My experience with HDD warranty returns has been that if you can show that if either
- the HDD itself is reporting it has a problem
- or that, as a data storage device, it is unable to successfully store data
then you can return it and have a replacement if it's within its warranty period.
For the 1st type use the HDD's built in reporting technology that most HDDs have these days, called S.M.A.R.T. monitoring. A tool like GSmartControl can tell you if one of the tests has failed. This is one of the things Seagate's SeaTools and Western Digital's Data Lifeguard Diagnostic looks at.
If the HDD fails the SMART analysis or tests then there is no need to proceed with gathering any more proof of it being a bad drive, as the drive itself is saying it is faulty. Most HDD manufacturers would prefer that you use their tool to read the SMART data if they have one, so that all their forms line up nicely with neat rows of tick boxes.
For the 2nd type of issue, you need to prove to yourself that the drive can't hold data. If the HDD is part of a RAID array with redundancy this is easy, and you can just tell it to scrub the array a couple of times. However for individual drives you need to
- format the disk
- write data to the disk until it either has an error writing(take note of how full it was when having the error, so you can tell the support people this), or is full
- if the disk could be filled with data without error then compare the data written to the source data taking note of which file on the disk has an error.
I find that simple scripts helps with the above better than a tool like the command line program chkdsk, as chkdsk spends too much time trying to recover data from problem areas. Simple scripts such as entering the lines below in to Windows command prompt are usually enough to test the whole disk.
for /L %n in (1,1,4000) do copy c:\somebigfile g:\fill\%n.dat
which will attempt to copy a file 4000 times to the folder "g:\fill" and number them 1 to 4000. I find that a 1GB source file is convenient to let you figure out where on the disk has an issue.
To verify that the files were written ok something like the below command line to compare all 4000 files to the source file
for /L %n in (1,1,4000) do @( fc /B c:\somebigfile g:\fill\%n.dat 1> nul 2> nul & if ERRORLEVEL 1 (echo %n is BAD ) ELSE echo %n ok )
Formatting the disk and writing new data to it will allow a disk that has "pending sectors" as listed by a S.M.A.R.T. program, to reallocate those sectors, thereby increasing the reallocated sector count and increasing the chances of the hdd manufacturer replacing the hdd under warranty.
However if the HDD can reallocate enough bad sectors during a first pass of the above process to make a 2nd pass succeed and still have enough spare sectors available for reallocation to still pass the SMART tests you are in a bit of a no-mans-land in terms of warranty. Some people have reported that they have been able to have replaced HDDs under warranty that have even one reallocated sector. Possibly due to some studies that show that HDDs with even 1 reallocated sector were significantly more likely to die sooner rather than later.
The above scripts won't find errors caused by excessive head seeking as it is all sequential, unless you are also reading/writing and deleting other files to that hdd at the same time.