Although I have trouble following what numbers you see increase / decrease see if this helps explain what you observe:
Reallocations RAW value: Can not go down unless using a tool than can edit SA (service area) of the drive or due to corruption of the SA. Vendor Specific Commands may allow SMART values to be reset entirely, Victoria can do this for some drives.
Pending RAW: Can increase or decrease: A pending sector can be put back to service (RAW value goes down) or can be reallocated (RAW value goes down). If the latter then RAW for reallocations will increase.
Reported Uncorrectable: Appears to be an event counter, not number of sectors counter. IOW if you send multiple read command to the drive to read a specific sector, the number will increase not because more 'bad sectors' were found, but because it runs into the same sector over and over. As long as a sector is 'pending' each read of the sector will result in a reported uncorrectable event.
At first I used Victoria 5.3.7, which healed 2 sectors before i turned
off the full scan.
I suggest you let it finish a full scan.
BTW, tools like Victoria don't heal sectors. What can happen is:
- Victoria (or some other tool) reads a sector
- Drive can (A) or can't read (B)
- If B then drive will try recover the sector. It can (N) or can't (M)
- If N it may write back in-place or reallocate the sector immediately ('healed'), if M the sector will become pending
- A pending sector will be re-allocated on write and replaced with a spare ('healed').
I reality nothing is healed. Either sector was recoverable and good enough to keep in use or it is simply replaced.
Decreasing RAW value for reallocate!?
IMO is would defy logic and contradict everything I ever read on the topic. But appears it is happening on your ST2000DM008.
Only explanation I can come up with is this:
This (ST2000DM008) is a SMR drive if I am not mistaken.
This means the drive does dynamically map LBA sector addresses to physical addresses. Which is in fact what sector reallocation does as well mapping a LBA address to a spare sector.
So both these mechanisms map LBA addresses to physical sectors. It's not at all impossible for the drive to simply stop mapping to bad sectors without having to rely on a more or less fixed spare pool. And even apply this is to sectors already in the reallocation list.
If this is what's happening you might consider this cheating and another reason to stay away from Seagate SMR drives.
Bonus: I made a little calculator for the RAW values on Seagate drives for:
1 Raw_Read_Error_Rate
7 Seek_Error_Rate