FreeNAS Replication - Target without redundancy

[N20Visuals]

Hi Everyone,

i've been reading through the forums all day but didnt find anything that would answer my question:
I currently have a FreeNAS box running with a couple of Jails and VM's and everything is working flawlessly.
I'm strictly speaking from a home users perspective, so bear with me...

My Plan now is to build a second machine for backup purposes which will only be online once a week to update the backup and then offline again.
Since i'm on a budget my plan is to use single-disk vdevs for the machine, so no redundancy whatsoever. I do know that this is not quite the optimal scenario but i just cant afford enough hard drives to make the second machine run any kind of redundancy.

My Question is: When i sync the Backup-Box via a weekly replication task using snapshots and one of the drives in the replication target dies, will i have to start all over?
In my understanding the way i plan to use the disks when one of the drives failes the whole pool is destroyed and i would have to sync all the data again? Am i corret?

Thank you for any help :)

 

[adrianwi]

If the replication target has no redundancy and one disk fails then you lose that pool, yes.

I ran a second machine for a few years as a backup target 24x7 and it had a 5 x 4TB stripped pool, so if any of the disks had failed I would have lost my backup. I took this risk as my main FreeNAS machine has a RAIDZ2 pool, so could tolerate 2 disks failing without losing data. I also had anything really important backed up to the cloud.

My backup machine now has a 9 x 4TB RAIDZ1 pool, which is still probably considered fairly high risk, as the rebuild of so many 4TB drives could well result in errors and potential data loss, although it gives me that chance.

 

[Chris Moore]

Since i'm on a budget my plan is to use single-disk vdevs for the machine,

Strict interpretation of your statement indicates that you will have multiple vdevs, but each vdev will be a single disk. To make a pool this way, yes you would loose the whole pool.

What you could do is make each disk a separate pool. Then you only loose the data on that disk. It makes backup a little more difficult, but you don't put all your eggs in one basket... It's an idea.

which is still probably considered fairly high risk,

Yes, but it is just a backup. I run RAIDz1 in my backup pool also.

 

[N20Visuals]

Thank you for your fast replies. So it's like I thought. Since I'm running a RAID10 with 8x4Tb in my main system im not worried when the backup machine fails, it'll just be rather inconvenient since the system will be connected through a rather slow DLan connection. I might upgrade to a three or more disk z1 down the road, but for now this seems good enough when I check drive integrity thoroughly.

 

[sretalla]

Just be aware that you could not only fall victim to a lost disk but also to data corruption/bit-rot (which could be detected but not corrected without at least one parity disk) on the striped replica... having an additional copy is already something, but if that copy isn't reliable, how good is it really.

 

[Mark St.]

its really difficult if you can afford only a limited number of disks and you try to find a optimum.

i read about the possibility of disk failure during a resilver in another thread, ordered instantly another disk to improve my pool from Z1 to Z2....and after i got this disk one day later i decided to use it as another backup-disk instead....