How to connect USB Disk (last resort)?

[amitkhas]

Hi All,

I have an existing 6 disk setup in RAIDZ1 (RAID 5 w/ ZFS). Unfortunately, 4 of these disks are failing. It is quite surprising since they are less than 3 years old and have an online time of ~1000 hours. They are Hitachi 7200 RPM disks. Probably last time I buy Hitachi... anyway...

I want to get my data out of there ASAP so I can replace the 4 disks. Transfers over SAMBA (Windows 8.1) are slow and often gets hung up. The box is maxed out at 6 disks, so I have to resort to an USB external disk. I am planning to create another pool with the USB external and perform a copy via shell. Questions:

1) I plugged in an NTFS USB 2TB drive, but it's not detecting it in the bios, nor on FreeNAS. Does the drive have to be formatted a certain way for it to be detected? Once it is connected, what command do I issue in shell to perform a copy?

2) The disks are failing as per a previous post I made (link). Post #5 and #6 are print outs of the SMART values. I don't quite understand what it means to be "failing." The drives are still functional, technically. Is it because the sectors, that currently have data, are going bad? Is there anyway to correct this? Could a ZFS scrub potentially correct this?

I decided to create a new thread since the topic is different than the original. The original question has been answered.

I am still very surprised that 4 of 6 Hitachi disks are failing within 3 years. It's for a home use setup and is online only when needed, with only one user.

Any help is MUCH appreciated. I am in a time crunch to save the data.

P.S. I now know that RAIDZ2 is a better option. I was not aware when I built the system in 2011. When I re-build the pool, what is the recommended RAID type for most tolerance. Performance is not a priority since it's for home use. I'm thinking RAIDZ2 or RAIDZ3. The problem with RAID 1+0 is that 2 disk failures could result in data loss depending which disks fail. Thoughts?

 

[solarisguy]

USB 2.0 speed will be your very low limit. I have a different advice.

I think it is worth trying to use FTP for file transfer from your FreeNAS, to your Windows 8.1 system using FileZilla FTP client ( https://filezilla-project.org/ ). And on your Windows you can either have fast internal disks or can connect an external disk using USB 3. FileZilla can be set to retry if the network connection breaks down for whatever reason. And unlike with Windows copy, you can see what had failed and requeue.

P.S. You connected 2TB disk, but earlier you had written that you have almost 3TB of data.

 

[amitkhas]

I have nearly 3 TB of data but only a 2 TB external drive. I am willing to save only the highest priority of 2TB and delete the rest of the data. However, I never was able to get the 2 TB WD External Drive to be detected, so I actually never got that far.

I will attempt FTP and let you know how it goes.

Another question:

When I recreate the pool, which option would be better? The box allows for 6 disks. I am planning to utilize the existing 2 working 2TB disks.

4x 3TB drives in RAIDZ2 + 2x 2TB drives in RAID 1 (8TB usable)
or
6x 2TB drives in RAIDZ2 (8TB usable)
or
6x 2TB drives in RAIDZ3 (6TB usable)

I want to purchase my replacement disks soon. I only require about 6TB of storage. Extra is bonus. Performance isn't as important compared to reliability.

 

[Yatti420]

Please read docs.. ntfs write on freenas - not gud idea.. Get data off.. are the drives secure mounted? 1000 hr on drives isnt even burned in :)

You wont wanna reuse rhe drives if had multiple droves fail..

Sent from my SGH-I257M using Tapatalk 2

 

[solarisguy]

6x 2TB drives in RAID-Z2

However, that is going to be around 6TiB useable, since you have to convert between capacity in TB (decimal) and TiB (power of two), and one should not plan on using more than 80% of ZFS pool capacity, so...

• 2 TB = 1.819 TiB
• 80% * 4 * 1.819 TiB = 5.8208 TiB

Moreover, since so many of your disks have already failed, I would strive for 6x 3TB drives in RAID-Z2. Even, if initially your RAID-Z2 would have 4x 3TB disks and 2x 2TB disks. That is, I would replace the “good” 2TB disks as soon as your budget allows. You can reuse your non-failed 2TB disks for external backup copies.

