• Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
    Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.

ZFS partitioned drives

D

davee

n00b
Joined
Mar 11, 2011
Messages
9
Can anyone point me to an incredibly simple guide for setting up ZFS RAIDZ on partitioned drives please?

I've 3*500, 1*1000, 1*1500 and 1*2000GB drives and I'm looking to get redundancy for 1 drive failure and of course, the maximum capacity possible.

The way I see it I have to partition down to 500GB chunks but when it comes down to doing that, and then working out the best way to allocate them to vdevs I come unstuck. So yea, I can't be the only person to have asked this right, is there a guide?
 
With the KISS principle in mind I would probably go with:

MIRROR - 1.5TB
1.5TB drive
2.0TB drive

Mirror - 0.5TB
0.5TB drive
0.5TB drive

Mirror - 0.5TB
1TB drive
0.5TB drive



All in one pool will give you 2.5TB of single disk per group fault tolerant storage. And it's really the simplest config to do that and use all your disks.
 
Not to sound ungrateful but you've just created a 2.5TB pool from 6TB worth of drives... While it may be easiest I'm maybe after something a little smarter than that. Thanks for your input though.
 
If you want something smarter buy some new drives then :)

If you partition your 2000GB drive into 4 500GB partitions to use raidZ then you have just basically rendered the raid-Z useless. Your one physical drive is now 4 "virtual drives". Redundancy is only on the "virtual" level. So one physical drive going out (if it's the 2.0TB drive) would take out 4 virtual drives and blow up your pool/vdev. You are also going to have terrible performance if you can get this to work as well, and only have any kind of protection if one of the 500Gb drives dies. if the 1Tb, 1.5Tb, or 2Tb drive dies your entire dataset is lost with a partitioning Raid-Z scheme.

Basically what you are suggesting is that you can treat each of your drives as a single 500GB drive and get 0.5 * 6 of raw space (3TB), and with a parity drive you will be back down to 2.5TB usable. And you only have protection against one drive failure.

With the 3 mirrors you have protection against one drive failure worst case, 3 drive failures best case. And the same amount of usable space.

If you are going to give up drive failure protection you can just run everything together as a JBOD setup.
 
The way I see it, something like splitting everything into 500GB chunks gives me the ability to create 4 vdevs each of 1500GB in size. That 1500GB coming from 500GB partitions from different drives meaning I can RAIDZ them and sacrifice 500GB per 1500GB to parity. That way if one drive dies I'm covered and I'll have 4TB of space for data. Am I wrong?

Pic related.

driveplan.png
 
davee said:
The way I see it, something like splitting everything into 500GB chunks gives me the ability to create 4 vdevs each of 1500GB in size. That 1500GB coming from 500GB partitions from different drives meaning I can RAIDZ them and sacrifice 500GB per 1500GB to parity. That way if one drive dies I'm covered and I'll have 4TB of space for data. Am I wrong?

Pic related.

driveplan.png
Click to expand...

its highly insecure by design.
disks are really cheap today (2TB=70 Euro), data is valuable

I would never do nor think about.

Gea
 
davee said:
The way I see it, something like splitting everything into 500GB chunks gives me the ability to create 4 vdevs each of 1500GB in size. That 1500GB coming from 500GB partitions from different drives meaning I can RAIDZ them and sacrifice 500GB per 1500GB to parity. That way if one drive dies I'm covered and I'll have 4TB of space for data. Am I wrong?
Click to expand...

That would probably work, given you make sure each partition lives in a separate VDEV. Performance will be pretty terrible (since what ZFS think is a disk is actually just a portion or timeslice of a real disk). And it's really outside what people would normally do, so the odds are higher that weird bugs or incompatibilities will pop up. And when a drive does fail if it's going to be a bit of a rubiks cube to replace.

Two more 2 TB disks would let you have a a 3 disk RaidZ1.
 
This is as much an exercise is making technology bow to me. True pretty much anything can be solved by throwing money at it but technology is supposed to do what we want it to...

So this is so far out there that there aren't any guides? If I manage it I'll write one. Thanks for your time.
 
You must log in or register to reply here.
Back
Top