So last night we had a power failure....

 

I rebooted the server today and the system booted and started parity check as expected.  I let it run and came back 2 hours later (about the length of the check).  I looked at the hdd light and it was still it up as if the parity check was on-going.  I refreshed the interface and I got the "internet explorer cannot display this page" message.  tried telnet.. nothing...

 

I cycled the power and after it booted, quickly stopped the parity check and the server seemed to be working fine.. I opened random files... ran a few avis for a while... everythign seemed ok.  retried the parity check and at some point the computer locked up again (i think it was early on.. somewhere around 3-5%)

 

I again cycled the power and turned off the parity check, ran telnet and typed the command " tail -f /var/log/syslog .

After the crash (the interface went non-reponsive) there was no new error to the syslog.  Below is a screen capture on my computer where the unraid interface screen cannot be accessed after a crash and the telnet client with the output to syslog.

 

 

How could parity check be causing this crash all of a sudden?  I've never had as much as a hiccup in previous parity checks. No sync errors or crashes.  How can I diagnose what the problem is?  I don't know anything about parity checking and the hardware/software demands that might be causing this crash.

 

When I was getting responses from the unraid interface, there were no sync errors reported, no errors on disks and all drives were green.

 

Matt

 

p.s. I'm going to run it without a parity check for the time being to see if the crash might be unrelated to parity checking.

 

It would seem that it is not directly tied to the parity check as it crashed anyways.. results in telnet are the same... but this time the crash was preceeded with a buzzer sound for about 10 seconds before the red hdd light came on.

Could this be power supply? or mobo?

My first thought was corrupted file system, but a parity check does not use the file system, or look at files.  A power spike may have damaged a drive, and the parity check is hanging when it reaches a certain point of the damaged drive, so I would obtain SMART reports for all of the drives, and examine them.  For help with grabbing SMART reports, see this or the new Console commands, smartctl section.  Or use UnMENU's MyMAIN plugin with SMART view.

 

A power failure is often accompanied by power spikes, both at the time of outage, and at the time of power resumption.  It is possible that an electrical spike has damaged other chips, on the board, a card, or within the power supply.  Make sure all fans are operating, especially the CPU fan.  And you might try a Memtest.  I do recommend the use of a UPS, or at least surge protection.

 

The syslog tail piece is completely normal, nothing wrong showing there.

RobJ, thank you for the response and Merry Christmas!

 

Ok smart tests completed... all say they have no errors from what I can tell...  I have 4 drives and have included a windows text file with the results of the smartctrl test for each drive as created by the command

 

smartctrl -a -d ata /dev/sadX > smartsdX.txt

 

Here is my system configuration (drive locations).

 

 

memtest didn't reveal anything either... cpu fan still spinning... all case fans are spinning...  Power supply fans spinning...  UPS is now quickly moving up my todo list (top).

 

Any other ideas on locating the source of the problem?

 

Cheers,

 

Matt

Try running

 

ps -eaf | grep emhttp

 

to see if if the unRAID interface is running or not.  If not, run it by hand, and then start the parity check, and see what messages you get (such as a segfault).

 

You may also want to hook a monitor up to the console, to see if there are any messages showing up there.

 

How loud was the "buzzer sound"  Did it sound like it came from a drive, or the mobo?  Was it constant or pulsating?

How loud was the "buzzer sound"  Did it sound like it came from a drive, or the mobo?  Was it constant or pulsating?

 

It sounded like a constant buzzer for about 10 seconds from the mobo.  When the buzzer went off is exactly when the red hdd turned on and the system was unresponsive.  (even holding down the power button for 10 seconds to force a turn off would not work).  I have an older Antec power supply @300 watts (about 4-5 years old).

 

I have the server on now with no problems from over 2 hours now, I turned off the array and spun down the drives and am letting it go to see if it will crash.  I'm wondering if my power supply may be damaged and can't keep up when the computer needs it for running all the drives(ie parity check)?

 

Merry Christmas!

 

Matt

RobJ, thank you for the response and Merry Christmas!

 

Ok smart tests completed... all say they have no errors from what I can tell...  I have 4 drives and have included a windows text file with the results of the smartctrl test for each drive as created by the command

 

smartctrl -a -d ata /dev/sadX > smartsdX.txt

 

Here is my system configuration (drive locations).

 

 

memtest didn't reveal anything either... cpu fan still spinning... all case fans are spinning...  Power supply fans spinning...  UPS is now quickly moving up my todo list (top).

 

Any other ideas on locating the source of the problem?

 

Cheers,

 

Matt

 

Nothing seems of much concern in the smart logs.  There is an unknown attribute (#188) that has a value of "1" on 2 of them.  Not sure what this is, but likely nothing too awefully bad.

In order to help, I need the answers to the questions I asked.

 

How LOUD was the buzzing?  Did it come from a drive or mobo (or PSU)? 

 

300 Watts is more than enough juice for your drives and system.

 

Try running a smartctl short test on each one.

In order to help, I need the answers to the questions I asked.

 

How LOUD was the buzzing?  Did it come from a drive or mobo (or PSU)? 

 

300 Watts is more than enough juice for your drives and system.

 

Try running a smartctl short test on each one.

 

It was pretty loud.  louder than the start-up beep... It sounded pretty horrible.  It sounded like it was from the mobo, not the drives. other than that I don't know how to quantify loudness for the purposes of posting online.  It was about as loud as a buzzer from your average board game.

 

I just ran another parity check and it completed 100% successfully with no sync errors... the server has been up and running for the last 8 hours (2-2.5 with the array started).  The smartctrl results are posted in an earlier post.  I captured them again and compared.. they are the same now as then, no need to re-post.

 

Thanks again for all the help,

 

matt

That's why I asked how loud... a moderate to ligh buzzing can be the servo in the drive... a very bad sign.  A very faint one would be from the PSU.  A loud one would be from the mobo (or alarm on a drive cage). 

 

You ran the smartctl info output... you didn't run the smartctl *tests* that would affirmatively test the drives.