ft -AutoSize

And here's the output:

TimeCreated           Message
-----------           -------
10/05/2014 7:43:21 AM Display driver igfx stopped responding and has successfully recovered.
10/05/2014 7:42:41 AM Display driver igfx stopped responding and has successfully recovered.
10/05/2014 4:01:21 AM Display driver igfx stopped responding and has successfully recovered.
10/05/2014 4:01:16 AM Display driver igfx stopped responding and has successfully recovered.
10/05/2014 4:01:11 AM Display driver igfx stopped responding and has successfully recovered.
10/05/2014 3:01:10 AM Display driver igfx stopped responding and has successfully recovered.
10/05/2014 3:01:04 AM Display driver igfx stopped responding and has successfully recovered.
10/05/2014 3:00:50 AM Display driver igfx stopped responding and has successfully recovered.
10/05/2014 3:00:25 AM Display driver igfx stopped responding and has successfully recovered.
10/03/2014 3:01:01 AM Display driver igfx stopped responding and has successfully recovered.
10/03/2014 3:00:26 AM Display driver igfx stopped responding and has successfully recovered.
10/02/2014 3:00:39 AM Display driver igfx stopped responding and has successfully recovered.
10/02/2014 3:00:26 AM Display driver igfx stopped responding and has successfully recovered.
10/01/2014 3:00:36 AM Display driver igfx stopped responding and has successfully recovered.
10/01/2014 3:00:27 AM Display driver igfx stopped responding and has successfully recovered.
09/30/2014 8:02:33 AM Display driver igfx stopped responding and has successfully recovered.
09/30/2014 8:01:34 AM Display driver igfx stopped responding and has successfully recovered.
09/30/2014 8:01:29 AM Display driver igfx stopped responding and has successfully recovered.
09/30/2014 8:01:05 AM Display driver igfx stopped responding and has successfully recovered.
09/30/2014 8:00:43 AM Display driver igfx stopped responding and has successfully recovered.
09/29/2014 7:01:43 AM Display driver igfx stopped responding and has successfully recovered.
09/28/2014 3:00:25 AM Display driver igfx stopped responding and has successfully recovered.
09/27/2014 3:00:39 AM Display driver igfx stopped responding and has successfully recovered.
09/27/2014 3:00:28 AM Display driver igfx stopped responding and has successfully recovered.
09/26/2014 3:00:54 AM Display driver igfx stopped responding and has successfully recovered.

Mike Bailey

Posted 2014-10-06T19:35:05.497

Reputation: 191

The logout/login solution suggests the problem is related to software rather than hardware. Are all of the machines configured identically, running the same software, etc.? Is there any commonality in what is being processed before the reboot failure? Any chance it relates to non-graceful shutdown or momentary power issue (are they on a UPS), that leaves something in a corrupted state before reboot? – fixer1234 – 2014-10-06T20:05:22.680

Yep. All the same exact software set. They're based on the same image and we have our software automatically pushed down via some configuration on the machine. We've seen this even without our own software installed. I did some follow up research and it looks people are having issues with the Intel Graphics Drivers. I also saw some event log entries corresponding to the graphics driver crashing. – Mike Bailey – 2014-10-06T20:08:20.620

@fixer1234: WRT to the reboot, we use a controlled reboot process. All machines are hooked up to UPS units and the UPS units are plugged in to outlets that can switchover to generator power. We haven't had any power related events when this issue crops up. We've duplicated this issue in our lab environment with a manual reboot as well. – Mike Bailey – 2014-10-06T20:12:10.143

1Have you consulted an exorcist? – fixer1234 – 2014-10-06T20:12:37.253

The graphics driver issue looks like a promising avenue. – fixer1234 – 2014-10-06T20:14:40.053

@fixer1234 The exorcist was a dead end. – C Bauer – 2014-10-06T20:22:02.597

No answers