First time user, server keeps powering itself down during first parity sync?

[bello2185]

Hey guys, I upgraded my desktop PC so I decided to use my leftover old parts + some new stuff to build an unRAID server. I build the server and everything went fine, after changing some BIOS settings unRAID also booted ok. I logged in via the web interface, made my array (2x SSD for cache, 1 HDD for parity, 1 HDD for data) and started the first parity sync. This took ~ 15 hours so I went to bed after a while. When I checked the server this morning I came to find that it completely powered down by itself. So I boot it up again, visit the web interface, start the first parity sync again. And sure enough, after a while I again notice that it powered itself down.

This was with unRAID 6.4.0, so I tried 6.4.1rc2 and 6.3.5, but the exact same thing happened. I ran Memtest86 but it came back clean. On 6.3.5 I tried monitoring the syslog via SSH, but after starting the parity check it didn't output anything. This is the complete syslog until it powered itself off: https://pastebin.com/NCDTrjQc

The specs of the server are:

 

• Intel 5820K with Noctua NH-U9S cooler
• Asus X99A
• Corsair DDR4 2400 MHz 16GB
• Asus GeForce GT 710 1GB
• 2x Samsung 850 Pro 256GB (Cache pool)
• 2x HGST Deskstar NAS (2017) 8TB (1 data, 1 parity)
• Silverstone Grandia GD08B
• Corsair AX 860, 860W (80 Plus Platinum)
• Sandisk Ultra Fit 32GB USB 3.0

 

I'm out of ideas of what it could be, hopefully some of you might have some suggestions.

[trurl]

6 minutes ago, bello2185 said:

This took ~ 15 hours

So did it ever complete the parity check, or was this ~15 hours the estimated completion? How long does it actually run?

 

[JonathanM]

Spontaneous power off events are generally hardware or environment. Double check CPU cooling and thermal paste first of all, then power supply stability all the way from the breaker to the outlet to the UPS battery condition to the computer's power supply.

 

It would be helpful to actually observe it happen, as that can help point to environmental causes like a power dip when a load kicks in elsewhere in the building.

[bello2185]

14 minutes ago, trurl said:

So did it ever complete the parity check, or was this ~15 hours the estimated completion? How long does it actually run?

 

 

When I went to bed it was at about 80%, the 15 hours were the elapsed + remaining time. Some time during the night it powered itself down. The tries I did today all stopped a couple of hours in.

 

11 minutes ago, jonathanm said:

Spontaneous power off events are generally hardware or environment. Double check CPU cooling and thermal paste first of all, then power supply stability all the way from the breaker to the outlet to the UPS battery condition to the computer's power supply.

 

It would be helpful to actually observe it happen, as that can help point to environmental causes like a power dip when a load kicks in elsewhere in the building.

 

I think the cooling should be sufficient, I used Noctua thermal paste (right amount, not to much, not to little) and the specs of the cooler say it's good for 165w TDP. The screws are properly tightened, so it's making good contact with the heatspreader. I've also got 2x 120mm intake and 2x 80mm outtake near the CPU. I think the power supply should be alright too, up until a couple of days ago it ran reliably under heavy loads in my gaming rig. Currently I've got the server set up in the same location as I had my HTPC. The HTPC ran 24/7 without problems from the same outlet, so I think that should be okay too. I could try swapping the PSU cable and trying another socket.

[bello2185]

This morning I cleared the bios and loaded the default settings + a different PSU cable on a different socket + a new USB with 6.4.0, but after starting the parity sync, it again shut itself down after some hours. I had the server connected to a TV and what I saw on the screen before it shut itself down some time later, was the command line interface that was frozen. It didn't show any errors or anything on screen.