Tom,

 

Is it possible to move the "support for >2Tb drives" from 5.2 to 5.1 (or maybe even 5.0)?

 

3Tb drives are not very expensive, (5.7 eurocent per Gb, compared to 4.12 eurocent/Gb for a 2Tb drive).

 

Upgrading a 1Tb drive to a 3 Tb drive is cheaper than upgrading it first to a 2Tb drive and then upgrading a 2Tb drive to a 3Tb drive.

 

Support for a HighPoint RocketRaid 620 PCIe 2.0 x 1 SATA III (6.0Gbps) card is also needed though.

 

See http://lime-technology.com/forum/index.php?topic=9741.0

 

Thanks

Personally I'd rather have it exactly where it's at.

 

This gives time for other manufacturers to bring out their own 3TB drives. Right now there seems to be only 1 internal 3TB drive actually available to be purchased by the consumer. Why bring out support for only 1 single 3TB drive? There will be other drives out there which will have different requirements for unRAID support. Why rush something when the market hasn't even stabilized on what's required?

+1 Brit.  There are numerous other items to be fixed / tweaked that are far more important fundamentally to unRAID, and Tom has done a good job of prioritizing those.  3TB drives won't be mainstream for a while, and they certainly aren't stable or reliable enough for server use yet.

I respectfully disagree   , and this is why:

 

I have a CM590, filled with 12 drives. 8x 2Tb and 4x 1Tb. Since the case is full, I have to replace a smaller drive with a bigger one to expand the array. This leaves me 4Tb headroom to expand at the moment.

 

If I buy a 2Tb drive to replace the 1Tb drive, then I have to find another use for this 1 Tb drive now. Also, my array is expanded with only 1Tb, for the price of a 2Tb drive.

 

If unraid would support in the next version 3Tb drives, then I would pay (after the parity drive) a 3Tb drive price for an 2Tb expansion of the array (when replacing the 1Tb drives).  Later I would get only 1 Tb expansion for the 3Tb drive if I have to replace a 2Tb drive

 

Those 4 1Tb drives are on the moment also on a PCI sata controller, I need a new controller anyway (a PCIe controller), and this is included with the WD 3Tb drive.

 

The price per Gb is reasonable (for the WD drive if you need a new HBA), not great, but reasonable.

 

The alternative: a bigger case, bigger PSU and new controller cards is not cheaper. And more difficult to "sell" to the wife.

 

Cheapest solution in the long run would probably be a second Unraid server (can reuse the "old" drives)), but this is also difficult to "sell" to the wife.

 

 

 

About availability: This is for people in The Netherlands, or Belgium:

 

Hitachi HDS723030ALA640 3 TB (Bulk, SATA 600, Deskstar 7K3000, 24/7)

http://www.alternate.be/html/product/Harddisks_3,5_inch_SATA/Hitachi/HDS723030ALA640_3_TB/492903/?tn=HARDWARE&l1=Harddisks&l2=Serial-ATA&l3=3%2C5+inch  230 euro

http://www.zercom.eu/index.php?p=35&f=1&s=1&a=265832     201.5 euro

 

 

Western Digital WD30EZRS (Kit) includes HBA

http://www.alternate.be/html/product/Harddisks_3,5_inch_SATA/Western_Digital/WD30EZRS/480022/?tn=HARDWARE&l1=Harddisks&l2=Serial-ATA&l3=3%2C5+inch   240 euro

http://www.zercom.eu/index.php?p=35&f=1&s=1&a=124138    210 euro

http://www.informatique.nl/110942/wd-caviar-green-3tb.html     239 euro

http://azerty.nl/8-857-340431/wd-caviar-green-wd30ezrsdtl-.html  217 euro

 

 

WD Caviar Green 2.5TB, 64MB, SATA2 WD25EZRSDTL includes HBA

http://www.informatique.nl/110941/wd-caviar-green-2.5tb.html    178 euro

http://azerty.nl/8-857-340430/wd-caviar-green-wd25ezrsdtl-.html   168 euro

 

Available enough for me...

 

I can fully understand and appreciate your situation and feedback, so don't take this discussion the wrong way. I too want to see 3TB drive support eventually.  It just seems like there's more shakedown and fallout to be had in this area before things become standardized.

