PC Powers on Randomly

[Stereodude]

So I built a backup server that is turned on every night at 2:00AM via Wake on Lan from my main server. That all works fine, but the backup server also turns on at random times as well.

I made a few batch files to allow me to log when Windows starts, when the shut down starts, and when the backup process starts and completes.

Here's a little over a week's worth of data.

Start Up Occurred: Sun 10/04/2009 2:01:04.73
Delay Started: Sun 10/04/2009 2:01:15.26
Backup Completed: Sun 10/04/2009 2:09:14.21
Shut Down Started: Sun 10/04/2009 2:09:46.95

Start Up Occurred: Mon 10/05/2009 2:01:06.42
Delay Started: Mon 10/05/2009 2:01:17.45
Backup Completed: Mon 10/05/2009 2:09:40.42
Shut Down Started: Mon 10/05/2009 2:10:13.14

Start Up Occurred: Mon 10/05/2009 6:30:37.89
Delay Started: Mon 10/05/2009 6:30:48.93
Backup Completed: Mon 10/05/2009 6:37:41.12
Shut Down Started: Mon 10/05/2009 6:38:13.92

Start Up Occurred: Tue 10/06/2009 2:01:06.59
Delay Started: Tue 10/06/2009 2:01:17.42
Backup Completed: Tue 10/06/2009 2:08:47.35
Shut Down Started: Tue 10/06/2009 2:09:20.31

Start Up Occurred: Wed 10/07/2009 2:01:08.59
Delay Started: Wed 10/07/2009 2:01:19.45
Backup Completed: Wed 10/07/2009 2:09:31.59
Shut Down Started: Wed 10/07/2009 2:10:04.32

Start Up Occurred: Wed 10/07/2009 3:50:45.04
Delay Started: Wed 10/07/2009 3:50:55.60
Backup Completed: Wed 10/07/2009 3:57:17.40
Shut Down Started: Wed 10/07/2009 3:57:50.23

Start Up Occurred: Wed 10/07/2009 3:59:07.04
Delay Started: Wed 10/07/2009 3:59:17.42
Backup Completed: Wed 10/07/2009 4:05:39.14
Shut Down Started: Wed 10/07/2009 4:06:12.07

Start Up Occurred: Thu 10/08/2009 2:01:09.54
Delay Started: Thu 10/08/2009 2:01:20.32
Backup Completed: Thu 10/08/2009 2:10:21.23
Shut Down Started: Thu 10/08/2009 2:10:54.18

Start Up Occurred: Thu 10/08/2009 6:18:53.57
Delay Started: Thu 10/08/2009 6:19:04.32
Backup Completed: Thu 10/08/2009 6:26:16.01
Shut Down Started: Thu 10/08/2009 6:26:48.67

Start Up Occurred: Fri 10/09/2009 2:01:11.23
Delay Started: Fri 10/09/2009 2:01:22.03
Backup Completed: Fri 10/09/2009 2:09:25.56
Shut Down Started: Fri 10/09/2009 2:09:58.57

Start Up Occurred: Fri 10/09/2009 7:31:30.09
Delay Started: Fri 10/09/2009 7:31:40.67
Backup Completed: Fri 10/09/2009 7:40:19.78
Shut Down Started: Fri 10/09/2009 7:40:52.87

Start Up Occurred: Sat 10/10/2009 2:01:00.14
Delay Started: Sat 10/10/2009 2:01:10.79
Backup Completed: Sat 10/10/2009 2:09:32.48
Shut Down Started: Sat 10/10/2009 2:10:05.37

Start Up Occurred: Sun 10/11/2009 2:01:02.81
Delay Started: Sun 10/11/2009 2:01:12.67
Backup Completed: Sun 10/11/2009 2:09:27.70
Shut Down Started: Sun 10/11/2009 2:10:00.65

Start Up Occurred: Mon 10/12/2009 2:01:03.50
Delay Started: Mon 10/12/2009 2:01:14.40
Backup Completed: Mon 10/12/2009 2:08:54.84
Shut Down Started: Mon 10/12/2009 2:09:27.78

Start Up Occurred: Mon 10/12/2009 6:10:34.48
Delay Started: Mon 10/12/2009 6:10:45.28
Backup Completed: Mon 10/12/2009 6:17:19.25
Shut Down Started: Mon 10/12/2009 6:17:52.23

Start Up Occurred: Mon 10/12/2009 7:31:19.64
Delay Started: Mon 10/12/2009 7:31:29.60
Backup Completed: Mon 10/12/2009 7:37:54.32
Shut Down Started: Mon 10/12/2009 7:38:27.31

There are 7 random extra power ons in there.

Now, the machine is on a UPS which makes my hypothesis all the more odd. I'm also 99% sure I have the machine set to not turn back on in the event of a power failure in the BIOS. I'll have to double check tonight when I get home.

However, my best guess is that the machine turns on during some AC power glitches. Examining the UPS log on my main server reveals that on the 10/7 AVR boost went active at 3:49:09AM, and the power failed at 3:49:16AM with the power being restored at 3:49:21AM. That same sequence occurred again at 3:58:01AM, 3:58:07AM, and 03:58:09AM. The backup server appears to have powered up for both of those, but there's nothing in the UPS log corresponding to any other random turn on events though I admittedly do not have the UPS's voltage tolerance set very tight.

