And I'm back again.  Tower is currently cleanly powered down.

 

Before I proceed I figured I'd post my syslog and see what the problem is.  The tower started its monthly parity check (uncorrected) and I noticed today that it was only at 4% after running a full 24 hours.  It did not show any parity errors but the syslog was full of red lines.  It cancelled the check without issues.  I rebooted it without issues and it came up fine.  Syslog looked good so I restarted the parity check.  It started throwing the same error as before right away.  Again I shut it down without issue. 

 

I've reseated a few drives but I have no idea what drive its complaining about.  Again it boots normally and begins to show the errors as soon as I start the check.

 

Here is the syslog from the last boot.  parity check start and a normal cancel.  It keeps complaining about ata13.  How do I tell what drive that is?

 

I think I've replaced 4 drives in the last 6 months.  Last months check completed normally without errors and in normal time.  So this problem is recent.  Also for history's sake.  Last time (a couple months ago) I discovered that slot 4 was bad.  So disk4 is currently in slot 14.  All slots are full other then the unused known bad slot 4.

 

Thoughts?

syslog-2014-12-02.zip

If you search the syslog for the early occurrences of ata13 it tells you what drive it is!

Dec  2 00:47:26 Tower kernel: ata13: SATA link up 1.5 Gbps (SStatus 113 SControl 300)
Dec  2 00:47:26 Tower kernel: ata13.00: ATA-7: ST3500630AS,             9QG2TN7S, 3.AAE, max UDMA/133

An earlier line suggested this was device /dev/sdl

 

Thank you.  According to my records I think that may be disk 2.  If so its one of the old almost original 500GB drives that has been in use for years.  Now that I know what I'm looking for I can pull the drives and quickly find it and confirm the serial number.  Guess I'll drop by my local store after work today and buy another drive.  I see they still have a sale on so thats good.

 

Does anyone know of a windows program that can preclear a drive for unRaid?  I suppose I'll preclear the new one I buy today in this slot but I have a couple other drives I'd like to test.  I have no free slots on the server however to let the other drives sit and do a preclear.

 

If there are no win programs that can preclear it.  Then what do you guys recommend as a program that can fully test all the sectors on a drive?

 

 

You could just install unraid on another usb drive and boot into that. You'd be down whatever that computer is of course. You could also preclear on your unraid server in that one slot but you should probably not have the array started until you can replace that drive. Having old small drives die is the reason I wish we could remove a drive from the array permanently and let unraid copy all it's files to the other disks.

You could just install unraid on another usb drive and boot into that. You'd be down whatever that computer is of course. You could also preclear on your unraid server in that one slot but you should probably not have the array started until you can replace that drive. Having old small drives die is the reason I wish we could remove a drive from the array permanently and let unraid copy all it's files to the other disks.

 

Or you could look for a used computer at a garage sale.  Just be sure that it has SATA ports and that it will boot up.  ( At least far enough to indicate the it can't find an operating system (or some other indication that the MB is OK).  For your purposes, you don't care if the hard disk is bad!)

I have in the past booted up unRAID (basic version) under VirtualBox on Windows and used that as the pre-clear environment.

Yes.  Current plan is to preclear the new drive in the slot where I will be replacing the drive.  I'll do this while the array is not started.  I know I can start the array and leave this slot unassigned and then do a preclear with the array running but unprotected.  But I'd rather leave it not started while doing to the preclear so thats the plan.

 

New drive has been reserved and is ready for pickup after work.

 

 

Thats a great Idea.  I've never even considered that. 

 

I can't give up my main box.  The other box is connect to the TV and Home automation stuff.  But I do have a netbook thats mostly unused that I could use.  Hm.... I think that only has one usb slot on it.  I'll have to check tonight.  I should have an extra usb hub around somewhere and I have a kit for connecting any drive to usb.  I wonder if I could boot from a usb stick in the hub and see the drive also in the hub.  I'll have to try that.   

 

 

Garage sale.  Well.  I do have a couple carcases left over.  I might be able to put enough together to make a bootable system without buying anything.

 

Hm... VirtualBox.  Another good idea.  I like that idea even better.  My main box has a slot on top built specifically to slide an internal drive into it for temp access without opening the case or plugging into USB.  I've used it a few times.  Its handy.  If I could access that with unraid running in vituralbox.  That would be perfect.

 

Great Ideas guys.  Thanks.

 

 

Bought a replacement drive last night.  Tower is up with array off currently doing a preclear on the new drive in slot 2 where the bad drive was.

 

Installed virtualbox last night to play with it.  Installed a VM of windows7 since thats what I had thats currently unused.  Worked great.  This is pretty slick.  I'll have to see now if I can get unraid running within it.  This would be handy for disk checking as I have no free slots.

 

I'll report back once the preclear and rebuild are done.  Thanks guys for the info and suggestions.

Preclear completed without issues.  I let the drive rebuild.  Its now Completed and all is good.

 

Thanks everyone.

 

 

Now to go further explore the ideas given above. 

Run a parity check to confirm successful rebuild.

?? ok.  Is it possible for it to rebuild but NOT show errors in the syslog?  Everything seems fine.  I've never done a check in the past after a rebuild.  Should I?  I've started the check.

You should always check after a rebuild.  The writing of the rebuild may have gone without error, but you need to verify that you can read the data without error as well.  Failure to do this can leave an unseen error which could result in an inability to rebuild again later if another drive was to fail.

Thanks for the info.  The check was started yesterday.  No issues so far.  I expect it to be done by tomorrow night assuming it goes normally.

 

Hourly status emails stopped coming so I took the home pc over from work to check the server.  Looks like it completed the check this morning.  No problems.  All is good.

As an aside, and on a possibly related note, I have on old Celeron (socket 478!) with a Promise 4-port PCI SATA controller.  Used it to preclear three disks the other day (I usually do one at a time but Black Friday cursed me into buying more drives....).  At SATA-1 speeds, it took almost 90 hours!  My cheapie PCIe SATA-2 card is now worth its weight in gold and fortunately I have a few old Socket 775 Core2's to use instead for next time.  Can't even imagine how long a preclear would take on a Netbook running a USB - SATA adapter on a modern 3TB+ drive.  If you get a garage sale PC for this purpose (or a super cheap MB/CPU combo on ebay) make sure it has at least SATA-2, otherwise you will be waiting a while for preclears to complete.  YMMV of course.