Disk speed check?

[joeschmuck]

I think my point was to not let people think that if you toss an L2ARC at it that things would magically be awesome over a piss-poor design in the first place.

I agree that benchmarks are artificial but they are a tool which can be used to help design a better product.

 

[ZippyZoodles]

You need 6TB of usable space but you only have 4 x 4TB drives? In mirror vdevs, that means you only have an 8TB pool, and that puts you at ~75% full. You don't want to be there, that will over time become painfully slow as fragmentation pwns your pool. Add at least two more 4TB drives (another 4TB mirror) if not two more... or larger drives... or both...

Ok thanks. I have purchased 8 x 4TB 7200 RPM HGST drives. Found a 300TB Intel 320 SSD I am going to use for ZIL and am trying to source some 10G network cards.

 

[Ruff.Hi]

To use dd here is an example...

1) Create a dataset which has compression turned off. This is important because compression will give you a false reading.
2) Open up a shell window.
3) Type "dd if=/dev/zero of=/mnt/pool/dataset/test.dat bs=2048k count=10000"
4) Note the results.
5) Type "dd of=/dev/null if=/mnt/pool/dataset/test.dat bs=2048k count=10000"
6) Note the results.
7) Lastly cleanup your mess and "rm /mnt/pool/dataset/test.dat" to delete the file you just created.

Note: /mnt/pool/dataset will depend on your specific pool name and dataset name.

Sorry about the thread necro.

I have 3 new 4TB WD red drives. They are under going disk burn in testing at the moment ... and one of the is about a 1/3 the speed of the other two (in the bad block testing). I want to use the above ... but on a disk level, not dataset / pool level.

Am I right in thinking that I can set up 3 pools / datasets (each pool with 1HDD) and run the above on each dataset (effectively - each HDD)?

 

[joeschmuck]

Am I right in thinking that I can set up 3 pools / datasets (each pool with 1HDD) and run the above on each dataset (effectively - each HDD)?

Yes, you can set up a single hard drive as a pool. This of course is not recommended for normal use but for testing, it's fair game and good common sense for troubleshooting.

As for if you do have one drive slower than the others, take note fo the interface SATA connection they are on, and the specific SATA data cable. Also take note of the drive serial number and model and manufacture dates. What I hope is happeneing is you have two drives on a faster SATA port than the slow one. After your burn-in testing is done, you could test that theory out by relocating the slow drive to a fast drive location. Of course using the three different pool is a smart way to go as well. Ensure you look at your SMART data on the drives after the fact, make sue thing looks off.

Good Luck and continue to use that good common sense.

 

[Ruff.Hi]

Thx for the confirmation. I was going to move the HDDs around (different power and data cables) to test if it is HDD or other. All three where purchased from same place and same time. If HDD is slow, I will be returning it.

 

[joeschmuck]

If HDD is slow, I will be returning it.

If it is significantly slower, if you can return for another new drive, I would too. Just make sure they didn't give you a drive that was a slightly different model. Could you imagine getting an Archive drive by accident, YIKES!

 

[Ruff.Hi]

Turns out I had a bad / fault data cable or seat. I am rerunning the HDD burn-in again. Back in about 80 hrs :(

 

[joeschmuck]

Turns out I had a bad / fault data cable or seat. I am rerunning the HDD burn-in again. Back in about 80 hrs :(

Take note of the SATA cable, maybe place a line on it with a Sharpie and if you ever have a problem with it again, you will know that the cable already failed once before, cut it in half and toss it in the garbage. Glad it was an easy fix, hope the testing comes out okay.

 

[Ruff.Hi]

Take note of the SATA cable, maybe place a line on it with a Sharpie and if you ever have a problem with it again, you will know that the cable already failed once before, cut it in half and toss it in the garbage. Glad it was an easy fix, hope the testing comes out okay.

It is sitting in the basement on the floor ... all by itself. It will be shortly heading to the big trash can in the sky.

Ran some speed tests on my 4 disks (1 old, 3 new).

Code:

root@FreeNAS2[~]# dd if=/dev/zero of=/mnt/Pada2/test.dat bs=2048k count=10000
10000+0 records in
10000+0 records out
20971520000 bytes transferred in 7.530618 secs (2784833918 bytes/sec)

root@FreeNAS2[~]# dd if=/dev/zero of=/mnt/Pada3/test.dat bs=2048k count=10000
10000+0 records in
10000+0 records out
20971520000 bytes transferred in 7.094443 secs (2956048848 bytes/sec)

root@FreeNAS2[~]# dd if=/dev/zero of=/mnt/Pada4/test.dat bs=2048k count=10000
10000+0 records in
10000+0 records out
20971520000 bytes transferred in 7.083413 secs (2960651711 bytes/sec)

root@FreeNAS2[~]# dd if=/dev/zero of=/mnt/Pada5/test.dat bs=2048k count=10000
10000+0 records in
10000+0 records out
20971520000 bytes transferred in 7.042892 secs (2977685686 bytes/sec)

root@FreeNAS2[~]# dd of=/dev/null if=/mnt/Pada2/test.dat bs=2048k count=10000
10000+0 records in
10000+0 records out
20971520000 bytes transferred in 2.344292 secs (8945781139 bytes/sec)

root@FreeNAS2[~]# dd of=/dev/null if=/mnt/Pada3/test.dat bs=2048k count=10000
10000+0 records in
10000+0 records out
20971520000 bytes transferred in 2.330415 secs (8999049394 bytes/sec)

root@FreeNAS2[~]# dd of=/dev/null if=/mnt/Pada4/test.dat bs=2048k count=10000
10000+0 records in
10000+0 records out
20971520000 bytes transferred in 2.350633 secs (8921647879 bytes/sec)

root@FreeNAS2[~]# dd of=/dev/null if=/mnt/Pada5/test.dat bs=2048k count=10000
10000+0 records in
10000+0 records out
20971520000 bytes transferred in 2.327011 secs (9012212015 bytes/sec)

Just ran ada2 again and it was 7.58 seconds - slower than before. Checked the S/N and it was the same drive that was dragging its heals on the badblocks test. This is with a different data sata cable.