 

I'm hoping for the 5.1 series to be in beta-testing within the next three months and with a month or two before it's final. That would put the 5.2 series development somewhere within the next 6 months. Ideally by then > 2TB drive area will be standardized and the picture will be clearer.

 

That seems to be more available than over here from the last time I looked before the Holiday.

 

Do you happen to know if the packages include the same SATA HBA or if the different SATA HBAs support all the different 3TB drives?

 

To have proper support it would seem one would need to purchase at least two of each of the 3 > 2TB drives available, test the different SATA HBA individually with each drive individually, test two of the same SATA HBA together with each drive individually, test the different SATA HBA individually with all the drives together, test the different SATA HBA together with each drive individually, test the different SATA HBA together with all drives together.

 

Sometimes things behave differently with 2 of the same SATA controllers, sometimes they don't. Sometimes things behave different with 2 of the different SATA controllers, sometimes they don't.

 

It's a quite extensive and expensive test scenario that will take a good bit of time to be properly tested.

Obviously having 3TB drive support is necessary - that is where the future is going.  But just like with USB 3.0, there is a high volume of hardware incompatibilities out there right now for 3TB drives.  Some controllers support them, and some don't.  The question is, does the dominance of 3TB drives and the adoption by the mass market line up with your timing, or Tom's timing.  Most likely it will fall somewhere in between.  Your issue is not a matter of if support will be there, but when.  We don't have the luxury of knowing what is simple or hard for Tom to develop.

This is a good discussion and exactly the kind of feedback that's valuable for prioritizing features.  I have looked into some of the changes necessary to support large drives:

• The linux kernel itself should be ok, even if not a 64-bit kernel.
  
• There are some changes required in the unraid driver, don't know about other drivers.
  
• unRaid will need to support GPT (GUID Partition Tables).  This constitutes the most significant code changes.  Becaues of this I want to roll in another feature, "Support foreign drives" as part of this feature.  This increases the number of changes slightly.
  
• Obviously the big unknown is motherboard and disk controller support.  Since HDD manufacturers are bundling controller cards, this does not bode well.  Note that it's not necessary to support bios boot of large drives for unRaid, but there may be API limitations in the physical controller and/or individual drivers.
  

 

That said, I expect large drive support to be more and more desired and probably will get included in 5.1.

 

As for release schedule - my goal right now is to get 5.0 out of beta and into "final".  So the current feature set defined for 5.0-betaX is pretty much finalized.  Beta-3 is almost ready and once all the forum/website/wiki management chores are done I'm hoping to concentrate fully on getting this done.

Obviously having 3TB drive support is necessary - that is where the future is going.  But just like with USB 3.0, there is a high volume of hardware incompatibilities out there right now for 3TB drives.  Some controllers support them, and some don't.  The question is, does the dominance of 3TB drives and the adoption by the mass market line up with your timing, or Tom's timing.  Most likely it will fall somewhere in between.  Your issue is not a matter of if support will be there, but when.  We don't have the luxury of knowing what is simple or hard for Tom to develop.

 

Very true. It would for me be much cheaper/easier to have 3Tb support now, than later.

 

It could be very hard for Tom to implement this, I dont know. And to be save, support for an extra HBA is needed.

 

However, I ask because "I already have a no, I can maybe get a yes" 

 

 

I respectfully disagree   , and this is why:

 

I have a CM590, filled with 12 drives. 8x 2Tb and 4x 1Tb. Since the case is full, I have to replace a smaller drive with a bigger one to expand the array. This leaves me 4Tb headroom to expand at the moment.

 

If I buy a 2Tb drive to replace the 1Tb drive, then I have to find another use for this 1 Tb drive now. Also, my array is expanded with only 1Tb, for the price of a 2Tb drive.