6x 3TB using raidz2 would give you 8.73 TiB, and that would allow you to have snapshots, unless 6TB already includes space for snapshots or your data is static and read-only.

P.S. Is your NTFS 2TB external drive USB 3.0 capable?!

 

[amitkhas]

Sorry in advance for the noob question:

How do I secure mount the disks? How can I tell if they are secure mounted?

The NTFS 2TB external is on USB 2.0. This may be a moot point cause I believe FreeNAS is not compatible with USB 3.o.

Thanks for the recommendation, solarisguy.

 

[solarisguy]

Secure mount... In simplest terms, would the disks move if you try to move them in your enclosure?

For example, (not necessarily your case, but it happens):

• an enclosure where a manufacturer invented some way of quick mount that allows the drives some freedom of movement
• using thumbscrews to fasten the hard drives and those screws are a tiny bit too long (they would have been OK, if some sort of pads or washers were used)
• mounting the drives just on one side or not with at least two screws on both sides

Quick mount and thumbscrew are not evil by themselves, sometimes execution is very poor...

 

[amitkhas]

Ah, secure mount as in whether the drives are physically mounted securely in the enclosure.

Yes, the drives are fixed pretty well. It's in a 6 bay, vertical tower. The screws are fastened tightly. There is about a 1 inch vertical gap between each disk. The SMART values indicate temperature within range. Temps are 32.

Could a ZFS scrub correct the "failing" disks?

 

[solarisguy]

No, scrub does not fix malfunction of a disk.

I have read again your earlier post, and you stated that you had managed to copy most of the files, so you should be good with FTP for the remaining ones.

 

[amitkhas]

I initiated the copy via FTP. It is working MUCH better than SAMBA/CIFS. The SAMBA/CIFS transfers were timing out. Thank you!

On RAID-Z1, I was getting transfer rates of ~50MB/s on SAMBA/CIFS.

Any idea what the transfer rates would be on RAID-Z2 and RAID-Z3?

 

[solarisguy]

RAID-Z2 is slower than RAID-Z1. However, I cannot tell you by how much on your hardware.

Once people deploy FreeNAS on currently recommended hardware (CPU from one of Intel server lines, ECC RAM, etc.) speed difference is not noticeable when using a single Gigabit Ethernet connection.

Ultimately, at this point, your biggest problem is that you do not have ECC RAM, and your CPU does not support ECC RAM.

P.S. Substantial speed difference between CIFS and FTP transfer rates might indicate some software configuration issue(s).

 

[amitkhas]

I am in the process of procuring new 2TB or 3TB HDDs (quantity 4).

I am clearly not going to go with Hitachi. Any brand recommendations?

I'm thinking WD Red. I can get a 2TB for $90 x 4 = $360.

 

[solarisguy]

WD Red is the series recommended here.

And then you might want to read

• http://forums.freenas.org/index.php...red-drives-with-high-load-cycle-counts.18095/
• http://forums.freenas.org/index.php?threads/hacking-wd-greens-and-reds-with-wdidle3-exe.18171/

I cannot advice on 3TB versus 2TB, since that depends on your budget and your predictions about data growth. Nonetheless, please take into account that if shortly (let's say 12-18 months) you might need six 3TB drives, you had just wasted money on four new 2TB drives.

Also, I would budget replacing the last two Hitachi drives. Unless they were manufactured months later than the failed ones. When looking at the disks in Storage section of the FreeNAS GUI, are the serial numbers very close?

 

[amitkhas]

Thanks for the links.

I was thinking the 3TBs might have higher failure rates because:

A. they are newer
B. larger capacity means higher potential failures

The higher capacities is one of the reasons RAID-Z1 is "dead." Therefore, I was thinking it would be 'safer' to get 2TBs for now.

I'm not sure if my thinking is correct. Then again, that's a good point about data growth - I'd hate to waste money now only to purchase new HDDs in 2-3 years.

