My 6+ yrs old 3TB external USB Seagate Barracuda ST3000DM001 had gone bad. I used it infrequently and unplugged the power and USB afterwards. The drive was always flaky with the initial connections and had occasional slow file copies but otherwise working. It finally became unrecognizable by the OS after sitting in storage for over a year. It was originally a single 3TB GPT partition formatted as NTFS. After failing it showed up with a wrong drive geometry (a NTFS partition of 375 GB only) and was unmountable/unrecognized in Windows and OS X.
To troubleshoot and save time, I took the drive out of the enclosure and hooked it up directly as SATA. When trying to rewrite the correct drive geometry, Testdisk-7.1-WIP told me that the drive wouldn't let me rewrite the partition ("Partition: Write error"). However testdisk could "see" the data fine after entering the right geometry. The only thing left to do was dumping the drive as a 3TB image to recover however much data possible.
The whole image process took 17 days (~407 hrs)... The destination USB drive was fairly new and in good working condition so no problem there. I ended up with a 3TB image with nearly all the files, as far as I could tell. There were a few hundred input/output and fixup_warn errors in the log, but not sure what they really meant. Only a handful of the files were corrupted or turned-binary in the image when I test-read them.
I then left the drive connected(SATA) in the computer for two months. I got another external USB drive last week and re-tried the image dump so I can unplug and retire the bad drive. To my surprise, not only did the imaging only take about 45 hours, but it also got a few more files than the previous dump!
I did a hash comparison with the files from the first dump, and found that the second time things only improved. The two dumps were about 99.9999% the same, and the difference were the previously missing, corrupted, or turned-binary files, now in perfectly good recognized and readable state.
When I tried a surface scan using Techtool Pro right after the 2nd imaging, the scan only took 26 hours and returned no bad sectors. Two month ago it was reading extremely slowly (bad sectors?) and would probably take weeks to finish. I couldn't wait for it so can't give a time comparison on the scan.
The SMART reading for the drive has tons of reallocated sectors and some pending and uncorrectable counts.
Code: Select all
ID# ATTRIBUTE_NAME FLAG VALUE WORST THRESH TYPE UPDATED WHEN_FAILED RAW_VALUE
1 Raw_Read_Error_Rate 0x000f 116 064 006 Pre-fail Always - 206712440
3 Spin_Up_Time 0x0003 094 092 000 Pre-fail Always - 0
4 Start_Stop_Count 0x0032 100 100 020 Old_age Always - 430
5 Reallocated_Sector_Ct 0x0033 100 051 036 Pre-fail Always - 1144
7 Seek_Error_Rate 0x000f 070 060 030 Pre-fail Always - 11830910
9 Power_On_Hours 0x0032 095 095 000 Old_age Always - 4902
10 Spin_Retry_Count 0x0013 100 100 097 Pre-fail Always - 0
12 Power_Cycle_Count 0x0032 100 100 020 Old_age Always - 307
183 Runtime_Bad_Block 0x0032 099 099 000 Old_age Always - 1
184 End-to-End_Error 0x0032 100 100 099 Old_age Always - 0
187 Reported_Uncorrect 0x0032 001 001 000 Old_age Always - 7823
188 Command_Timeout 0x0032 100 099 000 Old_age Always - 6 6 9
189 High_Fly_Writes 0x003a 100 100 000 Old_age Always - 0
190 Airflow_Temperature_Cel 0x0022 064 027 045 Old_age Always In_the_past 36 (Min/Max 35/41 #2152)
191 G-Sense_Error_Rate 0x0032 100 100 000 Old_age Always - 0
192 Power-Off_Retract_Count 0x0032 100 100 000 Old_age Always - 154
193 Load_Cycle_Count 0x0032 099 099 000 Old_age Always - 2903
194 Temperature_Celsius 0x0022 036 073 000 Old_age Always - 36 (0 18 0 0 0)
197 Current_Pending_Sector 0x0012 100 069 000 Old_age Always - 0
198 Offline_Uncorrectable 0x0010 100 069 000 Old_age Offline - 0
199 UDMA_CRC_Error_Count 0x003e 200 200 000 Old_age Always - 0
240 Head_Flying_Hours 0x0000 100 253 000 Old_age Offline - 873h+54m+29.237s
241 Total_LBAs_Written 0x0000 100 253 000 Old_age Offline - 29472053552965
242 Total_LBAs_Read 0x0000 100 253 000 Old_age Offline - 259894645155740
So what kind of physical drive problem might fit this behavior? I don't understand why it had such difficulty reading the drive two months ago but somehow way smoother now, even though still not trouble-free. I tried creating incomplete images multiple times back then in a span of several days, but the slowness was always consistent (very slow at some %s). Both image dumps and surface scan were done as an internal SATA drive. Why did the two runs differed so much? The computer and the drive weren't running hot, but always-on 24-7. I only reboot the computer maybe twice in between, and the drive was never recognized or mounted, although I am not sure if the spindle was spinning. Other hardware and OS variables were pretty much the same as far as I remember.
Thanks.