weird seagate RMA - 3rd bad replacement drive???

[venm11]

So, my 7200.11 1.5Ttb was failing and I tried to RMA it with Seagate. That was a month ago. They sent me a 2tb Constellation drive, which was about 50% bad blocks, per chkdsk. I talked to seagate and they sent me a 2nd 2tb Constellation which failed to even spin up. I sent both those back and today a 7200.11 1.5tb arrived, which I immediately tested, and it also has a huge number of bad blocks.

But I noticed something curious in the number of bad blocks of the 2 drives that worked. According to my chkdsk attempts, they both had 1073GB of "good" blocks. Is it possible that there's some compatibility issue with the PC I'm testing them with (an old Sata 1 intel D865GLV mobo), such that blocks above 1TB are detected as "bad" and so marked? I had no such problems with the original 7200.11 1.5tb on this machine, though. Maybe the newer drive firmware is incompatible?

Replacement #1 (Contellation 2tb):

1953512000 KB total disk space.
0 KB in 1 files.
4 KB in 9 indexes.
879764200 KB in bad sectors.
125604 KB in use by the system.
65536 KB occupied by the log file.
1073622192 KB available on disk.

4096 bytes in each allocation unit.
488378000 total allocation units on disk.
268405548 allocation units available on disk.

Replacement #3 (7100.11 1.5tb):

1465136000 KB total disk space.
0 KB in 1 files.
4 KB in 9 indexes.
391370848 KB in bad sectors.
110700 KB in use by the system.
65536 KB occupied by the log file.
1073654448 KB available on disk.

4096 bytes in each allocation unit.
366284000 total allocation units on disk.
268413612 allocation units available on disk.

The "good" block counts aren't exactly the same, but they're similar enough to wonderWTF !??

 

[drescherjm]

Is it possible that there's some compatibility issue with the PC I'm testing them with (an old Sata 1 intel D865GLV mobo), such that blocks above 1TB are detected as "bad" and so marked?

Yes it is possible that chkdsk is calling bad sectors that are not bad. Actually chkdsk should almost never see a bad sector since the drive should automatically remap those with good ones from its spare pool.

I would not look at chkdsk output to see if a drive is good or bad. Look at SMART instead. And yes 3 out of the last 4 RMA's I did for my department last year (I sent back 12+ drives) I got back bad drives from the RMA. I recommend thoroughly testing the drives you get back from an RMA before putting a drive back into service. At minimum do a full format or even better do a 4 pass badblocks test (available on any linux livecd).

 

[venm11]

Well, chkdsk is one of the things I use to to thoroughly test the disks (as a means of surface scan). I do not trust Seatools because it does not give such specific information and it doesn't fail drives that should be failed.

What do you mean "full format"? LLF? A windows format will accept the bad blocks and leave me with a "healthy" 1TB partition (despite disk physically 1.5tb or 2tb).

SMART indicated that both replacement drives were basically good but had been subjected to high heat (in the "yellow" state per the smart tool I'm using). I kind of expected them to reset the smart statistics on these drives after "repair". Considering all the problems, the appearance is that they just send you out someone else's returned drive to test.

 

[venm11]

I suppose the real question is: what tool should I be using at this point?

 

[drescherjm]

What do you mean "full format"? LLF?

I mean do not check the quick format box in windows and make windows write every single sector. A full format should take several hours.

I do not trust Seatools because it does not give such specific information and it doesn't fail drives that should be failed.

At work I have seen times where very bad drives have passed manufacturer tests.

 

[drescherjm]

I kind of expected them to reset the smart statistics on these drives after "repair".

If there are real physical defects the drives should record these in the reallocated sectors count. Use a program like CrystalDiskInfo to look at the raw results for that.

 

[klightspeed]

When you inserted the drive, was there already a partition present?

If so, then it's possible that either Seagate did not wipe the HDD (so it still had the original bad filesystem present), or they created a NTFS filesystem with blocks above about 1TiB marked as bad.

