I'm almost positive there is no easy solution, but I'm looking to convert from ZFS RAIDZ2 (12x 2TB and 2x 4TB) to unRAID.

 

Unfortunately the array is almost full so I can't just erase the drives and convert them so I think I need to find extra drives to transfer the data over and then format the existing ones but are there any other options?

 

Also, I haven't purchased unRAID yet but I have another question... is it possible to manually configure advanced settings for Docker or KVM?  For example, I'd like to use macvlan's with Docker and a few other features.

Edited December 27, 20178 yr by sofakng

30 minutes ago, sofakng said:

I'm almost positive there is no easy solution

There isn't, it's easy do add disks to unRAID but it's not possible to remove disks from FreeNAS so you'll need to more disks or some other temp storage.

 

32 minutes ago, sofakng said:

is it possible to manually configure advanced settings for Docker or KVM? 

There are no other limitations on the trial key other than the time limit.

Thanks for the information.  Do you know if the docker containers can be configured to use macvlan and other advanced features of docker?

Docker isn't really my forte but I believe you don't need macvlan for v6.4, it's already possible to assign different IPs.

I'd have a hard time recommending 2T drives for such a large unRAID build.

 

10 2T drives = 20T. WIth 3 8T drives (4 including parity) you could replace the whole lot of them and get an extra 4T, and keep the 2T drive configuration for a backup.

 

I recommend 8T drives due to them being at the sweet spot on price.

 

Many have been purchasing the Best Buy "Easystore" 8T external drives. They contain WD REDs and users have had good luck with them (me included). They have been down around $150 over the holidays, but a quick check shows them at $179 right now.

A user has 5 8T Seagate archives for sale for $125 each. These drives are a bit faster and community has had good luck with them. And they are still under warranty. Worth considering.

2 minutes ago, SSD said:

I recommend 8T drives due to them being at the sweet spot on price.

I prefer to buy a bit above the sweet spot, because cost/TB isn't the only important metric. Larger drives will consume less electricity/TB and have a longer economical life span before they become a waste of electricity and a waste of drive slots.

 

2TB was nice drives to buy somewhere around 2010. It's most definitely better to go with 8TB or bigger for a new installation and to try to get the biggest the economy can afford for parity - it saves lots of work later, shuffling around files.

3 minutes ago, pwm said:

I prefer to buy a bit above the sweet spot, because cost/TB isn't the only important metric. Larger drives will consume less electricity/TB and have a longer economical life span before they become a waste of electricity and a waste of drive slots.

 

2TB was nice drives to buy somewhere around 2010. It's most definitely better to go with 8TB or bigger for a new installation and to try to get the biggest the economy can afford for parity - it saves lots of work later, shuffling around files.

 

Generally agree - but the 8T are a lot less than the 10T. A 48T array would take 7 8T drives or 6 10T drives (including one parity).

 

Cost of 7 8T drives (@$180/drive) = $1260

Cost of 6 10T drives (@$330/drive) = $1980

 

I'll live with one extra drive.

1 minute ago, SSD said:

 

Generally agree - but the 8T are a lot less than the 10T. A 48T array would take 7 8T drives or 6 10T drives (including one parity).

 

Cost of 7 8T drives (@$180/drive) = $1260

Cost of 6 10T drives (@$330/drive) = $1980

 

I'll live with one extra drive.

I have earlier been popping in 10 TB parity drives - now have two 12 TB waiting to be added to new build. It saves a lot of work to have big parity drives - especially since the 10 TB and 12 TB drives will drop in price. And at least where I live there isn't so big premium for the larger drives from Seagate and Hgst, while there is a quite big price jump for WD Red drives.

Just an addendum.

 

Let's say a 8 TB drive costs $200.

A 12 TB drive fits 50% more. So at same cost/capacity it would be worth $300.

 

But if we also assumes that a 12 TB drive is meaningful to keep for 5 years while a 8 TB drive is meaningful to keep for 4 years, the longer economical life time means we get a break-even at $375.

 

And if we then note that the 12 TB drive is likely a helium drive that draws maybe 5 W idle while the 8 TB is not helium and draws 7 W idle while still having room for 50% more, then the 12 TB drive would consume about 44 kWh/year or 3.7 kWh/year/TB while one 8 TB would consume 61 kWh/year or 7.7 kWh/TB.

My power figures is for Seagate Ironwolf drives.

 

Let's assume $0.1 / kWh.

 

For 8 TB drive with a cost of $200 and a life span of 4 years, the storage cost would be $200 + 4*8*7.7*0.1 = $225 or $7.02 / TB / year.

For a 12 TB drive with a cost of $400 and life span of 5 years, the storage cost would be $400 + 5*12*3.7*0.1 = $422 or $7.04 / TB / year.

 

And a 12 TB drive is likely to have more than a 5/4 year economical lifetime advantage.

And the above doesn't take into account time savings from needing to perform fewer disk transfers by having a longer time interval between replacing drives.

