Hello every one ,

I just want to know that whether badblocks utility for checking hard disks bad sectors is reliable or not because it happens one time that it said no errors about one hard disk but there were still some issues with that hard disk. so with what parameters it will be reliable to use to check hard disk issues.


Thanks,
fahadaziz.

I think if it were particularly unreliable you'd probably find lots of comments about it on the net. I don't see anything of that sort.

You don't specify what your "issues" were, but you do realize that bad sectors are not the only problem that a hard drive can have, don't you? Have you tried running smartctl (part of the smartmontools) on the drive, for example?

@ David the H,

Thank you for your reply, The issue which I wanted to mention was that one of my colleague ran badblocks utility on one of our server's hard drive and badblocks reported that there is no issue in that hard drive but the problem was that there were still some issues such as its RAID was broken.

Now I am checking a lot of hard disks for further re usability and the steps which I am taking are as follows

1: fdisk -l
2: smartctl -a /dev/sdb
3: badblocks -wvs /dev/sdb.

In order to run some tests and to check its health so that we can re use these hard drives for further uses.

Thanks and regards,

Fahad Aziz.

I would make that

3. fsck -fcvy /dev/whatsit.

The 'c' calls badblocks as part of a filesystem check. Don't try and run two of these together - run them in line.

You mention that the disk's RAID was "broken"? Is the disk you are trying to check bad blocks on a physical disk or a volume made from a number of disks. Servers usually access disks through hardware RAID cards with a RAID setup utility available during POST. This is true of HP.Compaq Proliant servers, I assume DELL and others are similar.
You wouldn't run a bad block utility on a disk volume as the RAID card manages bad blocks, RAID rebuilds, etc, transparently, i.e. the operating system is unaware of the rebuild though there is an access overhead due to the extra work the RAID card is doing. You can adjust the rebuild priority when you set up the RAID. HP hot swap disks also log bad block info so that when a specific threshold is passed the disk is marked as "Degraded" and a warning led flashes on the disk, you can replace it then or, when it goes solid indicating a failed disk - you just pull it and replace it with a new one at this point... The RAID card rebuilds the data onto the new disk.
What? It's a software RAID? OK, forget all that, it's irrelevant!

bad blocks? Sorry, I just can't help becoming anal! disks have a hidden bad block file which is used to map any bad areas on the main surface of the disk. There are usually two spare cylinders within the disk geometry which are used to provide good storage. During a seek, the bad block file intercepts the address of a bad block and re-vectors the head to a block within the spare cylinder area. When you run a bad block utility the data is read from each block, known data is then written to the block and then verified by reading back, if it doesn't match, the block is marked as bad and the original data is written to one an area within the two spare cylinders. It's location is marked in the bad block file along with the original address so it can be accessed as if it hadn't been moved. If the block verify was good, the data is written back to the original block. Aren't you glad I told you all that? Er.. No, I thought you may have just given up and gone to the pub instead!
Oh! once the bad block utility has fixed the bad blocks, hopefully, if you were to re-run it it would be clean. Run it again a couple of days later, still clean? you may be lucky. If there are more blocks re-vectored I'd get a backup run as it looks like the disk may be on the way out!

Thats, my... er...


Play Bonny!