Hello Limetech Community,

 

I built my new server on Saturday.

 

MBoard: Asus P8B75-M-LX

CPU: Intel G550

Memory: G.Skill 1666 Ripjaws 4 GB

Case & PSU: Antec P183 V3 & Antec Neo Eco 620

HDD: 5 x WD Red NAS drives 3TB  (1 for parity, 1 for cache, 3 for storage)

Unraid version: Pro

 

Well, I started the simultaneous (different pty's) preclear process on my drives at 3pm on Saturday. It finished on Monday morning at 10am. All was ok before I had to go on overnight business. Everything passed, no errors all good.

 

Came back last night, did some bios tweaks that I read about (e.g. enable AHCI, disable sound, serial, parallel ports etc) and then installed the Pro key. Added my devices into an array and the parity sync started. Now boom:

 

Disk2 is showing: Reads: 4294967239 Writes: 17314587057 Errors: 11671800

 

I am confused because preclear gave the drives a workout and returned no issue, but Unraid now tells me there are problems. The parity sync is continuing as I type and the reads, writes and errors are increasing. The parity disk, cache disk and the two other disks in the array seem fine.

 

I have attached a screenshot of my server MAIN window as of 2 mins before posting.

 

Can I please ask for some support from this forum. Am I missing something, have I really got a dud disk here or have I done something wrong?

 

 

Thank you in advance.

 

Daniel

 

 

[EDIT] Was a loose cable. Doh, *Slaps Forehead*

A bit of an update. Parity sync is still running and the errors, writes and reads on this disk are increasing:

 

 

My syslog is showing masses of the following entries:

 

Feb  6 05:48:38 nas kernel: handle_stripe read error: 16230248/2, count: 1

Feb  6 05:48:38 nas kernel: md: disk2 read error

 

 

Weird thing, I tried to do:

 

 

smartctl  -a  -d  ata  /dev/sdd OR smartctl -a -A /dev/sdd

 

and get:

 

Smartctl open device: /dev/sdd failed: No such device

 

 

Very weird (to me anyway) given the web GUI is showing that the system recognizes it and also that preclear run on it. Hmm, cant be anything daft like a loose cable can it!?

The really confusing thing is that it is showing all of your drives unformatted. Is there data on them?

 

Is it doing a correcting parity check? I guess it doesn't matter whether it completes it or not since if it is non-correcting it won't change anything, and if it is correcting it is too late anyway.

 

Just don't tell it to format anything if you have data on these disks. Someone else will have to help you with this one, but in the meantime you might try to get a syslog.

It's fully possible that a disk failure happened after a pre-clear.  Hard drives generally have a bit of a bell curve failure rate where the odds of a failure are higher early, less so after a month or so leveling off to rare after about 6 months with a plateau to about 2 years, where the curve starts to pick up again (depending on all sorts of factors, of course.)  Being that you're in the first week or so, you're still in the (please excuse the crude-ness of this term, but it fits) high infant mortality period.

 

Or you could have a loose/bad SATA cable, that'll often give you errors but not kick the drive offline, and it's the cheaper thing to test.  If replacing the cable seems to make the errors stop increasing, I would certainly run a few rounds of pre-clear on the drives.  As a quick test you could simply swap the drive ends of the SATA cables between drives and see if errors stay with the drive or the cable.  It could even be a bad SATA port; rare, but it can happen.

The "no such device" seems to indicate a drive that is no longer responding... and the most likely cause is one of the cables to it is loose.  It could be the SATA cable, or the power cable, or a drive tray connector, or backplane connector. 

 

Next most likely, the drive died... and then finally, the disk controller port died, or the disk controller card is not seated well in the MB. (if it is an add-on card)

 

As far as the interface goes... it saw the drive at one time... If you were to reboot now, I'm pretty sure the drive would show as missing instead of just having "read" errors.

 

Joe L.

 

 

But why does the user interface show all drives unformatted?

But why does the user interface show all drives unformatted?

He never said he formatted them.  You do not have to format them before calculating the initial parity...

 

Joe L.

I guess that makes sense. It has to update parity anyway whenever you add a new drive and then format it so I guess as far as parity goes it gets recalculated when the file system is created anyway.

 

OK

 

The really confusing thing is that it is showing all of your drives unformatted. Is there data on them?

 

Is it doing a correcting parity check? I guess it doesn't matter whether it completes it or not since if it is non-correcting it won't change anything, and if it is correcting it is too late anyway.

 

Just don't tell it to format anything if you have data on these disks. Someone else will have to help you with this one, but in the meantime you might try to get a syslog.

 

@ trurl - My understanding is that the parity is calculated initially "automatically" on a new install of blank drives (at least that is my experience here as I certainly didn't initiate it) I have no data on them anyway so is all good. When I format and create the FS I imagine it will be updated again.

 

It's fully possible that a disk failure happened after a pre-clear.  Hard drives generally have a bit of a bell curve failure rate where the odds of a failure are higher early, less so after a month or so leveling off to rare after about 6 months with a plateau to about 2 years, where the curve starts to pick up again (depending on all sorts of factors, of course.)  Being that you're in the first week or so, you're still in the (please excuse the crude-ness of this term, but it fits) high infant mortality period.

 

Or you could have a loose/bad SATA cable, that'll often give you errors but not kick the drive offline, and it's the cheaper thing to test.  If replacing the cable seems to make the errors stop increasing, I would certainly run a few rounds of pre-clear on the drives.  As a quick test you could simply swap the drive ends of the SATA cables between drives and see if errors stay with the drive or the cable.  It could even be a bad SATA port; rare, but it can happen.

.

 

The "no such device" seems to indicate a drive that is no longer responding... and the most likely cause is one of the cables to it is loose.  It could be the SATA cable, or the power cable, or a drive tray connector, or backplane connector. 

 

Next most likely, the drive died... and then finally, the disk controller port died, or the disk controller card is not seated well in the MB. (if it is an add-on card)

 

As far as the interface goes... it saw the drive at one time... If you were to reboot now, I'm pretty sure the drive would show as missing instead of just having "read" errors.

 

Joe L.

 

 

 

 

@matguy & Joe L - I completely agree that the drive could have failed, but I am finding it hard to believe right now. Given the discovery of the drive is intermittent. I am at work now but will check the connections again when I get home and run another preclear on the drive in question. I havent migrated any data from my existing nas yet so all is good.

 

 

Thank you all for your quick responses! I will post back when I have an update.

 

 

Daniel

do not preclear a drive after it is already assigned to the array.  (actually, it will not let you, and there is no need regardless)

Just fix the cabling (most likely) or replace the drive if it really is defective.  Be happy it died before it holds your precious data.

 

Joe L.