Badblocks test of 1.5TB drive over USB

[cheaper_popcorn]

Time elapsed: 56 hours and counting... Anyone want to guess how much longer it will take?

String used

badblocks -n -v /dev/sde

[nuke12]

USB? A man that likes pain?

[cheaper_popcorn]

YAAAAAARGH, let me out of my misery already, have I not suffered enough?!?

Seriously, is there a way of interrupting or cancelling the process that does not risk damaging the hard disk? I thought this would take a maximum of 48 hrs, and need to get cranking on moving files and installing an operating system.

[nuke12]

This is the non-destructive read mode? Correct? Is there data on the drive?

---

CTRL-C in the shell where badblocks is running should stop it or in another shell, type "killall badblocks" should do it. Your not going to hurt the drive but badblocks won't have compleated the job, you set it out to do.

[cheaper_popcorn]

Yes, it is nondestructive mode, and the drive is fresh from the shop with no data on it.

Thanks, I think I will interrupt it and just assume that the disk is ok. I got a pointer that since no bad blocks have been listed so far, it means that up until this point, the program has not managed to find any bad blocks, which at least is a little reassuring.

[nuke12]

Well on a new drive, I would have used the destructive write mode (better test) and for god's sake, not over USB. The test is slow enough via SATA.

[xor]

Has anyone ever found a bad block in the last few years with these sata smart drives, using badblocks? In my observation, these drives work fine and then quit without warning if you use tools that don't actually work with the smart controller. The smart drives actually do sector reallocation on the fly and mask any problem the drive may be having until they run out of spare sectors. I can understand using badblocks on older scsi, ide(or pata as they are now called), and flash drives. If I am wrong educate me.

[cheaper_popcorn]

Noted, but I am still a ix/ux n00b... practice and frustration makes perfect.

I tried your killall method which appears to have worked fine. Out of foresight it appears the coders have set badblocks to output at which block it is interrupted, so with that info I can now calculate approx. how much longer I have to go if starting it again (it is possible to specify start and end block so I wont have to redo the whole thing).

Thanks a lot.

[nuke12]

@xor - in a perfect world, you are correct but;

1) the smart controllers only do their thing on the fly and normally with data. I still like to test new drives before I get data on there.
2) Strictly speaking, badblocks only tests for daaaaa, badblocks but when you run it on a new drive it exercises every function of the drive and also makes it run hot for a long period of time. And yes, I have found problems on new drives using this very old program. One I remember, the drive started out doing it's read/writes very fast but as time went on it kept slowing down.

@cheaper - I don't recall being able to do that but I don't think I've ever tried. You'll have to check the switches.

[xor]

So badblocks can exercise a new drive and make it run hot, and you found a drive that ran fast and slowed down when using badblocks. Did badblocks actually find any bad blocks though?

[cheaper_popcorn]

Total number of blocks on HDD: 1465138583
No. of blocks checked up until interruption: 1224322432
Blocks remaining to be checked: 240816151

Time spent to check 1,224,322,432 blocks: Approx. 63 hours

Gives a speed of 19,433,689 blocks checked per 1 hour.

Which means approximately 12.39 hours left to go... Lucky I interrupted the bugger.

[nuke12]

(07-16-2010 12:23 PM)xor Wrote:  So badblocks can exercise a new drive and make it run hot, and you found a drive that ran fast and slowed down when using badblocks. Did badblocks actually find any bad blocks though?

Nope. Lets look at this another way. You run badblocks, it does reads/writes to every sector. So now, you also have this smart controller that is watching what badblocks is doing. If badblocks starts having a problem, the smart controller should step in and reallocate the sector. Either way, the problem is picked up before you put real data on there.

As for the drive I mentioned, I exchanged it for the exact same model and the second drive worked fine. I assume heat was causing something to go wrong.

[xor]

(07-16-2010 12:34 PM)nuke12 Wrote:  Nope. Lets look at this another way. You run badblocks, it does reads/writes to every sector. So now, you also have this smart controller that is watching what badblocks is doing. If badblocks starts having a problem, the smart controller should step in and reallocate the sector. Either way, the problem is picked up before you put real data on there.

As for the drive I mentioned, I exchanged it for the exact same model and the second drive worked fine. I assume heat was causing something to go wrong.

I saw the "other way" also, I guess my original assertion was vague. I was trying to get across the point that badblocks as designed does not do what noobs think it does(when run against smart drives). Non-noobs use badblocks for the side-effects it has on disks. For that matter I could use "dd if=/dev/random of=/dev/sda" and achieve somewhat the same type of exercise on the drive.

[nuke12]

(07-16-2010 12:41 PM)xor Wrote:  I saw the "other way" also, I guess my original assertion was vague. I was trying to get across the point that badblocks as designed does not do what noobs think it does(when run against smart drives). Non-noobs use badblocks for the side-effects it has on disks. For that matter I could use "dd if=/dev/random of=/dev/sda" and achieve somewhat the same type of exercise on the drive.

Yep, but I still like badblocks best because when older drives start acting up the smart controller can run out out of spare sectors and then you will see them with badblocks.

I've also seen complete drive failure running badblocks on a new drive. It just died. Nothing.....no spinning, no actuator movement but your command would probably have achieved the same thing.

I like, if at all possible, to test all new drives. Badblocks, the mfg's destructive write tests, something.