Bad sectors

[#1]

I Guess the ideal situation is to have no bad sectors at all on a HDD but if they are starting to appear is there a figure which would start to get concerned about ?
If you were talking about a New HDD is there a figure that would allow you to claim under warranty ?

[#2]

Any increasing number of bad sectors usually rings alarm bells for me. If you have one or two and there haven't been any more for months then you may be ok but any time where they start increasing, backup, panic buy a new HDD then backup again !

[#3]

In theory, the "normalised" values from the drive itself are supposed to tell you when it's failed or not.
e.g. values from 26 May 2010
[tt]ID# ATTRIBUTE_NAME          FLAG    VALUE WORST THRESH TYPE      UPDATED  WHEN_FAILED RAW_VALUE
  5 Reallocated_Sector_Ct  0x0033  090  090  036    Pre-fail  Always      -      413
  9 Power_On_Hours          0x0032  100  100  000    Old_age  Always      -      62[/tt]
413 reallocated sectors, but the normalised value is only down from 100 to 90, nowhere near the 36 failure threshold.
Current figures:
[tt]ID# ATTRIBUTE_NAME          FLAGS    VALUE WORST THRESH FAIL RAW_VALUE
  5 Reallocated_Sector_Ct  PO--CK  003  003  036    NOW  3975
  9 Power_On_Hours          -O--CK  092  092  000    -    7421[/tt]
The drive has used up almost all its spare sectors, the number crossed the failure threshold around the end of March 2012, yet actually the drive still works, and I did wipe everything off it then copy everything back onto it at the time.

[#4]

Once a sector has been marked bad it is never recovered.
Well that was true in the days of SCSI - I don't know about SATA.
You should get concerned if you get sudden changes in error counts over a few days.
Make your backups more frequently.

[#5]

They can be marked as pending, which the drive is supposed to deal with at a later date where they can be marked as bad (reallocated) or returned to use.
I've had a few drives get bad sectors. One was a Hitachi which marked one on the initial full format - used it heavily for 3 years and no more ever appeared. Likely to have been a manufacturing flaw with that sector.
Another was a Western Digital Green. It marked a load of sectors as pending, then during a SMART offline test it decided they were okay. After running badblocks it changed its mind and marked them pending again, after which a SMART offline marked a few as bad and return the rest to use. This went on for a few cycles of badblocks until enough sectors got reallocated to get an RMA (from memory, it was several thousand). I suspect it was head failure rather than the platters due to the randomness and inconsistency of it.
Another was a Samsung Ecogreen. It marked one sector as pending, and later decided to return it back to use without reallocating it. Several days of badblocks failed to prompt it back into a failed state again. That sector might be a little weak, so could fail again in the future, but I couldn't make it.
Finally a Seagate 7200.9. One sector marked as pending during day to day use. Ran SMART offline test. Sector was reallocated. No more appeared for several years until retirement.