 

[Johnnie Black]

Just ran ada2 again and it was 7.58 seconds - slower than before. Checked the S/N and it was the same drive that was dragging its heals on the badblocks test. This is with a different data sata cable.

There are two different 4TB WD models, old one with four 1TB platters and the new one with 3 1.33TB platters, the new one has a noticebly higher sequential read and write speed, you can post the complete models for all to confirm it that's the issue.

 

[Ruff.Hi]

All three HDDs where purchased from BestBuy at the same time. The details are identical ...

WD - Red 4TB Internal SATA Hard Drive (OEM/Bare Drive)
Model: WD40EFRX
SKU: 2452034

 

[Johnnie Black]

WD40EFRX

I mentioned for you to post the complete model, the difference is the rest of the model after this, e.g. for the Blues, which also are available in the same different models:

WD40EZRZ-00WN9B0 - 4 platter model
WD40EZRZ-00GXCB0 - 3 platter model

You can also tell by looking at them, old one is missing the highlighted part:

 

[Ruff.Hi]

ok ... I will pull the drives and have a look at them.

Or just run a smartctl -i against them ...

First 3 are the new ones from Best Buy
ada2 ... Device Model: WDC WD40EFRX-68N32N0 <-- this is the one that 'looks' slow
ada3 ... Device Model: WDC WD40EFRX-68N32N0
ada4 ... Device Model: WDC WD40EFRX-68N32N0
ada5 ... Device Model: WDC WD40EFRX-68WT0N0 <-- this is a much older drive

 

[Johnnie Black]

WDC WD40EFRX-68WT0N0

This one is 4 platters and slower, all the other ones are the newer model, so that's not what's causing the difference, it might be a disk with one or more slow sector zones, is the SMART attribute #1 Raw_Read_Error_Rate still 0? That can be a clue for the existence of slow zones.

 

[Ruff.Hi]

Raw_Read_Error_Rate for all 4 drives are 0.

Some of the drives have non-zero number against 'udma crc error count' but that is caused by faulty data cable (known issue).

I have a question out to WD about this ... no reply as of yet.

 

[Ruff.Hi]

Running the badblocks test again. Three drives are at 29%, 29% and 28% on the write phase of the first pass ... after 1:49.

The first one is the slow one. Maybe not so slow?

 

[Ruff.Hi]

I have a question out to WD about this ... no reply as of yet.

I got a reply from WD ...

I understand that you are experiencing slower data transfer speed on one out of your 3 WD Red drives. I am sorry for the inconvenience caused to you. Please be informed that there are several reasons due to which the data transfer speed get affected, out of which few of reasons are mentioned below:

• Computer in use during data transfer
• Network Traffic
• Faulty network hardware (router, switch, cable)

I would recommend you to follow the link mentioned below in order to know the more about causes and solution of slow data transfer.

And they gave me a link that talked about slow transfer rates to the cloud. I was writing dummy info to a file and then reading dummy info from that file. I don't think any of those issues are real in my situation.

Well ... at least I got an answer reply.

 

[joeschmuck]

I got a reply from WD ...

I understand that you are experiencing slower data transfer speed on one out of your 3 WD Red drives. I am sorry for the inconvenience caused to you. Please be informed that there are several reasons due to which the data transfer speed get affected, out of which few of reasons are mentioned below:

• Computer in use during data transfer
• Network Traffic
• Faulty network hardware (router, switch, cable)

I would recommend you to follow the link mentioned below in order to know the more about causes and solution of slow data transfer.

And they gave me a link that talked about slow transfer rates to the cloud. I was writing dummy info to a file and then reading dummy info from that file. I don't think any of those issues are real in my situation.

Well ... at least I got an answer reply.

That was just a canned answer to address any generic slowness, has nothing to do with your issue of course.

Sorry to hear that one of your new drives is a slightly different model, I know it would drive me nuts for a while. Does even the slow drive meet your throughput needs when your pool is setup? If yes then try to ignore the slow drive.

Good luck!

 

[Ruff.Hi]

That was just a canned answer to address any generic slowness, has nothing to do with your issue of course.

For sure. I was going to reply ... but decided I couldn't be bothered. In the end, I ran badblocks again and the three new drives finished on average at about the 61 hr mark ... ranging from 60-62ish. The 'slow' drive finished in 61. I will stop worrying about the slow drive.

Sorry to hear that one of your new drives is a slightly different model, I know it would drive me nuts for a while. Does even the slow drive meet your throughput needs when your pool is setup? If yes then try to ignore the slow drive.

My comment above (this is a much older drive) was meant to mean ... this was purchased a while ago. I can see how it was confusing. The old model drive is my onsite backup drive in case any in my pool go bad. The three new drives (all from Best Buy) are the same model :).