I just built my second unRAID (3 500GB SATA II drives), and it seems to shut down after every parity check.  After I initially finished it, I cleared the two data disks, and it started a parity build just before going to bed.  The parity check was estimated to take 2 hours.  When I got up, the server was off. 

 

I restarted it, and it started the parity build again.  Again, the time estimate was two hours, but when I checked it again (more than 2 hours later), the machine was off again.  I'm pretty sure that both of these times, parity was still listed as invalid (but I'm not 100% sure at this point).

 

The thrid time, I tried to build parity, I stayed with the machine to try to see what was happening.  I clicked refresh throughout the 2-hour build/check, and it finished with valid parity and no errors, but within 5 minutes of completing, the machine turned off.  I was running that tail in a telnet session, but did not see anything prior to it shutting down (all I saw were some I/O errors every time I refreshed the status page).

 

When I restarted the machine this time, it listed parity as valid, but it started to check parity.  I guess this is because it had not shut down cleanly.  I decided to start testing data on the machine, and over the course of the day, I moved 100 GB onto one drive and 60 onto another without any problems or errors.  The machine ran fine all day, and I didn't see any type of data corruption with limited testing. 

 

But before going to bed last night, I ran another parity check, and this morning the machine was OFF.

 

Anybody ever experienced this?  Any ideas?

 

The setup is this:

Gigabyte GA-945GZM-S2 motherboard

Celeron D processor (around 3 GHz)

2x512 MB DDR2 533MHz cheap Corsair RAM

3x500GB Seagate Barracuda 7200.10 SATA II drives

Maybe it's not parity check related.  Had it running, doing nothing, all drives spun down.  Eventually, it shut off.  I'm not sure how long it took.

Could you have some type of power functions enabled in bios.  I would check and see if that is the case.  It something like that is enabled then the server would shut down after x amount of idle time.

Could you have some type of power functions enabled in bios. I would check and see if that is the case. It something like that is enabled then the server would shut down after x amount of idle time.

 

I really do hope this turns out to be a "duh" moment, and that's what I'm starting to think it is.  It was the fact that it happened immediately after a rebuild that threw me.

I'm not ready to say "duh" yet.  I cannot find any BIOS settings that I think would cause this.  I've pretty much disabled everything that I'm not 100% sure of.  To make the issue more bizarre, I was assuming it had something to do with the machine being idle.  It was pretty much humming along last night not shutting down (but not being touched either through shares, web interface, or telnet).  I invoked a parity check, and within a few minutes, it had shut off.

 

I had a tail going, but I just lost the final output.  Will try again.

I'm not ready to say "duh" yet.  I cannot find any BIOS settings that I think would cause this.  I've pretty much disabled everything that I'm not 100% sure of.  To make the issue more bizarre, I was assuming it had something to do with the machine being idle.  It was pretty much humming along last night not shutting down (but not being touched either through shares, web interface, or telnet).  I invoked a parity check, and within a few minutes, it had shut off.

 

I had a tail going, but I just lost the final output.  Will try again.

I would suspect the power supply...  Most all other failures (cpu, ram, etc) I can think of would result in a system freeze, not a power down.  Perhaps you have a fan that is stopping, and the bios monitoring it is shutting your power supply down to save the CPU from overheating?  Is the CPU heatsink still on securely.

 

When you assembled the unit, did you use heatsink compound? 

 

Some motherboards have a feature as follows... does yours?

CPU ThermalGuard™

The ThermalGuard™ technology prevents the CPU from overheating and burning up in the case of any hardware cooling malfunction. This hardware thermal protection automatically shuts down your system when the CPU temperature reaches a certain threshold.

Thanks for the ideas, Joe.  I've changed the power supply, and I've disabled the "CPU Smart Fan Control."  I had noticed that when I would turn the box on, the CPU fan would spin up, then stop for about 5 seconds before starting up again.  Within the BIOS, it read about 1300 RPM when the fan restarted and very resonable temps, so I didn't think much of the 5 second fan stop.  With it disabled, the fan spins at about 2800 RPM and it doesn't take that "break" on startup.

 

Hopefully either the PSU or the CPU fan was the problem.  I've made it through one parity build and I'm running another right now.  If it's awake in the morning, I'll feel safe in thinking the problem is resolved.

 

Thanks