Large performance gap with EARS on 4.7

[ixnu]

Your procedure would be:

0) Enable 4k align setting on unRAID 'settings' page

1) stop array

2) remove sde from array by unassigning it on the 'devices' page

3) go back to the unRAID 'main' page and start the array with one missing disk

4) run preclear_disk.sh -z /dev/sde from putty (the -z clears the mbr).  This should only take a few seconds or maybe minutes.

5) stop the array

6) assign the disk to the same slot on the 'devices' page

7) go back to the unRAID main page - it should say 'upgrading disk'.  Click the 'I'm sure' checkbox and start the array.  unRAID will rebuild the old data onto the disk.  This will take many hours, but your array will be available the entire time.

 

This worked perfectly! Rebuilt over night and writing ~38 MB/s this morning with no pauses through 500 GB.

 

This needs to go in the AF wiki page and I agree that this should be the default for 4.7 because I followed the set up wiki and it did not mention that setting. I also followed the AF wiki and it was confusing to be b/c it reads like no action needs to be taken for AF drives on my version - which is certainly not the case.

 

Again, thanks for the best kind of help: fast and precise.

[Rajahal]

I looked into the issue a bit more to see what could be done to prevent your scenario from happening again.  I agree that the wiki isn't clear enough about this particular case.  However, the unRAID 4.7 Announcement thread spells it out perfectly:

 

Whether to create Partition 1 starting with sector 63 or sector 64 is configurable using a new setting in the Disk settings section of the Settings page called "Default partition format".  The default value of this setting is "MBR: unaligned"' date=' which specifies an MBR-style partition table with partition 1 starting in sector 63.  If you intend to install a new AF  drive, you should first change this setting to "MBR: 4K-aligned", which specifies an MBR-style partition table with partition 1 starting in sector 64.  [/quote']

 

I'll work on the wiki and try to make this situation more clear.  Thanks for bringing it to my attention, and I'm glad we were able to help you.

[Johnm]

I definitely have a bigger issue then a jumper.

 

I pulled the new EARS out and put in a sammy F2 1.5TB and started my copy over. I just noticed it is taking an hour to copy an 8Gig MKV.

 

The rest of the rig is a well burned in, stable supermicro server. all drives are on the mobo headers. glad its getting decommissioned.

 

Is there a parity build or drive rebuild going on?

This was my WHS backing up so i could repourpose the hardware for unRAID. I looked in my event log. bad sectors. the drive I was using got 127+ smart errors overnight. pending failed sectors. best part is that drive PASSED the WD tools low tools and low level format just a week ago. my problem is solved.

 

thanks guys for helping me also in this hijack.

[swells]

Is ~40MB/sec normal from the EARS drives. I currently have 3 of them and only see above 30MB/sec in bursts. Usually stays between 25-30MB/sec. All of my drives are 4k aligned. Anything else i should check? I was under the impression that my speeds were normal from what i had read until i saw this post.

[chuck23322]

Is ~40MB/sec normal from the EARS drives. I currently have 3 of them and only see above 30MB/sec in bursts. Usually stays between 25-30MB/sec. All of my drives are 4k aligned. Anything else i should check? I was under the impression that my speeds were normal from what i had read until i saw this post.

 

4k Alignment should take care of the "AF" performance hit.  {Assuming NO jumper on them!}

 

What kind of controller are you driving them with, on which kind of bus card? {or is it a MoBo connector}?   SATA I?  In a Multi-port chassis?    What kind of parity drive?  

[mbryanr]

Also...transfer speed can be dependent on the client (type of NAS, OS)

[swells]

Im using the mobo connectors right now on a NVIDIA nForce 560 SLI MCP chipset. I have 3x EARS drives, one parity and two data. I recently added a Hitachi 750GB drive to expand a little until i could order another 2TB EARS drive, but adding that didnt affect my speeds at all. Im running windows 7 as my client where the data is being moved from and most of the data is off of my main 1TB WD Black OS drive.

 

Dont get me wrong, im not unhappy with 25-30MB/sec since the initial copy of >3TB of data after build was the only part that i would have seen great benefit. Now i will be moving smaller 50GB or less blocks of data over at a time and wont really notice it too much. I just thought if there was a way i could squeeze out more performance, why not.

[chuck23322]

Im using the mobo connectors right now on a NVIDIA nForce 560 SLI MCP chipset. I have 3x EARS drives, one parity and two data. I recently added a Hitachi 750GB drive to expand a little until i could order another 2TB EARS drive, but adding that didnt affect my speeds at all. Im running windows 7 as my client where the data is being moved from and most of the data is off of my main 1TB WD Black OS drive.

 

Dont get me wrong, im not unhappy with 25-30MB/sec since the initial copy of >3TB of data after build was the only part that i would have seen great benefit. Now i will be moving smaller 50GB or less blocks of data over at a time and wont really notice it too much. I just thought if there was a way i could squeeze out more performance, why not.

 

For Windows 7 as a client -- take a look at a copy/move replacement tool.  {Integrates straight into Win 7}..    I love it.  Dramatically improves the transfer throughput.

 

Teracopy @ http://www.codesector.com/teracopy.php

[swells]

I am actually already using teracopy. I also love it, especially how you can queue multiple transfers so that they do not run in parallel and degrade throughput.

[Johnm]

Dont forget, there are 3 generations of the 2TB EARS. the 3rd version is faster and cooler then the previous ones.

 

•1st version: WD20EARS-00S8B1 = 4 500GB platters (launched in January 2009).

•2nd version: WD20EARS-00J2GB0 = 4 500GB platters (made since early April 2010 in Thailand and on the market since May 2010). The PCB is more compact than the WD20EARS-00S8B1 and design of the hard disk has been modified and approximates that of the 1 TB version. The firmware WD20EARS 00S8B1-and-WD20EARS 00J2GB0 is: 80.00A80 (like many other models of Western Digital hard disks).

•3rd version: WD20EARS-00MVWB0 = 3 trays 667GB (the production of this new model began at the end of May 2010?). The firmware is WD20EARS-00MVWB0: 50.0AB50.

 

[swells]

They are all three of the newer ones. I checked an old syslog (at work so cant look at a new one).

 

Apr 18 17:52:20 Tower emhttp: pci-0000:00:07.0-scsi-0:0:0:0 host1 (sda) WDC_WD20EARS-00MVWB0_WD-WMAZA0690036

Apr 18 17:52:20 Tower emhttp: pci-0000:00:08.0-scsi-0:0:0:0 host3 (sdb) WDC_WD20EARS-00MVWB0_WD-WMAZA0334009

Apr 18 17:52:20 Tower emhttp: pci-0000:00:08.0-scsi-1:0:0:0 host4 (sdc) WDC_WD20EARS-00MVWB0_WD-WMAZA2763400

[Rajahal]

Is ~40MB/sec normal from the EARS drives. I currently have 3 of them and only see above 30MB/sec in bursts. Usually stays between 25-30MB/sec. All of my drives are 4k aligned. Anything else i should check? I was under the impression that my speeds were normal from what i had read until i saw this post.

 

These speeds look totally normal to me.