I have a strange one.

 

unRAID 4.3 beta 6 PRO. Currently 10 drives. Asrock MB. Two Supermicro 8 port cards, 1GIG RAM, Hiper Type 4 580W modular PSU. Lexar jump drive.

 

recently my server just turns of. I can find no pattern although it happens more when writing to Samba (potential red herring here). There are no logs generated in syslog that I can find.

 

It wont turn back on until the PSU is unplugged left for a few seconds then reattached (even if left for 24 hours). Using the PSU switch does not make it come back on. Two times i have had to reformat the flash drive as it would hang during boot up.

 

I am at a loss can anyone suggest anything. If there are strong suspicions I am prepared to buy new kit to replace components but obviously not at a whim.

 

appreciated

Your symptoms describe either power supply or motherboard issues.

 

For it not to boot when power-cycled makes me suspect the power supply the most. 

The flash drive, I think, is another potential red herring.  it might just be reacting to the poor power it is getting.

 

Looking at the specs for that supply, and the tests run here: http://www.extremeoverclocking.com/reviews/cases/Hiper_TypeR_580W_6.html

The tests at that site say that supply has two 12 volt rails.  One would be used by your motherboard, the other your disks. 

 

You are therefore, probably limited to about 20 amps to feed all 10 of your drives.  That might be cutting it pretty close, especially since it says "peak" output.

 

Joe L.

580W should be enough for 10 drives. (depending on the type and this is key).

 

If the power supply is overloaded or shorted somewhere, then it will exhibit symptoms as described.

I.E. Having to unplug it from the wall.

The switch on the back of the power supply would suffice if it were just a motherboard lockup.

 

You may try disconnecting a few drives from the power supply to see if it is overloaded.

 

If so, you could look into the power up in standby mode of some drives.

WD Green drives have a jumper or a software selectable via hdparm.

I think Seagate drives require a software setting.

 

This would reduce the huge power on surge.

 

However if your machine is turning off during writes, there is more going on here.

How many drives are active during these writes? How many drives are actually spinning?

 

When the machine powers off, is the LED on the motherboard still on?

 

Check the supermicro cards to see if there are any exposed edges over past the PCI slot.

if I remember correctly the Supermicro cards are PCI-X. I suppose you could use masking tape to cover any exposed copper just in case.

 

My number one suspect would be the power supply, agree with the others.  Number 2 would be temps, CPU temp or PSU temp or other motherboard chipset temp.  Third would be something on or a part of the motherboard, memory going bad or loose (because of temps nearby?) or northbridge or southbridge or power caps or loose addon card or ...

 

The above could all cause the symptoms you are seeing, power and temps most of all.  No USB issue I can think of would explain it.  Flash drive problems cause either a failure to boot, or small malfunctions later, with error messages in the syslog, nothing worse.

did you contact Tom?

 

 

Thanks for all the fast posts. I am certain that at least 2 of the crashes happened when only 2 drives were spun up. That tends to suggest its not PSU but definately doesn't rule it out.

 

I will have a new Gigabyte MB available soon thats far higher spec than this Asrock one. I think if that doesn't fix it it has to be the PSU as this problem also occurred with SATA1 TX4 cards as well.

 

the temperature one might also be an interesting option but since unRAID doesnt log this info its really hard to prove.

 

Will keep you posted.

 

Cheers all

Make sure the motherboard is not accidentally shorting out to the case with a mis-placed stand-off.  It would shut down the power supply for sure.  It might even be temperature related, with heat some expansion occurs and ..... zap.

 

Joe L.

It sounds to me like some component is hitting some protective temperature threshold which is powering down the computer.  My #1 guess would be the PSU.  Is it possible that it is a legitimate issue (i.e., your computer is really getting HOT)?  If not, it could be a bad sensor.

 

If the PSU is going down in this way, you'd never see it in any unRAID log.

 

You could run smartctl on your drives and see the max temps for them.  If might give some indication if you are having temp issues in your case.