Here are the Serial Numbers:
MN1221F305XSMD
MN1220F305HWTD
MN1221F305XXRD
MN1220F305YBAD
MN1220F3056DAD
MN1220F3057EYD

I am not sure how to interpret that. Are they close?

 

[solarisguy]

To me there are two clusters of serial numbers. Are these two the ones that did not fail?
MN1221F305XSMD
MN1221F305XXRD

Your thinking that newer is not better, is applicable to bleeding edge products. Going with 6TB Hitachi or 5TB Seagate would be exposing yourself to first generation products. WD Red 4TB are being sold since September of last year. And WD Red 3TB are being sold since 2012.

 

[amitkhas]

Strange enough, those are not the "good" drives.

The good drives are:
MN1220F305YBAD
MN1220F3057EYD

 

[Yatti420]

WD Red is the series recommended here.

And then you might want to read

• http://forums.freenas.org/index.php...red-drives-with-high-load-cycle-counts.18095/
• http://forums.freenas.org/index.php?threads/hacking-wd-greens-and-reds-with-wdidle3-exe.18171/

I cannot advice on 3TB versus 2TB, since that depends on your budget and your predictions about data growth. Nonetheless, please take into account that if shortly (let's say 12-18 months) you might need six 3TB drives, you had just wasted money on four new 2TB drives.

Also, I would budget replacing the last two Hitachi drives. Unless they were manufactured months later than the failed ones. When looking at the disks in Storage section of the FreeNAS GUI, are the serial numbers very close?

Wouldn't call the reds recommended.. Nas drives sure but the reds themselves need additional work like wdgreen as you indicate..

Sent from my SGH-I257M using Tapatalk 2

 

[amitkhas]

If not the WD Reds, then what would be better? Aside from enterprise disks - those would be vastly out of budget and probably overkill?

 

[solarisguy]

New drives from WD Red series might not need to be adjusted, that is why I gave two references.

WD Re is not always better, for example WD Green would operate in a wider (0-°60C) temperature range than WD Re, WD Se and WD Xe (5-55°C), while WD Red drives have operating temperature range (0-70°C).

WD Red drives have TLER (time-limited error recovery) – good, WD Green ones do not – not good. (See discussion elsewhere in the forum.)

The manufacturer does not publish MTBF for WD Green drives and gives them 2 year warranty, while MTFB for WD Red drives is 1 000 000 hours (= AFR 0.872%) and warranty is 3 years.

--

There is no disk that is better in every single category. Is seek time important for you? May be maximum sustained I/O operations per second (IOPS)?

For example, let's look at non-recoverable read errors per bits read for different 4TB drives from Western Digital****:
* 1 in 10^13 for WD AV, WD Purple
* 1 in 10^14 for MegaScale DC 4000.B, WD Black, WD Green, WD Red and WD Se
* 1 in 10^15 for Ultrastar 7K4000, WD Re

If you want better values of that parameter, then 1.2TB drive is the largest available:
* 1 in 10^16 for Ultrastar C10K1200 (SAS only 10,000 rpm, 1.2TB, MTBF 2Mh)
* 1 in 10^16 for WD Xe (enterprise performance series 10,000 rpm, 900GB, MTBF 2Mh)

The very existence of drives with non-recoverable read errors per bits read with a value of 1 in 10^14 was one of the driving forces behind recommending RAID-Z2 over RAID-Z1.

Consequently, if I really valued my data, using RAID-Z2 with six drives that have 1 in 10^15 would be safer than using RAID-Z3 with seven drives that have 1 in 10^14 non-recoverable read errors per bits read...


**** I could not find those error rates for 4TB hard drives made by Toshiba. And for 4TB Seagate drives it was 1 in 10^14 for Terascale drives, and 1 in 10^15 for Enterprise Capacity series.

 

[cyberjock]

Some of the Reds that came out like 6 months ago needed some extra work. But the Reds work fine(as does the Greens if you use WDIDLE).