I was in the process of transferring a large # of video files to my unRAID server (4.3.3) when I heard my computer beep.  I found my Explorer session locked up and I could no longer get to my unRAID server via my mapped network drive.  I went to the browser UI and tried to "Stop" unRAID (in order to shut it down and recycle it), but that too hung.  I then went in at the OS level and issued an "init 0" command, but that hung as well.  So I finally had to pull the power cord.

 

Well, unRAID is now back up, but the drive I was using is now has a blue dot beside it and my only option appears to be to "Refresh" (Start & Restore are grayed out).  What next?  There's a lot of data on that drive that I don't want to go without.

 

Thanks.

 

BillK

Whatever you do... DO NOT PRESS "RESTORE"  It is actually an "Initialize Configuration and Immediately Invalidate Parity" button.

 

 

The first thing you need to do is post a syslog.

 

Thanks Joe.

 

Here's the file (although I've rebooted a few times).

 

BillK

syslog.txt

Thanks Joe.

 

Here's the file (although I've rebooted a few times).

 

BillK

See if you can get smart reports from the failed drive.

 

Since you rebooted, and have no log of the events that caused the crash,  we have no idea of the cause.

Joe,

 

How would I get a smart report?

 

And I understand about the reboots, next time I'll get the syslog the first time.

 

BillK

Joe,

 

How would I get a smart report?

 

And I understand about the reboots, next time I'll get the syslog the first time.

 

BillK

I have to go out on a few errands, but the wiki describes how to go about it:

http://lime-technology.com/wiki/index.php?title=Troubleshooting

Thanks Joe.  Here's the results...

 

smart.txt

Joe, et all.,

 

Should I go ahead and re-"Start" unRAID?

 

Thanks.

 

Bill

Joe, et all.,

 

Should I go ahead and re-"Start" unRAID?

 

Thanks.

 

Bill

The syslog seems to indicate that disk1 was replaced.  Is that the one that is "blue"?

 

if you check the checkbox under "Start" does it enable it?

 

Joe L.

Although the icon next to disk 1 is blue, it was not replaced.  It was, however, the drive I was writing to when transferring files (& it or the parity drive hung).

 

And "yes", if I check the checkbox under "Start", but button becomes enabled.

 

BillK

Somehow the server thinks it is new... I have no idea why.

 

I suspect if you were to start the array it will clear it.  Probably not what you want to happen.  what does the text alongside the "Start button say?

 

(And do NOT press "Start" just yet, not if you want the files on the disk marked as "new")

 

Correct.  I believe the contents are still on the drive and I DON"T want to lose them.  In essence, I want unRAID to accept the disk (& data) and, if necessary, rebuild parity.

 

The text beside the "Start" button says, "Start will bring the array on-line, start Data-Rebuild, and then expand the file system (if possible)."

 

Might this be a case where I might be better off removing the drive, instructing unRAID to rebuild parity, then use a Reiser/NFS driver (sorry, don't remember the name off the top of my head) to read the contents off the drive and copy them onto the unRAID NAS (i.e., different drive)?

Correct.  I believe the contents are still on the drive and I DON"T want to lose them.  In essence, I want unRAID to accept the disk (& data) and, if necessary, rebuild parity.

 

The text beside the "Start" button says, "Start will bring the array on-line, start Data-Rebuild, and then expand the file system (if possible)."

 

Might this be a case where I might be better off removing the drive, instructing unRAID to rebuild parity, then use a Reiser/NFS driver (sorry, don't remember the name off the top of my head) to read the contents off the drive and copy them onto the unRAID NAS (i.e., different drive)?

 

OK, that description seems to indicate the disk will be re-constructed.  Here is how I would proceed.

 

Go to the "devices" page and un-assign the drive marked as new.  (Or, un-plug it physically)  Either will ensure it is not going to be written to and both are treated exactly the same as a failed disk. 

 

Then, click on "Start" to start the array.  You should be able to get to the contents of the missing disk by unRAID's simulation of it via parity in combination with all the other disks.