If the partition was not already present, then if those blocks were truly bad, it would have taken you 5 years or more to mark them all as bad - 879764200 KB = 219941050 blocks = 7 years at 1 block per second.

edit: It's also quite possible that there's a bug in the SATA driver or the BIOS for that 7 year old motherboard where it uses a signed integer to address sectors - 1073654448 KiB = 2^31 - 174752 sectors.

edit 2: The 7200.11 drives are Desktop SATA drives, while the Constellation drives are Enterprise SATA/SAS drives. Surely the Intel driver isn't trying to use the Read/Write 10/12 SCSI commands, which only support 32-bit LBAs, and taking the LBA as a signed 32-bit integer? Have you tried the drives in another computer or in an external drive enclosure?

Try:
* From Control Panel -> Administrative Tools, run Computer Management as Administrator
* Under Storage, select Disk Management
* Right-click the bad volume and select Delete Volume
* Once you have deleted the volume, right-click the unallocated space and select New Simple Volume
* Next Next Next Next Finish

 

[Bl4ckM0onk3y]

Seagate as any other company that has a mass production of any kind only tests 1 out of 10 units in a assembly line, so it is very possible to get a malfunctioning drive. Curiously i never did a full format on my drives and got any problems, maybe it's because i'm using Samsung ;]

 

[mwroobel]

Seagate as any other company that has a mass production of any kind only tests 1 out of 10 units in a assembly line, so it is very possible to get a malfunctioning drive. ]

Uh, are you kidding me? Where in gods name did you read that and why did you believe it enough to repeat it? That would be like saying Intel only bothers to test 1 out of 10 die that come off a wafer.

 

[Bl4ckM0onk3y]

Every drive gets basic tests, but its the same as any other manufacturer, no time for thorough tests in the assembly line ;]
Everyone would assume that every drive is 100% tested, it is so but only 1 out of 10 or 1 out of 100 is put thru torture test :]
Tried my best to put it to words ;] don't want to spread some kind of heresy.

 

[Old Hippie]

Seagate as any other company that has a mass production of any kind only tests 1 out of 10 units in a assembly line,

He's probably getting refurbed drives.

AFAIK there's no assembly line involved.

 

[Bl4ckM0onk3y]

Refurbished drives are not worth the trouble, some time ago had two refurbs from seagate, they all had unbearable whine from the drive motor, and the retailer that sent them told the clients that they are "new" ;]

 

[mwroobel]

Every drive gets basic tests, but its the same as any other manufacturer, no time for thorough tests in the assembly line ;]
Everyone would assume that every drive is 100% tested, it is so but only 1 out of 10 or 1 out of 100 is put thru torture test :]
Tried my best to put it to words ;] don't want to spread some kind of heresy.

You are still wrong. Every drive that comes off the assembly line has every standard sector checked during the low level format(a real low level format, not what people here call a format), that is how the low level diags find, identify properly and remap any initial bad sectors which are written to the map. They also lay down hard boundries to the sector map at this point. The other hardware also go through electrical diags multiple times. When the PCBs arrive they pass a basic test, and after they are affixed to the drives they go through still more tests. They probably pull 1 out of every 5,000 or 10,000 drive run for more detailed and selective tests, but not 1% as you proposed.

 

[Old Hippie]

Refurbished drives are not worth the trouble, some time ago had two refurbs from seagate, they all had unbearable whine from the drive motor, and the retailer that sent them told the clients that they are "new" ;]

Some are good....some are bad...but 95% of the time that's what you'll get with an RMA.

 

[patrickdk]

Normally I can tell easily when I get a refirb drive vs a new drive from an rma.

So far, every refirb one I have gotten has died within 3 months, if it even worked on day one.
I have yet to receive a working rma drive from seagate.

WD though, I normally get new drives as replacements, though lately they keep replacing my 512b sector disks with 4k sector disks.

I'm getting ready to rma my first hitachi disk here, wonder how they will handle it.

 

[mwroobel]

Normally I can tell easily when I get a refirb drive vs a new drive from an rma.

So far, every refirb one I have gotten has died within 3 months, if it even worked on day one.
I have yet to receive a working rma drive from seagate.

WD though, I normally get new drives as replacements, though lately they keep replacing my 512b sector disks with 4k sector disks.

I'm getting ready to rma my first hitachi disk here, wonder how they will handle it.

99% of the drives I get back from Seagate and WD have new labels that say recertified on then, so it is easy to tell.