What might be possible, assuming your disk controller can handle a 3TB drive, or the controller card supplied with the drive is recognized, is for you to purposely create an HPA (host protected area) on the drive making it look like a 2TB drive. (this would be temporary until 5.X when 3TB drives are supported)

 

By doing this, you can install the 3TB drive, run the appropriate

hdparm -N pXXXXXXXXXXXXX

command on it, and it will be a 2TB drive as far as unRAID is concerned.  (Your hardware investment is safe)  You would replace the 1TB drives only as needed (don't want to get more complcated than needed).  You'll not be able to access the upper 1TB of the data drives with the HPA, but think of it as built-in expansion room for when it is supported.  If you can replace the parity drive and one of the 1TB data drives with 3TB drives and make unRAID think they are 2TB drives (again, assuming the hardware supports them) you'll gain 1TB in usable space.  If you replace a second 1TB data drive with a 3TB pretending to be a 2TB, you'll gain another 1TB of data space.

 

Then, once native 3TB drives are supported you'll probably need to stop the array, remove the HPA on the parity drive, and re-compute parity.  (unRAID will think you upgraded the parity drive) After you do that, you'll probably need to empty the 3TB drive masquerading as a 2TB drive, then un-assign it, calculate parity without it, remove the HPA from it with the hdparm -N command, then assign the drive back to the array as a new drive.  It will need the new GPT structure to support it, so no easy conversion is likely (but who knows)  At that point in time, you'll gain the other 1TB (per upgraded drive) of usable space.  

 

There will need for  a method to replace a smaller MBR based data disk with a larger GPT based disk.  (because at some point, it would be less expensive to purchase a new 4TB drive than to replace an older 2TB drive.  It is already that way for me when replacing ANY of the older IDE drives.  I can get a new 750Gig IDE drive for about $200.  I'd be stupid to do so)  For that reason, who knows what will be involved in replacing a data disk with an MBR based partition with a larger disk with a GPT based partition.  unRAID will need to be able to handle it since it will happen.

 

The migration to >2TB drives will not be without some pain... and you'll need to recognize you'll be without parity protection for some periods of time, but I think it would be possible.

You'll probably be the first, and perhaps the only person to try faking unRAID with an HPA to le it use a 3TB drive as a 2TB drive, and with nobody to confirm it works before you, there are risks. (the new disk controller might not work as desired)

 

Joe L.

That said, I expect large drive support to be more and more desired and probably will get included in 5.1.

 

This is quite a statement, and actually accomplishes what Wimpie asked for in the beginning:

 

Is it possible to move the "support for >2Tb drives" from 5.2 to 5.1 (or maybe even 5.0)?

That said, I expect large drive support to be more and more desired and probably will get included in 5.1.

 

This is quite a statement, and actually accomplishes what Wimpie asked for in the beginning:

 

Is it possible to move the "support for >2Tb drives" from 5.2 to 5.1 (or maybe even 5.0)?

 

He made a convincing argument.  Also HDD manufacturers have all started the rush to large drives, so I think it's wise to keep pace.

Hi Joe L.,

 

Seems like a risky proposal  .

 

I dont't know if any hardware I have would support 3Tb drives.

 

WD included a controller card because most disk controllers would not support 3Tb drives (a HighPoint RocketRaid 620 PCIe 2.0 x 1 SATA III (6.0Gbps) card).  As far as I know, this card is not supported by Unraid.

 

Anyway, I am happy with Tom's reply that >2Tb support will probably get included in v5.1.

 

Now it's waiting for firmware upgrades for all the controller cards and BIOS upgrades for all the MB's out there...

 

 

Hi Joe L.,

 

Seems like a risky proposal  .

 

I dont't know if any hardware I have would support 3Tb drives.

Exactly the problem Tom is facing.  Only one drive currently available, and only one controller.

Yes... VERY RISKY.  But possible if you understood and accepted the RISK. (not for the newbies....)

On the subject of newer faster bigger hardrives.  Would it make sense in the future to support USB3 drives in the array?  This would allow another way of expanding when your case runs out of HD slots.  This is also a way to bridge the issues of >2 Tb drives.

 

 

On the subject of newer faster bigger hardrives.  Would it make sense in the future to support USB3 drives in the array?  This would allow another way of expanding when your case runs out of HD slots.  This is also a way to bridge the issues of >2 Tb drives.

Interesting thought.

 

It might be slower, it might not support spin-down or temperature readings  (I don't have a clue what features the external drive and interface support), but for some who know and accept the limitations, it would be a solution for expanded storage.

