• Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
    Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.

How do you test your HDDs for reliability?

E

evilsofa

[H]F Junkie
Joined
Jan 1, 2007
Messages
10,077
Slashdot did a really interesting thread on this question. One point I took away from that was that SMART implementation varies significantly. WD is discussed as a bad player on SMART, presumably to reduce warranty costs.

In one part of the conversation, it was suggested to have dban write out every sector and check smartctl for pending sectors. A reply suggested doing stuff with dd that I don't understand.

Is there a simple step-by-step for an effective reliability test of a new (or old) hard drive?
 
I use badblocks for a full write/read pass and then check SMART data through smartmontools afterwards.

How comfortable are you with Linux? Do you have a spare PC (or can dedicate a PC) to Linux while testing HDDs?
 
xnoodle said:
I use badblocks for a full write/read pass and then check SMART data through smartmontools afterwards.

How comfortable are you with Linux? Do you have a spare PC (or can dedicate a PC) to Linux while testing HDDs?
Click to expand...

+1, badblocks is pretty rough, but thorough. Note that if you use badblocks it will take a decent amount of time.
 
badblocks is a good program. I just did chkdsk /r on my a new drive I purchased recently.
 
I do not have a linux machine, but do have a spare laptop I could use to test few external USB HDD's I purchased around black friday.

I did some research, and it appears I can boot up with Parted Magic which has badblocks on it. However, is there a good tutorial on the best commands to use? Is it possible to hook up 3x USB drives and test them all at once?

Thanks!
 
xnoodle said:
I use badblocks for a full write/read pass and then check SMART data through smartmontools afterwards.

How comfortable are you with Linux? Do you have a spare PC (or can dedicate a PC) to Linux while testing HDDs?
Click to expand...

yeap...I do "mkfs -t ext4 -c -c -v /dev/something"

It takes a LONG time....
 
are either of these programs on the hiren boot disk or anything similar? last one i downloaded said "fix Ntldr missing" even ;) i also had this question going to try download those then after check hiren.
 
I've thought of this before but it feels like too much. I feel I'm paranoid enough as it is without testing my HDDs every month :p
 
I only test mine before the install. Recently I used Crystaldiskinfo and it was the only program out of a couple that cautioned me per the SMART test.

I then used HD TUNE 2.55 to test the speed of it and found out it was slow as a mule. I RMA'd the drive. It was refurb to begin with.
 
ziggo0 said:
+1, badblocks is pretty rough, but thorough. Note that if you use badblocks it will take a decent amount of time.
Click to expand...
Do you have a sense, for the purposes of this thread anyway, what the difference would be between using DBAN and doing, say:
badblocks -vws /dev/blah

I'm running that right now (is there anything the PartedMagic disk doesn't have?!), and no doubt it's way faster than DBAN, at least so far.

I pulled the following from the Slashdot thread, but am wondering what particular "registers" he's referring to? I'm looking at a smartctl -x report right now and am not sure if he simply means the SMART data or perhaps the somewhat mysterious "SATA Phy Event Counters" data at the end of the report.

I run smartctl and capture the registers, then run badblocks, and compare smartctl's output to the pre-bad-blocks check.

If there are any remapped blocks, the drive goes back, as the factory should have remapped the initial defects already, and that means new failed blocks in the first few hours of operation.
Click to expand...

Another post there further down said:
I do:
$DRIVE="/dev/sdb"
smartctl --xall $DRIVE > sm_data_0.txt; smartctl -t short $DRIVE; smartctl -- xall $DRIVE > sm_data_1.txt ; badblocks -vws $DRIVE; smartctl --xall $DRIVE > sm_data_2.txt
In addition, if I copy a significant amount of data to the drive I also recheck (and save) the smartctl ouput.

For larger than 1TB drives am am seeing lots of raw read errors on new drives, so I have to agree with other observations I've read about the larger than 1TB drive technology -- its not there yet
Click to expand...
 
In addition to rseller's question above, I'm using UBCD 5.1.1, which has Parted Magic 6.6, but I see that's over a year old now compared to Parted Magic 2012_12_25, so I guess I should use that instead. I'm reasonably comfortable using these sorts of tools, but when you start doing CLI commands like
$DRIVE="/dev/sdb"
or
dd if=/dev/zero of=/dev/sdX bs=8M
I get queasy because I'm not sure what that's going to do.
 
Yes, I'd always go with the latest.

I'd have to lookup what he's trying to with the if/of example you cited, but since I chose not to use dd, fortunately, I didn't have to. :)

But the example I quoted isn't a problem:

$DRIVE="/dev/sdb" --This is just defining the $DRIVE variable to the specific drive name on your system for the purposes of the script, before using that variable repeatedly later on. In my case, it was /dev/sb1. You can tell what yours is any number of ways, but one easy way is to just run the partition management icon on the desktop to see what it's called there.

