Hello, all.  I've been troubleshooting my issue for awhile now, and thought I'd toss it out here for some feedback.

 

Here is the issue I've been having so far.  Recently (i.e., in the past two weeks), I've had random power offs of my UnRAID server.  It powers off in two ways.  Complete power off (no blinky lights) and everything powered off except the power light (when this happens, the system must be fully hard powered off by the PSDU switch). This isn't due to any power issue within my house.  I have the server plugged into the same circuit/wall socket as my main gaming rig, and I've had the system lose power while browsing the internet on the gaming system.  The power off seems to be random, sometimes happening within 30 minutes of power on, and sometimes taking hours (sometimes over 12) to eventually power off.  I am not running any heavy loads on the system, and have even disabled virtually all dockers to minimize the strain the system may experience.  I suspect some capacitor somewhere is failing, but I've already removed/replaced parts and the issue does not resolve itself.  The thing is, this isn't the first time this happened.  I had the exact same symptoms about 6 months ago.  Swapped motherboard/CPU, no change, swapped PSU, no change, swapped PSU back to original, problem went away?

 

Here is a list of the components currently installed:

 

GA-78LMT-USB3 motherboard

AMD FX-8320 processor

EVGA 750B powersupply

16 GB DDR3 RAM

1 4TB Western Digital Blue Hard Drive (parity)

3 1TB Western Digital Green hard Drives (storage)

1 IcyDock quick bay (connected via molex connectors to the power supply)

1 case

1 USB Stick (UnRAID OS 6.6.6)

3 80MM fans

 

That's literally all that's in there.  No wifi dongle, no GPU, etc.  I've swapped out the power supply (the previous one was a CX750 from Corsair), and the issue resolved for awhile, but it is happening again.  So, probably not the PSU.  I have a complete spare motherboard and processor (exact same models) that I've swapped out, which also did not resolve the issue.  So, probably not those.  I have not swapped out the RAM (I suppose that could be the issue, but I'm not ready to justify buying another 2 8GB sticks just yet).  There seems to be no damage to the motherboard (no bulging caps or board discoloration).  The entire system has been blown out of all dust.  I haven't moved the drives out of the IcyDock, but I may go ahead and do that to rule out the IcyDock itself (it is a bit older).  The case is also my original case (I mention this since I found one outlier situation where the person experiencing the power offs tracked it to a shorting case power switch).

 

Has anyone experienced issues like this before?  If so, what was your solution?

 

My plans are to remove the drives from their bays and hook them directly to power (no IcyDock) and to reapply thermal paste (Arctic Silver, my go-to choice) to the CPU, just on the off chance that could still be the issue.

 

Thanks in advance.

IS it a clean shutdown?  Or a dirty shutdown?   Difference is that with a clean shutdown, there will be no Parity Check on restart.)  

 

Things to check for. 

 

1 -- Pets and Kids.  Both of these have been involved in unexplained shutdown in the past.  Cats seem to love those blinking HD LED's!

 

2-- Some mischievous person who thinks it is funny.  Change password to your server.  Double check WiFi security if there is no one on the wired network that would do this.  Change the passphase to one that is 12 of more characters long with numbers and symbols included.

 

It's a dirty power down.  Parity check starts when I boot it back up.

 

The system itself has actually powered off while I was sitting next to it, so no, not kids/pets.

 

I have nothing open on my router facing the internet (i.e., no forwarded ports), and my wifi password is 16+ characters (combination of upper/lower and numbers/symbols).  No unauthorized connections detected.  Additionally, a forced power off would leave the system in a fully powered down state.  Typically, the case still shows a power light.  Just no fans spinning, no hard drive activity, no network activity. 

19 hours ago, solnoid said:

This isn't due to any power issue within my house.  I have the server plugged into the same circuit/wall socket as my main gaming rig, and I've had the system lose power while browsing the internet on the gaming system. 

Have you tried swapping outlets between these 2 computers? Moving to another circuit? Does it happen if you have your gaming system powered down?

