Topic: How to remotely force a "full reboot" (Read 893 times)

Personally I would fix the actual problem vs worrying about rebooting all the time. That sounds like it could only get worse in the long run catastrophic end result for that rig.

fix the underlying problem with the rig

If you are still having issues, here is a thought. It is not a method for the reboot, but a possible reason for the issues. What about changing to or from UEFI in the Bios settings. The worst thing that will happen is the need to change it back on a bad boot. I swear that is not the exact solution, but I think that your PC is fast booting and missing the proper set up on the way.

yep, we used to call this the COLD boot and HOT boot, now its just reset, and the power never 'really goes' off most system stay running,

thus to really 'reset' COLD-BOOT U must pull the plug either a double-relay on a rasberry pi type device, but certainly no way to cold-boot without pulling the plug especially with windows,

with all the spyware most devices like android are always on and listening  even when off, same for windows, they never really turn off unless you pull the power and ALL battery's,

another reason not to use windows, on linux it would be easy to do a full software cold-reset, but on windows, well windows 10.0 as they say is spy-ware

luckily the MOBO bios has lots of switches that windows can't control, so it sounds like you figured out how to get a REAL COLD BOOT without pulling the plug

Thanks all for the suggestions!
You know what, I actually managed to find a cure... A combination of things, I think. Forcing the proper shutdown mode in Windows, disabling all fast boot options... And waiting longer after cutting the power (30sec). Solved!

I use 8 of these Watts Clever INPLUG

I have that too. There is a USB device that overloads the computer when the port setup is silent, but these devices are expensive. There are even devices that using a mobile phone triggers the reset. In the settings of the motherboard there is a function on and off by the signal from the router. But it seems to me that if the computer froze then it can be reset only by mechanical means.

Easy mate
I use this and control all from my mobile, even i can check consumption .
easy and stable
https://www.aliexpress.com/item/Itead-Sonoff-Pow-Wireless-Intelligent-automation-module-Switch-WiFi-Smart-home-automation-Remote-Power-Consumption-Measurement/32816963005.html?spm=a2g0s.9042311.0.0.Rrey0U

Yeah, and what chipset is your mobo?

I realized the only reason why my rigs freeze is due to incomplete gpu initialization (something about failed opencl calls ) while starting miner(usually when miner restarts that happens - claymore miner) or when one or some of my gpus hang during mining after several days or a month of mining
and thats on a rig with asrock h81 btc v2.0 mobo and 6 gpus

my other rig with msi z170a pro and 4 gpus usually freezes on average a lot more frequently than h81 mobo,(the worst case scenario being a restart per day) the reason because z170 is purely class A shit chipset for mining...
and the only way to restart it is by cutting the power off for at least 30-40 sec and then turning the power back on
usually by turning the switch on the psu off (0) and after 30-40 sec back on (I)

even pushing reset switch doesn't do anything in this instance,
only sort of the alternative is holding the power button for several secs but only to turn the rig off,
you cant turn it back on, 30-40 secs is needed for the mobo to be completely power cut in order to power the rig back on - boot up.

anyways, I'm interested in trying out those double relay usb watchdogs but I kinda doubt they'll work since I don't believe they are able to cut power, only force power off.
but since they are 9€ on aliexpress I'll give them a try...

Out of curiosity, what motherboard brand/model is it that's acting flaky?  I don't know if a usb watchdog would fix the issue or not if it's bios related, but this is on the edge of my comfort zone too, so hopefully someone has an idea for you (short of replacing the motherboard).

simple rig resetter could help... but is costly

Any ideas?
It's definitely a bios/mobo issue. When I restart from windows, the board explorer in bios only sees 4 gpus. I need to turn off the rig and cold boot it for the bios to see all GPUs again.

F***ing annoying.

Actually the state after power loss was correctly set to "on".
There seems to be a problem with the mobo not recognising all GPUs unless a cold boot is initiated. Not sure wtf is wrong with it. I've now turned off "fast reboot" in the power options in Windows, fingers crossed that it'll help (I have doubts though).

What annoys me most is that cutting the power to the rig doesn't even force a cold boot after the power is restored. Not sure what can be done there.

That one's pretty sweet. I reckon that's exactly what I'd need.
On a second thought, I'm gonna check the bios settings of that rig. I'm suspecting I might have set "state after power loss" to "last known" instead of "on". That would explain everything. If not I'm gonna have to think about a "tactical solution" (aka fudge)...

https://www.ebay.com/itm/Digital-Loggers-LPC-3-Remote-Web-Network-Power-Switch-Rebooter-8-Outlets-/332414301594?hash=item4d656fb59a:g:KxsAAOSw~R1ZyF7V

I love these.  One needed for 2 rigs.

This one for instance, the double relay option, has the ability for force power off by holding the power button for 6s and then restart

https://www.aliexpress.com/item/USB-Dongle-Watchdog-Watch-Dog-for-Mining-Miner-Rig-Unattended-Operation-Crash-Auto-Recover-Reboot/32827885781.html?spm=2114.search0104.3.70.AgRZdb&ws_ab_test=searchweb0_0,searchweb201602_3_10152_10065_10151_10068_10344_10345_10342_10343_10340_10341_10540_10304_10307_10301_10060_10155_10154_10056_10055_10054_10539_10538_10537_10059_10536_10534_10533_100031_10099_10338_10103_10102_5590020_10052_10053_10107_10050_10142_10051_10084_10083_5370020_10080_10082_10081_10177_10110_10111_10112_10113_10114_10312_5610011_10313_10314_10078_10079_10073-10051_10111,searchweb201603_17,ppcSwitch_5&btsid=4e7e72a9-c024-4590-a46a-446460987053&algo_expid=f4dae36a-0a97-462f-bca3-3f20c5737b01-9&algo_pvid=f4dae36a-0a97-462f-bca3-3f20c5737b01

Some USB Watchdogs have the option to hard turn off/on a rig.

Look for them here: https://www.aliexpress.com/wholesale?catId=0&initiative_id=SB_20171018020148&SearchText=USB+watchdog

It seems the bios does not re-initialise after power loss but just goes straight back to Windows. That causes problems because when the rig crashes, it seems to lose connectivity to the GPUs. So now even by cutting power and turning it back on, Windows only sees 1 GPU on startup. If I physically turn the rig off and on again, everything's loaded as it should, all 7 GPUs are detected.

Hmm... I don't understand why the smart plug does not force a full reboot... can you explain that?

Right. So one of my rigs is acting up.
Restarting Windows using Teamviewer does not reinitialise the GPUs for some reason so they do not all show up in device manager.
The only way to get everything to start properly after a crash is to physically turn off the rig, and on again.

Cutting the power using my TP-Link smart plug is not cutting it either. The mobo doesn't seem to re-initialise while powering back on, so devices are not loaded again when Windows restarts. Maybe that's a feature of the mobo, i.e. it's not resetting when power is cut, and rig reboots without going through POST processes?

Is there a way to force a full boot after power loss? Have I missed anything?