 

If the drives are stopped for most of the time, then the 8 TB drives will come off a bit better. If we assume each drive is powered 10% of the time, and the 12 TB drive has 0.8W sleep consumption and the 8 TB drive has 0.6 W sleep consumption then the 12 TB drive would have a yearly consumption of 0.89 kW/TB/year and the 8 TB drive would have a yearly consumption of 1.36 kWh/TB/year.

 

So the final lifetime costs with 90% sleep would be:

8 TB drive: $6.39 / TB / year

12 TB drive: $6.76 / TB / year

 

But in conclusion - drive that is 50% larger and costs 100% more has quite similar total lifetime costs.

 

For people running out of drive bays or controller card connectors, buying the cheapest drives can end up really expensive, since it requires earlier replacement of drives or the addition of  an additional controller card.

Thanks for all of the replies everybody!

 

I'd love to replace my 2 TB drives with 8 TB but I'd have to sell the 2 TB drives first.

 

I thought one of the biggest advantages of unRAID is letting me use my existing 2 TB drives and then upgrade them one-by-one to 8 TB drives as I can afford them?

 

Currently I have a total of 12x 2 TB and 2x 4 TB drives.  My plan was to use the 4 TB drives as parity and then upgrade everything to 8 TB as I can afford them, etc.

 

If anybody is interested in purchasing some older 2 TB drives to help me fund newer 8 TB drives please let me know

4 minutes ago, sofakng said:

I thought one of the biggest advantages of unRAID is letting me use my existing 2 TB drives and then upgrade them one-by-one to 8 TB drives as I can afford them?

You can, you just need to copy the current data elsewhere as you need to format before starting to use them in unRAID.

4 minutes ago, pwm said:

Just an addendum.

 

Let's say a 8 TB drive costs $200.

A 12 TB drive fits 50% more. So at same cost/capacity it would be worth $300.

 

But if we also assumes that a 12 TB drive is meaningful to keep for 5 years while a 8 TB drive is meaningful to keep for 4 years, the longer economical life time means we get a break-even at $375.

 

And if we then note that the 12 TB drive is likely a helium drive that draws maybe 5 W idle while the 8 TB is not helium and draws 7 W idle while still having room for 50% more, then the 12 TB drive would consume about 44 kWh/year or 3.7 kWh/year/TB while one 8 TB would consume 61 kWh/year or 7.7 kWh/TB.

My power figures is for Seagate Ironwolf drives.

 

Let's assume $0.1 / kWh.

 

For 8 TB drive with a cost of $200 and a life span of 4 years, the storage cost would be $200 + 4*8*7.7*0.1 = $225 or $7.02 / TB / year.

For a 12 TB drive with a cost of $400 and life span of 5 years, the storage cost would be $400 + 5*12*3.7*0.1 = $422 or $7.04 / TB / year.

 

And a 12 TB drive is likely to have more than a 5/4 year economical lifetime advantage.

And the above doesn't take into account time savings from needing to perform fewer disk transfers by having a longer time interval between replacing drives.

 

If the drives are stopped for most of the time, then the 8 TB drives will come off a bit better. If we assume each drive is powered 10% of the time, and the 12 TB drive has 0.8W sleep consumption and the 8 TB drive has 0.6 W sleep consumption then the 12 TB drive would have a yearly consumption of 0.89 kW/TB/year and the 8 TB drive would have a yearly consumption of 1.36 kWh/TB/year.

 

So the final lifetime costs with 90% sleep would be:

8 TB drive: $6.39 / TB / year

12 TB drive: $6.76 / TB / year

 

But in conclusion - drive that is 50% larger and costs 100% more has quite similar total lifetime costs.

 

For people running out of drive bays or controller card connectors, buying the cheapest drives can end up really expensive, since it requires earlier replacement of drives or the addition of  an additional controller card.

 

Hmmm ....

 

Interesting analysis and I agree with a lot of it, but do have some quibbles:

 

1. I cannot find a 12TB drive for $400. I see them at $500. Appears your price is below market. (Apologies if I missed something)

2. The 8T drives have been at $150. They are currently at $180. Appears your $200 price is above market.

3. The WD Reds are 5400 RPM drives and are Helium filled. I could find no 5400 RPM 12TB drives in my quick search. But I'd think that the 5400 RPM would be more power efficient due to slower spinning and Helium. And drives are frequently spun down anyway.

4. Why do you assume that the 12TB drives have a longer lifespan (5 vs 4 years)? That doesn't make much sense to me.

5. You are not factoring in the cost parity drive. Protecting an array of 8T drives costs $179. Protecting an array with 12T drives costs $500. Double those numbers for dual parity. The cost of parity goes up considerably for 12TB vs 8TB. Strictly looking at $/TB doesn't tell this story.

