Upgrading from 2 disk stripe to 4 disk zfs1

[JCasanova]

Hi all, I've got a few questions...

I have an aging freenas server that has been up for ~2 years. I use it primarily as an NFS target for a few application servers.

It has only 2 disks in it... both are also aging. 2x 2TB disks in a stripe.

I just bought 4x 2TB drives to replace those two drives. The data on these drives isn't 100% mission critical but expanding to a ZFS1 pool with 4 disks should increase performance and give me a single disk failure without losing data, which is better than I have now.

I'd also like to install a fresh copy of Freenas to the server while I perform this upgrade (before or after the data migration). It's currently running 9.10 but I've managed to muck up the install as I was very naïve when I first installed it and went too much under the hood... (jails don't work, and many other issues, all my fault.)

My preliminary gameplan is to:
1. Install the new disks, build a new zfs1 pool with them.
2. Copy data over to new pool.
3. Remove old disks.
3. Install new copy of freenas, reimport new pool in place of old pool.
4. Reestablish NFS shares and such.

Do you foresee any issues with that plan? Any pointers/advice? Would that be best practice in this situation?

 

[BigDave]

How many hours do your current two drives have on them?
How much redundancy do you want/need to protect your data?
How much data do you currently have, and could you move it all off the original disks?
How much room for more data would you like to end up with?

 

[JCasanova]

How many hours do your current two drives have on them?
How much redundancy do you want/need to protect your data?
How much data do you currently have, and could you move it all off the original disks?
How much room for more data would you like to end up with?

Both drives are over 40,000 hours of powered on, last I checked. End of life.
One disk of redundancy is enough for my purposes. This is merely bulk/archive storage for a cyber security lab I run for the educational purposes of my coworkers and myself - a lot of logs, notes, screenshots, video captures, reverse-engineered malware, not-in-use virtual machine images, etc. All critical data is not stored on this server. Virtual machine data is not RUN off of this server, merely stored on it when it isn't being used.
I currently have roughly 3.1 TB of data. I could move it to the new disks, obviously. I don't have a third place to put it.
Any extra space is helpful. We haven't had to start deleting old data but we are getting to that point. The nature of our work will cause this library of files to continuously grow. The more we can keep the better, but it isn't so critical that a multiple disk failure would result in a crisis.

 

[BigDave]

With that in mind, take the four new drives and create a RAIDz1 volume (one disk capacity to be used for redundancy,
three remaining disks for storage). See my signature for @Bidule0hm RAIDz calculator for what you would end up
with as USABLE space. My recommendation would be to add one more 2TB drive for room to grow. Perhaps use
one of the older drives and buy another quickly, in case the old girl kicks the bucket.
Make sure you have smart testing set up on your drives and burn in the new ones before copying any data to them.

 

[Ericloewe]

ZFS1 pool

Please use proper terminology. There is no such thing as "ZFS1" and there is no need to propagate an improper designation that will only lead to confusion, which inevitably leads to lost data somewhere.

 

[jgreco]

Please use proper terminology. There is no such thing as "ZFS1"

ZFS 1 was included in Solaris 10, June 2006 edition.

But we do request that you use proper terminology to minimize confusion. Random abbreviations and imprecise terminology do indeed lead to bad places.