I had a red-balled disk a weeks back. Replaced it with a brand new disk and the array was rebuilt and everything was normal. Until couple days back even that disk showed as red-balled. I tried re-seating the cables etc but no luck. I even tried changing the sata port on the mobo to which the drive is connected to eliminate a mobo port issue. But no luck. syslog attached. please help, thanks.

 

syslog.txt

Try a new SATA cable.

Sorry I forgot to mention that I did that. Will be trying a new power cable today too.

Attach a Smart Status report for the disk that is red-balled.

 

Did you pre-clear the new disk before you put it into the array?

No I didn't pre-clear it. I just added it.

 

After a little experimentation, I got the disk to be detected as a new disk. So I let it get rebuilt. Now the problem is that its being rebuilt at super slow speed. 24 hours down the line I'm still at 2% of data rebuild. So something is definitely wrong. Please advise. Screenshot and new syslog both attached.

The syslog was about a 25 MB .txt file. Most of it looked repetitive. Had to delete over 50% of the ending text and compress it to attach here.

syslog.zip

Post a SMART report.

Smart test does not report any errors. Attached.

smart.txt

Your Seek_Error_Rate is extremely high as compared to my Seagate ST3000M1000 drives which are now over 10,000 hours.  (Mine are in the range of 4500000)

 

I suspect you have a bad drive.  (Hope someone else who more of an expert than I will jump in!)

 

Again I ask, did you preclear the drive before you put it into the array?? 

No, I did not preclear it. Should I get a new drive and pop it in place of this one?

I would get a new drive, once rebilt i would preckear this old drive and see if the smart settings go bad or if it sounds bad, if so i would warenty replace.

 

Sent from my YP-G70 using Tapatalk 2

 

 

No, I did not preclear it. Should I get a new drive and pop it in place of this one?

And before you rebuild, I'd preclear the new drive first. It's amazing how many drives fail in the first few hours of hard use in unraid. Windows machines typically don't use more than a small percentage of the whole disk, so many times a marginal drive will work just fine in a desktop machine. Unraid relies on the entire drive capacity of every drive to provide parity protection, so any defect is one too many.

No, I did not preclear it. Should I get a new drive and pop it in place of this one?

 

Most people on this forum would recommend preclearing any new drive before adding to any unRAID array.  Somewhere in the neighborhood of 10% to 20% of all NEW drives are defective when received or go bad in the first few hours of use.  Getting and 'popping' a new drive into an array without preclearing is asking for the type of headache which are you are going through right now. 

 

Preclear the new drive.  If (and only if) it passes, should you commit it to the rebuilt process.  I personally run three preclear cycles on each new drive.  For your 3TB drive, that could take about a hundred hours.  That period of time will get you past 99.% of all infant mortality failures.  Once past that infant mortality period, you can expect a couple of years of service from 99+% of the remaining drives. 

If a drive fails pre-clear, I doubt the manufacturer of the HDD would consider it as reason enough for a warranty replacement. Or would they? Becuase wouldnt it work normally for the 99.9% of the pouplace who would take the drive and run Windoze on it?

 

And if it were just one machine on which I was doing the unraid business it would be ok. But I almost build servers as a part of my profession. So its very time consuming for me to pre-clear SO many drives. Its less painful for me to just remove a problematic one, and slide in another. Btw, by sheer coincidence, the server that has given me the most trouble is indeed my personal server and not a client's thankfully!

If a drive fails pre-clear, I doubt the manufacturer of the HDD would consider it as reason enough for a warranty replacement. Or would they? Becuase wouldnt it work normally for the 99.9% of the pouplace who would take the drive and run Windoze on it?

 

And if it were just one machine on which I was doing the unraid business it would be ok. But I almost build servers as a part of my profession. So its very time consuming for me to pre-clear SO many drives. Its less painful for me to just remove a problematic one, and slide in another. Btw, by sheer coincidence, the server that has given me the most trouble is indeed my personal server and not a client's thankfully!

 

I have only returned a few drives over the years but I have never had any questions asked about the disk.  I would just apply for a RMA and ship it in.  They generally shipped out the new drive within a day of getting the returned one.  Now granted, Two or three years elapse between my returns so I would definitely not be 'flagged' as one who might be working the system.  I would also think they would give less hassle with the  return of a new drive than one which is within ten days of having the warranty expire!  After all, you didn't buy the drive so you could ship it back to them at your expense without having a good reason.  (And testing every returned drive as it comes in the door to see if it is defective at that point isn't really cost effective or a good idea from a CR standpoint as 90+% of them will be obliviously defective on a through test and 90% of the remaining ones probably have a intermittent problems that is difficult to find.)

I don't think I am ever going to be able to have a protected array. I have replaced the suspect disk with a brand new disk and the dat re-build is again ~ 1MB per sec. 

 

syslog attached.

 

It may be worthy to note that when I opened the case to retrieve the HDD, the inside of the cabinet was burning hot, and a couple of the disks (including the one that I took out) were really really hot although the server wasnt even in use or rebuilding data at the point since I had just disabled the "bad" disk.

syslog.txt

That syslog shows ata2 continually resetting so, you have some sort of problem.  That would explain why everything is so slow  I would suspect that it is either power or SATA cable related.

 

If the drives were running excessively hot I would check hat fans are working correctly.  Your description makes it sound like they might not be.

I have reshuffled the power and Sata cables completely so its not possible that it could be a specific faulty cable. With the problematic disk disabled everything seems to be working fine and disk temperatures are also normal.

If a drive fails pre-clear, I doubt the manufacturer of the HDD would consider it as reason enough for a warranty replacement. Or would they? Becuase wouldnt it work normally for the 99.9% of the pouplace who would take the drive and run Windoze on it?

 

And if it were just one machine on which I was doing the unraid business it would be ok. But I almost build servers as a part of my profession. So its very time consuming for me to pre-clear SO many drives. Its less painful for me to just remove a problematic one, and slide in another. Btw, by sheer coincidence, the server that has given me the most trouble is indeed my personal server and not a client's thankfully!

 

I've never heard of a manufacturer refusing an RMA for any reason. Most of them offer advanced replacement so there is no way to refuse the RMA.

Since two bad drives seemed to be too unlikely to be a coincidence, I got a brand new power supply, took the whole server apart, aired everything out, and popped in everything with new sata cables and power connectors. When I turned the server back on, I enabled the "bad" disk, and now its rebuilding at the usual 100 mbps odd speed. Around the 75% mark now and everything is chugging along fine. Thanks for all the help.

your drive temps (according to your screenshot) are way high. check that your fans are all working and cables arent blocking the air flow or the fan.