 

The HUGE issue is that the overall speed of writing to the array would be dictated by the write speed to the multiple USB3 3TB drives (one of them would have to be the parity drive, and another for data storage)

 

Other than that, and the fact that the person who needs the expansion the most is far more likely to have a motherboard supporting only USB2, it might be a short term solution for some. 

 

Joe L.

it might not support spin-down or temperature readings  (I don't have a clue what features the external drive and interface support),

 

It depends on the USB bridge.  I have many of them, and some, don't pass anything ... even simple IOCTLs like identify.  Others, like JMicron, pass a full suite of IOCTLs and even pull all the smart attributes (look at the docs for the -d switch for smartctl).

 

BTW, version 5.40 of smartctl has better support for auto-determination of bridges, and setting the appropriate -d parameter.

Transfer speeds aren't keeping up with capacity so its taking longer and longer to read/write a whole disk.  Not such a problem for unRaid but in traditional raid arrays you are going to end up with multiple day rebuilds on even small arrays.  Not a good situation with old disks!

I would not rush for 3TB disk support. Especially not until there are a) more drives available and b) lot more controller and mainboard support 3TB disks. Just wait and see what will happen. That one 3TB drive may remain single as the industry is moving to 2.5" anyway.

Being that >2TB support is slated for 5.1 at the soonest, I think Tom has plenty of time for the industry to catch up to the new standard they expect consumers to test for them.  And given the price / size disadvantage of 2.5" solutions, I don't think you'll see 3.5" drives die off.  I expect to see 5TB on 3.5" format before hybrids finally catch up.

if the big players start using 2.5" drives in their products then they will soon be cheaper than 3.5"

The largest 2.5" disk I see is a 1T.  And they are priced considerably higher than 2T 3.5" disks.  And they are slower.  I can't imaging that overnight 2.5" disks with 3T are suddenly going to be released and supplant 3.5" disks.  If they do it is bad news for us.

 

I have watched the 750G, 1T, 1.5T, 2T, and now beginning of the 3T drive progressions.  They seem to follow a similar trend ...

 

1 - Jubilation (initial announcements)

2 - Sticker shock (initial pricing)

3 - Denial (we don't really need these, the priror size is cheaper / big enough, etc.)

4 - Niche use (someone buys them, ooos and ahs from the forum)

5 - Price tumbles, demand increases (everyone buys them - every week a new deal)

6 - Price bottoms out (rumors of new drives emerge)

 

We're at step 6 on the 2T, and step 3 on the 3T.

 

The twist with the 3T drives is that they can't be added to an array without an enhancement to unRAID.  But once 3T drives come down to the $150 price point, people will want to buy them and use them in their arrays.  When they hit $99, there will be a small riot if people have to keep buying the 2T drives.  I expect by March the pressure to support these drives will be very strong.  It seems that the manufacturers are looking to support these drives with GPT partitions and exposing the illusion of 512 byte sectors (rather than fully exposing the actual 4K sector structure).  I think unRAID needs to support that relatively quickly.

There isn't just 1 3T drive out there. I see 3 now as well as some 2.5T drives. Seagate, Hitachi and WD all have 3T products. Seagate is just sitting on the release of the 3T internals likely because of the early tech support issues. I don't believe Samsung has a 3T drive yet but they can't be far behind.

 

Don't forget that this is the year we're really going to see UEFI begin to replace the old BIOS and with UEFI rolling out I'd bet most if not all of those motherboards/systems will support 3T drives. So, this makes me agree completely with bjp on this. I expect the 3T's to start really rolling soon and for them to become cost competitive very quickly. Around March is not unreasonable to expect these to turn the corner, especially once the major competitors 3T drives all hit retail shelves.

 

Peter

 

Speaking of UEFI, will unRaid work on the current UEFI motherboards?  I was thinking of picking up a UEFI H67 based board in the next few days.

Speaking of UEFI, will unRaid work on the current UEFI motherboards?  I was thinking of picking up a UEFI H67 based board in the next few days.

 

I don't believe that anybody has tested it yet.

If UEFI can boot from a USB device, it might work. I think Linux was one of the first group of OS's to support UEFI style systems. It's a matter of that support being in the unRAID 4.7 kernel.

 

It should be well supported under unRAID 5.0 beta 3 since it's supposed to be rebased on newer Slackware distro (13.1? or 13.2/Current) than the older 12.2 series.