The downside of large disks

Some time ago I've blogged about the downside of large disks being long the huge amount of time that it needs to check them with badblocks for medium errors. As I told there, I bought two Seagate Barracuda ES.2 ST31000340NS and most of you might have

I bet they are so "slow" because they are remapping bad blocks with "spares". You won't see any errors via badblocks until the disk(s) run out of blocks for remapping...

You can continue running badblocks until this happens, or... use smartclt to list the errors that the disk has encountered.

True as well, but as already stated, the badblocks test I'm running is about finding out *if* there are already serious errors that would cause me to *immediatedly* return the drives to the dealer.
Of course the smartctl test will be run as well, but after the badblock run has finished.

And yes, I already had a drive that died within a day with lots of badblocks. So I choose that way to go... ;)

True. Although each of the disks have started with >100 MB/s (0xaa test) which dropped down to approx. 80-85 MB/s in the last test (0x00 test) for some unknown reasons.

So, using 6 hrs per test will give me 5 hrs x 4 test x 2 times (write and read/compare) = 40 hours for a complete run.

It's faster than that so I will end up with roughly 30 hrs for a complete write and compare test run with badblocks.

But what I'm curious about is the performance drop during 0xaa and 0x00 test within one day (-20 MB/s)...??

AFAIK, badblocks won't actually tell you anything useful at this stage in a drive's life. You need to ask the drive to do its own internal test run. You can do this with the 'smartctl' program from 'smartmontools':

# smartctl -t long /dev/hda

Then wait until the test completes. Once it does, you can view the errors with:

# smartctl -l selftest

You may have some issue. badblocks should run against a modern desktop drive at 60MB/s. That's 4.8 hours for 1TB.