I should correct what I said before: badblocks isn't "way" faster than DBAN, but somewhat faster. It looks like it's going to go about 24 hours on a 2TB drive.
 
rseiler said:
Do you have a sense, for the purposes of this thread anyway, what the difference would be between using DBAN and doing, say:
badblocks -vws /dev/blah
Click to expand...

I just did a bunch of reading and I found out that, and I quote, "The badblocks -wvs command will run four wipes on the drive, writing the patterns 0xaa, 0x55, 0xff, and 0x00 across the entire disk, respectively. It also performs a read test after each full write." It looks like my 1TB HDD will take 6 to 8 hours to complete that test. Meanwhile, you can choose what DBAN does, from a totally unreasonable 35-pass paranoia test that will take days to complete, to just a single pass of writing zeroes which will finish relatively quickly.

The man page for badblocks is here.

I also found this page, in which it is suggested to run badblocks and then the SMART long (extended) test. I did find occasional posts complaining that badblocks found bad sectors while SMART long test didn't, and vice versa, so it would seem to make sense to run both on a new drive.

So the process I'm working through now is:

1. Make a Parted Magic boot disk, and boot your PC with it.
2. Double-click "Disk Health", which opens GSmartControl, and it will be made as clear as it can be which drive is which. In my example, the drive on which I want to run read/write tests (which destroys all data!) is /dev/sdb, while the drives that have data that I don't want to lose are /dev/sda and /dev/sdc.
3. While still in GSmartControl, double-click the drive you want to test, then select the Attributes tab. Make note of the raw values of "Reallocated Sector Count" and "Current Pending Sector Count" (both are zero in my example).
4. Close GSmartControl and click the ROXTerm icon on the taskbar on the bottom to get a command line. Type
lshw -C disk
and you will get a list of the disks with info about each one, where you can confirm that you know which drive you want to do the test on by looking at the logical name.
5. Now for the badblocks command, which will take many hours. If you want to do a destructive read/write test on a new (or blank) drive with no data on (for example) the /dev/sdb drive, type
badblocks -wvs /dev/sdb
If you want to do a non-destructive read/write test on (for example) the /dev/sdc drive with data you don't want to lose, type
badblocks -nvs /dev/sdc
6. When it's done, go back to Drive Health (GSmartControl), double-click the drive you just ran badblocks on, select the Attributes tab and check the raw values of "Reallocated Sector Count" and "Current Pending Sector Count" to see if they have changed from before.
7. Click the Perform Tests tab and choose to perform the Extended Self-test, which should take several hours, and again check the Attributes to see if they have changed.
 
Ten hours later, badblocks had finished two passes and started on the third with no reported errors, so I aborted with control-c. Reallocated Sector Count and Current Pending Sector Count were still zero so I'm satisfied the drive is good.
 
We think alike! I gave up on it at that point too, not wanting to endure the next two passes.

We did pretty much the same thing otherwise: I ran the 3 tests in GSmart first though, since they're much quicker, and if the drive is bad you'll probably know it then without having to suffer through badblocks (note that you can save the Gsmart output to a file). But it checked out, so I then moved on to badblocks before finally revisiting Gsmart to see the latest Smart data.
 
stope21 said:
Heres an article with some more mainstream tools supported by the manufacturers:

http://www.jayceooi.com/2011/05/25/how-to-diagnostic-faulty-hard-drive/
Click to expand...

From the article: "Most of the hard drive companies like Seagate, Western Digital (WD) and SAMSUNG need you to check your hard drive condition with their own hard drive diagnostic tool before sending faulty hard drive for replacement (if still under warranty)."

From the slashdot thread I linked to in the OP: "Running the tool needed to generate an RMA does a full SMART scan of the drive, remaps some bad sectors, and then says everything is good. But it's not--each drive is still broken, in a way the firmware seems downright evasive about. Try to use it again, it doesn't take long until another failure. It does seem like the sole purpose of SMART and its associated utilities on WD drives is to keep people from returning a bad drive, by providing a gatekeeper in that process that never says there's a problem."
 
I plug my hard drives in. Do a quick format. Copy about 500GB of data to the drive.

If that works, I expect the drive to work as advertised.

---

Writing a whole drive 4 times when most of my drives are right once; never read, seems a bit much.