 

If there are critical files on that disk, I'd copy them to a different disk at that time.  Just to be doubly safe.  They'll be read from the "simulated" drive, so you do not need to worry about the drive failing while copying the files off it. 

 

Then, once your files are safe, you can stop the array, re-assign the drive, and have it re-construct it from the other drives.

 

We still do not know why it was taken off-line originally, and if it has problems it may fail again when it is being rebuilt.

 

Joe L.

Joe,

 

I came up with a slight variant which I think is an even better option.  I went to Microcenter and picked up a new 1TB drive.  So...I plan to:

 

1.  Remove the old "offline" drive (now marked as "new"),

2.  Click "Start" (as you suggested), so the system treats it as a failed drive.

3.  Using unRAID's simulation to copy the contents onto one of my currently good, empty drives in the system.

4.  Stop the array, install my new 1TB drive and let unRAID rebuild it using the parity drive.

 

That will allow me to keep the old drive in its current state (which, I hope, will allow me to read the contents using a NFS/Reiser driver at a later point if my steps don't work).  If everything goes smoothly, however, I'll probably install the drive in my workstation and use Spinrite to torture test it.

Joe,

 

I came up with a slight variant which I think is an even better option.  I went to Microcenter and picked up a new 1TB drive.  So...I plan to:

 

1.  Remove the old "offline" drive (now marked as "new"),

2.  Click "Start" (as you suggested), so the system treats it as a failed drive.

3.  Using unRAID's simulation to copy the contents onto one of my currently good, empty drives in the system.

4.  Stop the array, install my new 1TB drive and let unRAID rebuild it using the parity drive.

 

That will allow me to keep the old drive in its current state (which, I hope, will allow me to read the contents using a NFS/Reiser driver at a later point if my steps don't work).  If everything goes smoothly, however, I'll probably install the drive in my workstation and use Spinrite to torture test it.

Sounds like a good plan as it preserves your original disk and lets you test it at your leisure.

Joe,

 

Curiouser and curiouser.....

 

I stopped the array, pulled out the old drive, and powered up the server as you suggested.  I noticed it was taking longer then expected, so I flipped my KVM and found that it was recognizing the drives, loading the boot menu (i.e., Memtest or unRAID), but wasn't actually booting up.  But, after three power cycles, it booted up slowly (i.e, bzimage took a while to load).  I then flipped back to my PC and launch the browser interface.  When it first came up it knew the old drive was no longer present, and told me it was trying to "mount" it (under the Free column).  But I also noticed it was trying to mount an older, and as far as I knew, working 500GB drive that I hadn't been actively using.

 

So I hit my browser's refresh button.  As before, it saw the old drive was missing, but listed the size as 500GB, not the 1TB size I expected (i.e., it seemed be using the size of the other drive I saw it briefly trying to mount).  All the other drives appeared normal.  I then tried to read the contents of the simulated drive through Explorer, both directly and through it's mapped network drive name....but no luck.  It just wouldn't allow me to touch (i.e., read) the simulated drive.

 

So I powered it down, put in the brand new 1TB drive, and hit the power button.  This time it didn't even get to the boot menu....so I powered it down, removed and reinstalled the thumb drive and powered it back up.  This time it booted up normally.  I flipped over to the web browser UI and, after confirming that it saw the new 1TB drive, started up the array.  It's now in the process of formatting/rebuilding the drive.  I'll let you know what happens after it comes fully online (i.e., will the content be there or not).

This isn't the first time this unit has crashed under heavy load (i.e., transferring many files).  If I can stabilize it, I am seriously considering rebuilding this thing from scratch (i.e., new MB, new 5-in-3 units, etc.).

What size and model PSU are you using?

 

An Enermax NoiseTaker II 600W PS

 

Here's the rest of the setup....

