I have no idea whether this is now known here, but it seems that during february and march, many ST2000DL003 (Seagate Barracuda LP Green 2000GB) drives have failed for users all over the world. Mine died a few days ago (2 days after purchase), so I tried connecting to its monitor to see what happens (and maybe try the N1/m0,2,2... fix), so here's what I've found out:
The dead drive is:
I managed to connect to the disk's serial interface, and this is the diagnostic output it gives upon power-up. Hopefully someone qualified can now step up and tell us what it actually is that the ST2000DL003s are choking on.
(P) SATA Reset
SIM Error 1009
RW Error 00000080
User Data Base 00990DE8
Check MCMT Version: Current
MCMainPOR: Non-Init Case
MC Seg Disc and Cache Nodes: 4011982C 4011793C
Seg Write Preamble VBM start: 000010A7 end: 000010CE
Footer - start: 000010D0 end: 000010F7
Seg Read Preamble VBM - start: 000010F9 end: 00001120
Footer - start: 00001122 end: 00001149
Reconstruction: MCMT Reconstruction Start
Max number of MC segments 22E0
Nonvolatile MCMT sequence number 000070B0
Reconstruction: EXCEPTION: Segment Overall Sequence Number Mismatch
00004221 00000000 A
Recon Last Chance Header ID FFFFFFFF SeqNum FFFFFFFF Current
MC Internal LPC Process
Apparently the firmware is running into an assertion based on something it reads from the platters when starting up.
The bad news is there's no way to fix this using the "seagate fix" described all over the various Internet boards (that is, issuing commands N1 to reset the SMART log and m0,2,2,0,0,0,0,22 to re-format the user partition), because one never gets to the monitors prompt (ctrl-z won't work after the firmware has stumbled upon the exception). Insulating the contacts to either the motor or the heads (or both) doesn't help in this regard, because Seagate, as it seems, has crippled the electronics on these new drives by making the monitor (and presumably everything else) dependent on something that first has to be read from the platters. In other words, while older disks would communicate on the serial monitor even with the PCB totally disconnected from the motor and the heads, these new drives don't even start to log anything into the monitor until after they've spinned up and read a couple sectors from the platters. I tried taking the electronics off from several drives I have lying around (7200.9, 7200.10, 7200.11 and also one 2.5-inch drive), and this new drive is the only one that needs to read something from the platters before starting the serial monitor. I can't resist commenting on this observation... It simply seems that Seagate has made a pretty bad [cost cutting?] decision here.
Also, if you want to try to connect to your dead disk's monitor, too, note that:
- this new drive needs 5V to send/receive on the TX/RX lines (as opposed to 3.3V with all the older drives),
- while some older drives would communicate at 9600 bps, the ST2000DL003 communicates at 38400 bps,
- using a PCB from another, completely healthy disk doesn't help, because although the PCB will be able to spin up the drive, it won't then be able to read the initial couple of sectors (perhaps because of some platters-specific calibration data stored in the 512K Winbond flash?) and thus start the monitor.
Hope this helps someone eventually.
Edited by viinikala, 17 March 2011 - 04:24 AM.