Checking new HDD's in RAID

[Ericloewe]

Looks like only the conveyance test is not supported on my jbod. Just finished running the short test on all the drives and those worked fine.

Can you read the S.M.A.R.T. data? Do long tests work?

Are the drives involved the WD Reds or the Hitachis? If it's the Hitachis, they may not have a conveyance test for some weird reason, which would mean the rest of the setup is working normally.

 

[Hugo Ochoa]

Yes, I can read the reports for all the drives. The short and long tests both worked fine. Faster on the Hitachis than WD drives. Conveyance tests worked fine on the WD drives but apperantly are not supported on my Hitachi drives. I'm currently running some badblocks test on all the drives.

Sent from my HTC One_M8 using Tapatalk

 

[trionic]

Just a note that the System Rescue CD Linux distro contains many useful data recovery and testing utilities, including memtest+, dd, badblocks and iozone. The distro can be booted from CD/DVD/USB.

In addition, it looks like FreeDOS and other boot floppy images can be started directly from SysRescue's boot menu, providing an environment to execute wdidle3.exe (yet to be proven but I'll update this post when I know for sure).

 

[trionic]

Thank you for the advice. RMA created. Do WD ship new or refurbished replacement drives?

The replacement drive was "recertified". Crap considering the defective drive was new. The replacement looks like a Black drive. Sticker says "NASWare".

Not feeling too confident about a using recertified this drive in a ZFS VDEV. Thoughts?

 

[cyberjock]

I have mixed feelings about recertified drives. Some are just idiots that RMA stuff and have no idea that the drive is actually good. The *real* problem though is this... is the chances of a recertified drive failing "prematurely" the same as a "new" drive failing "prematurely"?

You buy a drive, if it's got no sticker could you really tell the difference and predict when it would fail? If you had a new and recertified drive can you run some kind of test and prove that a recertified drive isn't as good as new? If so, you just hit jackpot and many IT geeks everywhere will thank you. But if it stores your data right now, passes all SMART monitoring, SMART tests, bad blocks tests, etc does it really matter?

 

[michaelkoehler]

A second recommended way to check on drive performance and do some stress testing is to do a "dd" read of the each of the drives.
From the Shell you can run the following command - note this is to test 6 drives, adjust it if you have fewer or more drives.
# for i in 0 1 2 3 4 5; do (press return it will then start a new line where you continue)
> dd if=/dev/ada${i} of=/dev/null bs=1048576 & (press return again)
> done (after pressing return it will actually run the above commands)
this will take a long time (hundreds of minutes) and as long as you keep the Shell open, when it does complete it will report how long it took for each of the drives, if one is much longer than the others, it could have a performance issue (even if it passes SMART) (when I did this test on 6 drives, I left it overnight and saw the next morning they were all within 10% of each other)

Great post, Brian.

When I run this, I get an immediate response, though. Am I not waiting long enough, or am I doing something wrong? Is this what the results should look like?

[root@freenas ~]# for i in 0 1 2 3 4 5; do
> dd if=/dev/ada${i} of=/dev/null bs=1048576 &
> done
[1] 4925
[2] 4926
[3] 4927
[4] 4928
[5] 4929
[6] 4930
[root@freenas ~]#

 

[scurrier]

Great post, Brian.

When I run this, I get an immediate response, though. Am I not waiting long enough, or am I doing something wrong? Is this what the results should look like?

[root@freenas ~]# for i in 0 1 2 3 4 5; do
> dd if=/dev/ada${i} of=/dev/null bs=1048576 &
> done
[1] 4925
[2] 4926
[3] 4927
[4] 4928
[5] 4929
[6] 4930
[root@freenas ~]#

Maybe those are the process id's? Did you check to see if they correspond to what's running in "top"?

 

[michaelkoehler]

Maybe those are the process id's? Did you check to see if they correspond to what's running in "top"?

Aha, I think you're probably correct. I can't check without closing the shell, though. I'll let it go a while longer.

 

[Fraoch]

(deleted)

 

[pjc]

Then you run iozone in a seek-heavy manner. Then you keep that running a few weeks (no, seriously, weeks is on the short end).

What do you recommend for that? Random read/write? How do you make it run for weeks?

 

[jgreco]

What do you recommend for that? Random read/write? How do you make it run for weeks?

Actually several discussions here have convinced me to revisit an old tool ... I may have something more specific for you to run soon.

 

[Dennis.kulmosen]

FN comes with a nice tool called badblocks. In write mode you need to do it on each raw device. Badblocks writes a pattern to the device and reads it while comparing it to the pattern. Its slow, but its a complete surface scan plus stress test of the drive.


Sent from my iPhone using Tapatalk

 

[cyberjock]

Actually several discussions here have convinced me to revisit an old tool ... I may have something more specific for you to run soon.

So what are you doing with my mom now?

 

[pjc]

Actually several discussions here have convinced me to revisit an old tool ... I may have something more specific for you to run soon.

Let me know...I'll be interested.

This may be a silly question, but will seek-heavy usage also increase the drive temperature, or does read/write tend to increase it the most? I may also use this opportunity to do a thermal test on my fan configuration...

 

[Ericloewe]

Let me know...I'll be interested.

This may be a silly question, but will seek-heavy usage also increase the drive temperature, or does read/write tend to increase it the most? I may also use this opportunity to do a thermal test on my fan configuration...

Random accesses will cause the greatest increase, from an intuitive point of view.

 

[pjc]

Random accesses will cause the greatest increase, from an intuitive point of view.

That's what I would have thought (power required to move the head, etc.), but I wasn't sure, since most data sheets just list power consumption for read/write vs. idle.

Now I'm looking forward to jgreco's seek-heavy stress test even more...

 

[Fraoch]

Just wanted to say this thread is helping out tremendously for testing.

A quick note: turn off snapshots during those intense iozone tests! You may run out of space if you have aggressive snapshot settings where the snapshot will try to capture those 100 GB+ files you create in iozone.

 

[jgreco]

A quick note: turn off snapshots during those intense iozone tests! You may run out of space if you have aggressive snapshot settings where the snapshot will try to capture those 100 GB+ files you create in iozone.

HAHAHAHAHAHHAHAHAHAHAHAHAHHAHAHAH! That is AWESOME!

 

[Fraoch]

HAHAHAHAHAHHAHAHAHAHAHAHAHHAHAHAH! That is AWESOME!

Yeah, I wondered what was going on there...the snapshots don't care and will happily and dutifully create snapshots for you as they're supposed to. All the while your total capacity is dropping even after iozone is finished and the large files are deleted.

 

[Mlovelace]

Thanks for the information about the IBM ServeRAID M1015. However that will not work for my application. My server board only has 2 PCIe ports. My case is a Norco RPC-4224 (24 drives). So 16 drives would not suffice.

You can get a SAS expander and connect all 24 bays to a single serveRAID M1015 card. That card will support up to 32 drives per the IBM tech sheet.

http://www.newegg.com/Product/Product.aspx?Item=N82E16816117207