Six WD20EARS-00J2GB0 drives - all bad?

[Rainbow]

I was given an decommissioned storage "server" (based on a desktop PC) with six 2TB WD green drives connected to a 3ware RAID card. They're WD20EARS-00J2GB0 (Dragon Fly 4 4K LT). I planned to sell them or use as data drives for PC with small SSD but found that every single one has a problem. There are pending sectors in SMART but rewriting them does not help (auto reallocation does not work). I've zero-filled all drives using dd in Linux and the pending sectors count remains the same.

User area scan in WD Marvel shows thousands (e.g. 20000) of sectors over 1000ms (mostly around 1900ms) and also some UNC sectors (inserting them manually into G-List decrerases pending sector count in SMART).
One drive has probably bad head 0 (scan very slow, showing errors on almost every sector on head 0).

I've fixed WD800JD drives before using selfscan in WD Marvel (after PCB repair) but self-scan seems not be working here. It starts but ends too early with user area inaccessible (probably because self-scan script contains references to module not present in module directory).

Is there any hope to fix these drives? Maybe ARCO (never tried that)? Or just throw them into trash as they're utter crap?

[Rainbow]

The self-scan script is very short, only 8 entries (sorry, I've confused it with another newer 4TB WD drive where it referenced non-existing modules). I've found some self-scan scripts - there's DragonFly4 ARCO, but I don't know if it's safe for DragonFly 4 4K.

[Rainbow]

I've tried "DF4 Complete Original ARCO Script" on the worst (bad head 0) drive. It ended after 45 minutes with BG_PST_UNABLE_TO_LOAD_PTM error. Now the drive does not work at all. The model name is empty. When I load overlays from modules backed up, the name changes to ROM ID. But it's not able to initialize directory so the modules cannot be restored.

[arztt]

I've tried "DF4 Complete Original ARCO Script" on the worst (bad head 0) drive. It ended after 45 minutes with BG_PST_UNABLE_TO_LOAD_PTM error. Now the drive does not work at all. The model name is empty. When I load overlays from modules backed up, the name changes to ROM ID. But it's not able to initialize directory so the modules cannot be restored.

First write back ROM, then restart drive.
Now 'Emulate SA Translator (ROYL)' and 'Restart FW and block loading FW', load overlays->'yes'
Write back all modules and restart drive

If dir can't be initialized, use 'Write and initialize dir', load overlays->'yes'
Then restart, 'Emulate SA Translator (ROYL)' and 'Restart FW and block loading FW', load overlays->'yes'
Write back all modules and restart drive


If self scan and arco fails, just cut the bad heads and change them to 1.5TB drives.
Weak heads on these drives are sometimes hard to detect, because they work good when the drive is warm.
Do a write test for some minutes when the drive is cold. If a head is weak, it will write garbage. With a read test afterwards check if all written sectors are readable.
Or look in G-List, which heads have most reallocations.

[Rainbow]

Oops, I've lost the ROM backup (overwritten by accident). I think it could be rebuilt from modules.

[fzabkar]

What software tools do you have?

[diybit]

I dont know exactly for 00J2GB0 drives, but I can say you that selfscan works fine for others old WD20EARS like 00MVWB0 or others old Green.

I suggest you just trying the C4 46 ARCO recalibration, that can be easily started with old WDMarvel 2.3 (verify you get this C4 module)
Then done an SS without ARCO steps, that will recreate your P-List. (like with the SS "WD all" from hddoracle)
And then just depop the bad head(s) if drive is always problematic.
So I suggest you not trying anything more because it will probably be a lot of wasting time...

You are saying the drive have a lot of bad spots and so you are risking to brick it if you do others ARCO tests like the one that recalculate sectors by tracks, that will probably fall in error.
Hotswap can be a way to resurrect it if you fall in this situation...

I think fzabkar will help you to rebuild your ROM.

Can you share this "DF4 Complete Original ARCO Script" and others you have found please!?

[Rainbow]

I have WD Marvel. There's an option for rebuilding ROM but it seems to require working SA access - I don't see an option to load module files.

The script is included in "Spildit WD SS Scripts Collection.rar" located here:
http://www.hddoracle.com/viewtopic.php?f=176&t=1582

[Rainbow]

Thanks to arztt, I've used 'Emulate SA Translator (ROYL)' and wrote back all modules. The drive 'works' again (the same way as before) without touching ROM.

Both "Create ROM image from copies in SA" and "Create ROM image from copies in folder" (I was wrong before about this function not present, sorry) fails the same way "Copies of ROM modules are empty!". Modules 102-106 are really empty (contain strings" "ROYL and "NOT_INIT"), 109 seems to be OK.

[fzabkar]

I tried using the demo version of WD Marvel to rebuild a ROM from the files, but the resultant ROM was corrupt. I have now written a tool to do this, but it requires good copies of the 0x10x modules. These are only present when the ROM is embedded within the MCU.

I think most tools try to rebuild the ROM from 0x109, 0x40 and 0x35.

[arztt]

If the drive is detected again, you may not need to write back ROM. Maybe you are lucky and the ROM was not touched. But normally in Arco process WDMarvel will write a default ROM.

If you click on Read Flash while there already is a flash backup in folder, the program will keep both copies.
But modules will be overwritten when reading again.

[Rainbow]

Now I know that I should copy both modules and ROM to a safe place before doing anything.

Ran self-scan again, this time the first ARCO Main code passed Now it's running ARCO TPI Calibration.

[arztt]

Never tried ARCO scripts.

I always use the the built in ARCO process in WDMarvel and it's quite successful.
You can even use Firmware and ROM from a different drive and repair drives where original PCB is missing.
On this drive family it needs 2-3 days to finish.

[Rainbow]

The process was aborted two times because of test system's own HDD (Maxtor/Quantum) failure. Now it runs with replacement drive.
ARCO TPI Calibration passed with two errors (one DISK_DAM_ERROR and one unknown 53D8) after almost 20 hours :shocked:

[Rainbow]

It seems to be stuck at "ARCO VDT HeadQual" for 5 days.
State = 08 00 Unknown state
Last VSC command = Execute DVT
Last command end code = 00 Success

Track = 9614 (changes to some other number once in 5 seconds and returns back)
Head = 3
Zone = 1

Drive produces repeating quiet seek sounds.

[diybit]

SS steps can respond with a "success" even when there are error so be careful... about all SS I have seen have a bad errors handling and pass to next steps even when error happen, so I suggest to modify those SS to be able to right follow what is going on.

Another thing I have forget is, depending of your tool, you can also disable the defective zone(s) instead trying refurb a drive by disabling an entire head.

The thing you describe with a return back to the same position for about 1-2 hours means the step have fail.

Does your drive is now bricked?

[Rainbow]

Aborted self-scan and power-cycled drive. It is detected, although any access to user area is aborted.

Looking at the script, the step 4 (where it stopped) has "if error" column set to "stop" - I guess that's what happened.

Maybe I should try it with one of the better drives.