Motherboard:  Asus P5PE-VM Socket T (LGA 775) Intel 865G Micro ATX Intel Motherboard

CPU:  Intel Celeron D 326 Prescott 2.53Ghz LGA 775 EM64T Processor

Memory:  Two sticks of Corsair ValueSelect 512MB 184-Pin DDR SDRAM DDR 400 (PC 3200)

Controllers:  Two Promise SATA300 TX4 PCI SATA Controller Cards

Drives:  Four WD Green Caviar 1TB drives, one WD Green Caviar 500GB drive, and three Seagate 500GB drives

4-3 drive bay:  Two No-Name 3141SATA Back Plane Drive Bays (holds four SATA I or II drives in three 3 1/2" bays)

Fans:  There's a fan built into each of the 4-3 drive bay units (pulling air through the front, past the drives, and then into the case), a CPU fan (stock), a PS fan (stock), plus two fans at the back of the case pushing air out

 

 

What size and model PSU are you using?

 

BRiT,

 

I think I know where you're going with this, but not sure if this is really my "weakest link".  Yes, Enermax's split rail design does leave only 22A for the drives (versus Corsair's 650W PS w/ a single 52A rail).  But if this was really a limitation I would think I'd see huge problems when initially building parity or rebuilding a drive (as I'm doing know) when all 8 of my drives are spun up and working continuously.....not when I'm writing to a single disk (plus parity).

 

Make sense?

What size and model PSU are you using?

 

BRiT,

 

I think I know where you're going with this, but not sure if this is really my "weakest link".  Yes, Enermax's split rail design does leave only 22A for the drives (versus Corsair's 650W PS w/ a single 52A rail).  But if this was really a limitation I would think I'd see huge problems when initially building parity or rebuilding a drive (as I'm doing know) when all 8 of my drives are spun up and working continuously.....not when I'm writing to a single disk (plus parity).

 

Make sense?

You are assuming one of the 12 volt rails is exclusive to the disks.  I would not.  The second 12 volt rail might be exclusive to the PCIe connectors for power hungry video cards.

 

This review was not as glowing when loaded with more that about 150 watts on the molex connectors.  The power supply would not even boot the PC attache to it in the review.

http://www.3dvelocity.com/reviews/enermax600/noisetaker_5.htm

 

I could not find any description anywhere of which connectors were on which 12 volt rail. 

You are assuming one of the 12 volt rails is exclusive to the disks.  I would not.  The second 12 volt rail might be exclusive to the PCIe connectors for power hungry video cards.

 

This review was not as glowing when loaded with more that about 150 watts on the molex connectors.   The power supply would not even boot the PC attache to it in the review.

http://www.3dvelocity.com/reviews/enermax600/noisetaker_5.htm

 

I could not find any description anywhere of which connectors were on which 12 volt rail.  

 

Joe,

 

You're right.  My statement implicitly assumed one 12 volt rail was exclusive to the MB and power hungry video cards and the other was exclusive to the peripherals (i.e., hard disks, CD-ROM drives, floppy, etc.), which may not be true.

 

That said, I'm still a little skeptical the PS is the true culprit.  Otherwise, wouldn't I see even bigger problems when unRAID tried to first build the parity drive, or rebuild a data drive???  In either case there are 7 drives spun up performing nearly continuous reads and 1 drive spun up performing nearly continuous writes.  That seems much more strenuous than my attempts to copy large numbers of files from my PC to a single drive on my NAS.  In the latter case, isn't unRAID just spinning up a single data drive for writes and the parity drive for writes (i.e., it already knows the parity calculation before the new data arrives, so it just needs to factor in the new data and rewrite the parity if necessary...no need to spin up the other non-used disks).

 

On the other hand, I didn't want this problem to linger.  So I went to Microcenter and picked a Corsair CMPSU-650TX 650W ATX12V / EPS12V power supply, the same unit Tom uses in his servers due to it's high quality and single 52A rail.  It's now running happily inside my rack mounted case.

 

P.S.  The great news is that unRAID successfully rebuilt the missing drive & all its contents without a hitch!

 

P.P.S.  Joe....thanks for all the advice.  The next time the system crashes (knock on wood), I'll try to do a better job of capturing the error logs.