 

Unrelated to the financial analysis, another consideration is parity check time. My parity checks are taking just under 20 hours now. I can start it at 11/12 at night, and array is ready to go at prime time next day 7/8pm. With 12T parity, that would go close to 30 hours. Hopefully we'll have a parity parity option soon, but I'm not quite ready to go to larger than 8T parity right now. 

 

But I do agree that there is a better argument to go to 12TB than 10TB. And people continuing to invest in drives <= 4T should really be reconsidering, especially if slots are running low.

29 minutes ago, sofakng said:

Thanks for all of the replies everybody!

 

I'd love to replace my 2 TB drives with 8 TB but I'd have to sell the 2 TB drives first.

 

I thought one of the biggest advantages of unRAID is letting me use my existing 2 TB drives and then upgrade them one-by-one to 8 TB drives as I can afford them?

 

Currently I have a total of 12x 2 TB and 2x 4 TB drives.  My plan was to use the 4 TB drives as parity and then upgrade everything to 8 TB as I can afford them, etc.

 

If anybody is interested in purchasing some older 2 TB drives to help me fund newer 8 TB drives please let me know

 

Yes - if you already had an unRAID array, that would be an easy option. But you don't, and not sure how you are going to magically have space to copy your 32T of data to individual disks. But if you have a way then you can execute that strategy. But it is also a valid strategy to buy new drives and migrate existing data to new unRAID array, keeping old platform for backups. Many users do that for a variety of reasons including larger drives and having a good backup / fallback position in moving to a new platform which they have little experience.

 

You would be able to use the 2 4T drives (which can be copied to a new 8T drive) in unRAID without much trouble. But you are still looking at some new drives to unload the 2T ZFS volume.

My prices isn't in USD but in SEK.

Some prices I have:

Seagate Barracuda Pro 8TB 2500 SEK

Seagate Ironwolf Pro 8TB 3000 SEK

Seagate Ironwolf 8TB 2350 SEK

 

Seagate Barracuda Pro 12TB 4500 SEK - so 80% more expensive than 8TB from same family

Seagate Ironwolf Pro 12TB 4500 SEK - so 50% more expensive than 8TB from same family

Seagate Ironwolf 12TB for about 3900 SEK - so 66% more expensive than 8TB from same family

 

But I made my computations using the assumption of twice the price for 12TB than for 8TB, giving the 8TB a bit of advantage compared to my actual prices.
 

 

I'm not talking about longer physical lifetime - I'm talking about longer economical lifetime, i.e. how long it would be meaningful to keep the drives in the storage server. I throw out too small drives because they waste drive slots - installing a bigger drive from the beginning means I can keep the drive for a longer time before it's time to throw it out and replace with a larger drive.

 

 

I'm most definitely taking into account the cost of the parity - I'm just assuming that the initial server will have fewer disks but that the actual storage needs will grow with time and that there will be a reason to pop in additional data disks. Having a large parity from the start means it's ok to just pop in large data drives without having to first replace the parity drive. And having large data drives means there are more free drive slots for additional disks.

 

 

When considering parity checks - the runtime isn't necessarily proportional to the disk size. It's also a question of sustained transfer rates. The Ironwolf drives are 7200 RPM, which affects sustained transfer rates.

1 hour ago, pwm said:

My prices isn't in USD but in SEK.

Some prices I have:

Seagate Barracuda Pro 8TB 2500 SEK

Seagate Ironwolf Pro 8TB 3000 SEK

Seagate Ironwolf 8TB 2350 SEK

 

Seagate Barracuda Pro 12TB 4500 SEK - so 80% more expensive than 8TB from same family

Seagate Ironwolf Pro 12TB 4500 SEK - so 50% more expensive than 8TB from same family

Seagate Ironwolf 12TB for about 3900 SEK - so 66% more expensive than 8TB from same family

 

But I made my computations using the assumption of twice the price for 12TB than for 8TB, giving the 8TB a bit of advantage compared to my actual prices.
 

 

I'm not talking about longer physical lifetime - I'm talking about longer economical lifetime, i.e. how long it would be meaningful to keep the drives in the storage server. I throw out too small drives because they waste drive slots - installing a bigger drive from the beginning means I can keep the drive for a longer time before it's time to throw it out and replace with a larger drive.

 

 

I'm most definitely taking into account the cost of the parity - I'm just assuming that the initial server will have fewer disks but that the actual storage needs will grow with time and that there will be a reason to pop in additional data disks. Having a large parity from the start means it's ok to just pop in large data drives without having to first replace the parity drive. And having large data drives means there are more free drive slots for additional disks.

 

 

When considering parity checks - the runtime isn't necessarily proportional to the disk size. It's also a question of sustained transfer rates. The Ironwolf drives are 7200 RPM, which affects sustained transfer rates.

 

The BestBuy Easystore 8T are an unusual commodity, costing (during the holiday) $150 for 8T. That is well less than 1/3 the cost of the 12T.

 

If prices in US were similar to prices in Sweden, I would be buying 12T too!