I guess if my data was worth enough I would keep several copies of the data around. I do that for my valuable data and have not had any valuable data lost in 20 years.
 
evilsofa said:
So the process I'm working through now is:

1. Make a Parted Magic boot disk, and boot your PC with it.
2. Double-click "Disk Health", which opens GSmartControl, and it will be made as clear as it can be which drive is which. In my example, the drive on which I want to run read/write tests (which destroys all data!) is /dev/sdb, while the drives that have data that I don't want to lose are /dev/sda and /dev/sdc.
3. While still in GSmartControl, double-click the drive you want to test, then select the Attributes tab. Make note of the raw values of "Reallocated Sector Count" and "Current Pending Sector Count" (both are zero in my example).
4. Close GSmartControl and click the ROXTerm icon on the taskbar on the bottom to get a command line. Type
lshw -C disk
and you will get a list of the disks with info about each one, where you can confirm that you know which drive you want to do the test on by looking at the logical name.
5. Now for the badblocks command, which will take many hours. If you want to do a destructive read/write test on a new (or blank) drive with no data on (for example) the /dev/sdb drive, type
badblocks -wvs /dev/sdb
If you want to do a non-destructive read/write test on (for example) the /dev/sdc drive with data you don't want to lose, type
badblocks -nvs /dev/sdc
6. When it's done, go back to Drive Health (GSmartControl), double-click the drive you just ran badblocks on, select the Attributes tab and check the raw values of "Reallocated Sector Count" and "Current Pending Sector Count" to see if they have changed from before.
7. Click the Perform Tests tab and choose to perform the Extended Self-test, which should take several hours, and again check the Attributes to see if they have changed.
Click to expand...

Awesome!!! Exactly what I have been searching for. Wish this could be stickied somewhere as this question always comes up.

Ordered over 16 HDD's this past 2 weeks, need to test them before the 30 day return is up.

Two Questions:

1) "badblocks -wvs" command automatically runs 3 passes before it stops?
2) Can I run the above command on more than one drive at a time? If so, how? (I plan to run badblocks through latest Parted Magic)

Thank You!
 
ZzBloopzZ said:
Two Questions:

1) "badblocks -wvs" command automatically runs 3 passes before it stops?
2) Can I run the above command on more than one drive at a time? If so, how? (I plan to run badblocks through latest Parted Magic)
Click to expand...

I can answer the first question regarding the flags you want to use with badblocks. It appears -w will run four different patterns on the drives. Remember this option erases data!

-s Show the progress of the scan by writing out the block numbers as they are checked.

-v Verbose mode.

-w Use write-mode test. With this option, badblocks scans for bad blocks by writing some patterns (0xaa, 0x55, 0xff, 0x00) on every block of the device, reading every block and comparing the contents. This option may not be combined with the -n option, as they are mutually exclusive.
 
I hook up the disk and let it spin till it dies or gets retired.
 
Gut instinct tells me if the drive may be having issues...

I test for bad blocks using HDTune if the drive is 50/50 questionable, I do a drive clone using Norton Ghost 2003 (boot from disk) and clone to a new drive with the /fro switch if i think it's not too bad a drive, or /ia switch if I think it's a really bad drive, or if /fro failed. This has worked for me 99% of the time, with the remaining 5% being electronic or mechanical failure, in which case it's lab time (if a 1:1 controller swap doesn't fix it).
 
Hi,

Using Crystal Disk Info I have a caution on the reallocated sectors count.

Current: 100 Worst: 100 Threshold: 5 Raw Data: 000000110037

All other counters ok

I ran a program recommended on here"smart mon tools" and get as follows:

reallocated sector count: 0x0033 100 100 005
55 (0,17)
Is this showing a reallocated sector count of 55?

What is the typical range of the reallocated sector count before the disk dies?

Is this an imminent concern?

Power on count: 666

Power on hours: 2620

Thanks,

Ben

p.s. how do you create a new post on this forum?
 
What is the typical range of the reallocated sector count before the disk dies?
Click to expand...

From the dozens of drives that I have seen a non zero reallocated sector count there is no number that signifies a drive will die soon. I have had drives with > 100 bad sectors run for years without issue. However what should concern you most is seeing this number grow. If the number grows daily ... it will die soon or at least this has always happened in my small non scientific sample size..
 
drescherjm said:
From the dozens of drives that I have seen a non zero reallocated sector count there is no number that signifies a drive will die soon. I have had drives with > 100 bad sectors run for years without issue.
Click to expand...

If the drive has reallocated sectors when it's fresh from the factory, all that this means is that the manufacturing process for the disk platters resulted in some small number of defects in the media that could not be fixed with error-correcting codes. Normal.

However what should concern you most is seeing this number grow. If the number grows daily ... it will die soon or at least this has always happened in my small non scientific sample size..
Click to expand...

Agreed. This would indicate a post-manufacturing flaw of some kind.
 
You must log in or register to reply here.
Back
Top