I also know that on at least weekdays (maybe weekends too) there is a pretty serious power glitch at 7:30AM. I'm usually in my library during that time and without fail both UPS's in my library will momentarily trip at 7:30AM every day. Two of the extra power ons in the above log occur at 7:31AM and I know there were more 7:31AM Windows starts previously.

I guess ultimately in the grand scheme of things the random extra power ons don't matter, but the whole thing is very puzzling considering the system is on a UPS and turned off when these occur. :scratch:

 

[P5-133XL]

Check the power settings in CMOS setup. There are a variety of events that can typically power up a machine there: Things like network activity, or keyboard/mouse/USB activity can power up a machine.

 

[Bozo]

Are you sure the UPSs are not doing a self check?

 

[Stereodude]

Check the power settings in CMOS setup. There are a variety of events that can typically power up a machine there: Things like network activity, or keyboard/mouse/USB activity can power up a machine.

I double checked the BIOS today and all that stuff is turned off. It's set to stay off after a power failure. All the other power on options are disabled.

 

[Stereodude]

Are you sure the UPSs are not doing a self check?

The UPS data is from my APC Smart-UPS 700 XL connected to my main server. It does self checks at some interval (not sure what it is).

The backup server is plugged into a "dumb" APC BackUPS Pro 1100. My understanding is that it does not do any sort of regular self test.

 

[P5-133XL]

I would then tighten the UPS thresholds as the next step.

 

[Stereodude]

I would then tighten the UPS thresholds as the next step.

I can only do that on the UPS that feeds the main server, which will only improve my logging of power disturbances. :crap:

 

[P5-133XL]

Have you checked the log? If both machines are on the same circuit then the log (on the input side) should apply to both machines.

Other than the CMOS settings I don't know what would cause such behavior other than a general bios failure. Have you tried resetting the CMOS to default/optimized settings then make sure the power settings are totally off. If you recently did a bios upgrade without resetting CMOS the settings can dissociate from what the display says they are.

As a solution to your symptom, you can always use the power switch on the PS rather than the front power-on button. It really is not a cure but just a fix to the symptom.

 

[Stereodude]

Well I figured this one out.

The APC BackUPS Pro 1100 that powers the backup server also has a RS-232 / Serial connection to the PC. Whatever signal / message the UPS sends during a brownout or power glitch to the machine on the serial port causes the system to turn on. If WoL is disabled in the NiC properties in Windows, or the PC wasn't shut down gracefully (meaning the system won't respond to the magic packet) the messages from the UPS don't turn the system on. :arge:

I want the UPS to be able to communicate to the system so it can start a graceful shutdown in the case of the power going out while it is running a backup, so unplugging the serial cable isn't an option. It looks like I need to get a newer UPS with a USB connection.

 

[MaxBurn]

Interesting info on the wake by RS232.

Does the current UPS have SNMP capability? Could just use that for a graceful shutdown.

 

[Stereodude]

Does the current UPS have SNMP capability? Could just use that for a graceful shutdown.

No, it's a cheapy. It only has RS-232 with their bastardized pinout. Right now I'm leaning toward getting a APC SmartUPS with USB like I did for my main server.

 

[Handruin]

I recently bought this APC Back-UPS XS 1500VA and it connects via USB. Windows 7 shows it as a battery icon in the task tray just like you would see in a laptop. I set my power management via windows for when my machine is on battery to either shutdown or go into sleep mode after a certain number of minutes. With my Core i7 the estimated run time is about 45 minutes based on the front panel of the APC.

 

[Stereodude]

I should have bought two of these back when they were $149.99 instead of just one.

 

[Bozo]

Have you tried installing APC's free software? If i remember correctly, you could set the UPS parameters to what you need.

 

[Will Rickards]

I think it is the wake on mouse setting in the BIOS. I know you said they are disabled. But unless the bios/chipset is malfunctioning, it seems odd. What motherboard is this?

 

[Stereodude]

Have you tried installing APC's free software? If i remember correctly, you could set the UPS parameters to what you need.

No, I haven't. In the past when I tried it these APC BackUPS Pro 1000's didn't offer any functionality with their software that the included capability in Windows XP didn't. Besides, I'm not sure you can tell the UPS to not send messages when the power fails or sags since that would defeat the purpose.

 

[Stereodude]

I think it is the wake on mouse setting in the BIOS. I know you said they are disabled. But unless the bios/chipset is malfunctioning, it seems odd. What motherboard is this?

It's a Supermicro P4SCi. It has the most bare bones bios settings I've ever seen in any modern system.

I don't think it's Wake on mouse causing the problem (though there isn't a setting for that anyhow). I think it's Wake on Ring (also, no setting for it).

 

[Will Rickards]

I see what you mean about the bios based on the screen shots.
The only suspects I see are the Power On function in the I/O Device Configuration
And the Console Redirection settings.

 

[sechs]

I recently bought this APC Back-UPS XS 1500VA and it connects via USB.

FYI, the 1500VA and 1300VA models (and 1250VA model, if you can find it at your local chain store) use the same battery and are practically identical, except for that max load. The RS versions are exactly the same as their XS counterparts.

If you decide to go with one of these, you'd probably be fine with whatever is cheapest.