 

I used to have a laser printer on the same circuit and it would make the UPS on my server switch to battery briefly on occasion.

 

You really should have an UPS on your server anyway.

9 minutes ago, trurl said:

Have you tried swapping outlets between these 2 computers? Moving to another circuit? Does it happen if you have your gaming system powered down?

 

I used to have a laser printer on the same circuit and it would make the UPS on my server switch to battery briefly on occasion.

 

You really should have an UPS on your server anyway.

I have swapped them to different outlets.  I haven't swapped it to another circuit due to it needing a network connection (I have it hooked into a wireless range extender), and I have had the issue with the gaming system powered down.  In fact, typically the gaming system is off throughout the day.  The only things plugged in and powered on are my desk lamp, the range extender, and the server.

 

As for the UPS, I've been eyeballing it, but my understanding is that most consumer grade UPS systems do not have the battery inline, which means that the power coming into the server would remain on wall power until/unless power dropped.  This concerns me, as the swap over could cause drops as well, and if there is dirty power from the wall, the lack of inline battery means no power conditioning.

 

As an interesting side note, the gaming system has a very similar hardware profile, being that it also contains a 750w PSU, GA-78LMT-USB3 motherboard, FX-8320 processor, and 16GB RAM.  The primary difference is the drive configuration (1 256GB ssd, a 1TB hard drive, and a 2TB hard drive) and the addition of a Gigabye NVIDIA GTX 1050 OC video card.  No power off issues with that one.

Edited December 19, 20187 yr by solnoid
accuracy

What is the idle power load on this server?  I noticed that you are using a 750W PSU on a system that could easily run on a 450W PSU.  I am wondering if the PSU might be shutting down because of low percentage of power draw at the times when the server is virtually idle.  (I know this is a bit of a stretch but...)  That gaming system has that Video Card which I bet uses more power then the rest of the system!   

 

By the way, even the cheap UPS have power line filtering  and peak voltage clamping to help control dirty power.  I am not aware of any consumer UPS's that run full time using the battery and inverter.  You have to move into the commercial UPS's to find that feature.   You will find that there is more concern about the use of pure sine wave as opposed to approximate sine wave conversion.  If you are concerned about that, you will find a wide difference of opinion.  Some folks think it is a non-issue and others think sine wave is the only reasonable option.  

12 minutes ago, Frank1940 said:

What is the idle power load on this server?  I noticed that you are using a 750W PSU on a system that could easily run on a 450W PSU.  I am wondering if the PSU might be shutting down because of low percentage of power draw at the times when the server is virtually idle.  (I know this is a bit of a stretch but...)  That gaming system has that Video Card which I bet uses more power then the rest of the system!   

 

By the way, even the cheap UPS have power line filtering  and peak voltage clamping to help control dirty power.  I am not aware of any consumer UPS's that run full time using the battery and inverter.  You have to move into the commercial UPS's to find that feature.   You will find that there is more concern about the use of pure sine wave as opposed to approximate sine wave conversion.  If you are concerned about that, you will find a wide difference of opinion.  Some folks think it is a non-issue and others think sine wave is the only reasonable option.  

Thanks for the info about the UPS.  Is there a place for recommendations on budget UPS systems for UnRAID?

 

As for the PSU, I had considered that, but I don't want to have to worry about expandability later on down the road when I eventually add more drives, hence why I went bigger to begin with.  So, how would I go about estimating the best PSU for the build I have?  And do you think the IcyDock drive cage could be overdrawing the molex connectors?  I'm actually considering moving those drives completely out of the cage and into the tower proper in the open drive bays just to rule it out.

It is better/simpler if you get an UPS that is compatible with APC protocol so Unraid can monitor its status. APC and Cyberpower both should work fine. Between hurricanes and winter storms I have lost power several times recently. Never had an unclean shutdown since Unraid knew it had been running on battery too long and shut itself down long before the battery ran down.