vdev or new pool

[samehfattouh]

i have pool Raid z 1 with 3 HDD 10 t for each
i have new 3 HDD 4 T for each
i want best way to give me more space i have 2 choice
1- add new 3 HDD 4 t as VDEV
2- create new pool with 3 HDD 4 t for each
please help me for best choice
thanks

 

[samehfattouh]

i have pool Raid z 1 with 3 HDD 10 t for each
i have new 3 HDD 4 T for each
i want best way to give me more space i have 2 choice
1- add new 3 HDD 4 t as VDEV
2- create new pool with 3 HDD 4 t for each
please help me for best choice
thanks

please any one Help me

 

[Whattteva]

RAIDZ1 on 10 TB drives, brave soul you are. Has any of those drives ever failed you? It would probably take at least a day to resilver that, possibly longer. We haven't even talked about how long it would take to upgrade those 10 TB drives to bigger drives. A process that could take days or even weeks.

3x 4TB on RAIDZ1 is a lot saner, probably 8+ hours resilver?

As for which choice you want to do, I personally think it's more of a personal preference. You can either add a vdev of another RAIDZ1 or make another pool with that. The only difference is you will end up with either 1 big pool or 2 smaller size pools. Also, redundancy is at vdev level only. So if one of your vdevs fail, the whole pool fails whereas if you split into 2 pools, failure of 1 vdev means only failure of that pool, not both.

Personally, I prefer 1 big pool, but I also wouldn't run a RAIDZ1 though. I prefer striped mirrors as I feel like it provides the best balance between performance, reliability, and time efficiency.

 

[Davvo]

Brave is a kind word.
Best option in my opinion is to create a RAIDZ2 pool with all 6 drives.
However, depending on the use case mirrors could be a better choice.

 

[samehfattouh]

RAIDZ1 on 10 TB drives, brave soul you are. Has any of those drives ever failed you? It would probably take at least a day to resilver that, possibly longer. We haven't even talked about how long it would take to upgrade those 10 TB drives to bigger drives. A process that could take days or even weeks.

3x 4TB on RAIDZ1 is a lot saner, probably 8+ hours resilver?

As for which choice you want to do, I personally think it's more of a personal preference. You can either add a vdev of another RAIDZ1 or make another pool with that. The only difference is you will end up with either 1 big pool or 2 smaller size pools. Also, redundancy is at vdev level only. So if one of your vdevs fail, the whole pool fails whereas if you split into 2 pools, failure of 1 vdev means only failure of that pool, not both.

Personally, I prefer 1 big pool, but I also wouldn't run a RAIDZ1 though. I prefer striped mirrors as I feel like it provides the best balance between performance, reliability, and time efficiency.

what's best scenario you think

please step by step

and keep my data

 

[samehfattouh]

Brave is a kind word.
Best option in my opinion is to create a RAIDZ2 pool with all 6 drives.
However, depending on the use case mirrors could be a better choice.

what's best scenario you think

please step by step
how to convert to raid z 2

and keep my data

 

[Davvo]

what's best scenario you think

please step by step

and keep my data

Backup your data, destroy the pool, create the new pool (RAIDZ2 with all 6 disks).

 

[Whattteva]

what's best scenario you think

please step by step

and keep my data

Well few scenarios. You either keep going the way you are doing now or do what @samehfattouh said above (RAIDZ2 with all 6 disks).

I'm personally not as rigid as others here and I think RAIDZ1 on 4 TB drives are ok, but I think doing it on 10 TB drives are way too risky though.

 

[Davvo]

I want to stress that you can't convert a RAIDZ1 into a RAIDZ2, you have to destroy your pool and recreate it.
So backup, backup, backup!

 

[pschatz100]

i have pool Raid z 1 with 3 HDD 10 t for each
i have new 3 HDD 4 T for each
i want best way to give me more space i have 2 choice
1- add new 3 HDD 4 t as VDEV
2- create new pool with 3 HDD 4 t for each
please help me for best choice
thanks

As you haven't said anything about how much storage you need, your use for the storage or your system configuration, it is difficult to make an informed recommendation but I can share a couple of observations about ZFS pools:

• RaidZ1 with disks as larage as 10TB is considered risky
• If you combine all six disks into one RaidZ2 pool, then your gross capacity will be limited to 16TB which is less than you have now. When mixing disks of different sizes in one pool, the amount of space used from each disk is limited by the smallest disk in the pool. In your case, that means the system will only use 4TB of the 10TB disks if you mix them in one VDEV.

If you create a new VDEV with the three 4TB disks and add it to your existing pool, then you will increase capacity - but you will still have what is considered a risky setup based upon RaidZ1. If either VDEV becomes unreadable, then you lose the entire pool. Not a likely event, but I wouldn't want that to happen to me. So if it were me, and my only option was to use the disks you have then I would create a new pool with the 4TB disks. That way, if one pool goes down you are not at risk of losing the other pool.

I presume you have a good backup of your data?

 

[Morris]

The paranoia expressed in this thread is baseless. Risk management must take into account how much down time is permissible, the ability to restore from backup (if there is no backup then any storage system is huge risk), and the ability to meat the recovery time objective.

Disks are very reliable and the odds of a second disk failing in a day or two after redundancy is lost in a RAID Z1 is very low yet not zero. The wise designer takes all of this into account.

I witnessed two double disk failures in my 30 year IT career. In both cases the restore from backup was successful. In one case the array that failed was a backup target and the alternate target was still fine. A full backup was run and nobody was affected. In the second it was a production system and the recovery time objective was easily met as a proper DR drill had demonstrated it was possible.

 

[pschatz100]

There has been quite a lot of discussion and information published about the characteristics of ZFS filesystems and why RaidZ2 is preferred over RaidZ1 when the disks are larger than 1 or 2 TB. I won't repeat it here (search is your friend.) We're just pointing out best practice.

 

[Whattteva]

The paranoia expressed in this thread is baseless.

I witnessed two double disk failures in my 30 year IT career. In both cases the restore from backup was successful.

It's not baseless. You've witnessed two in 30 year IT career. Congratulations, you're witnessing things in an ideal enterprise setup and you have backups to boot. A lot of people frequenting this forum don't have those optimal setups and a lot tend to run Frankenstein setups and using the RAIDZ as the ACTUAL backup. Just a few days ago, there was a post about someone having dual disk failures after a power outage (no UPS) here. This guy here, runs some funky setup that causes even his RAIDZ2 pool to fail. This guy here lost his pool while resilvering a RAIDZ1 pool. And yes, all had no backups and I could keep going with examples from these forums. I'm sure you can find them too using the search function! One could say that what is baseless here is actually your confidence in assuming that everyone that visits this forum all have setups mimicking your 30-year IT career.

I have literally a collection of "greatest hits" of people that lost their pools due to (most commonly) power outages and other sub-par setups with no backups over the years littering this forum. We're giving advice while being mindful of the type of audience that tend to visit these forums. You would notice also, as is the case in this thread, the post count of the OP tends to be in the single digits. They generally aren't as well-versed as you may be and only visit the forums to fix their problem. And usually, only now will they learn more about safer best practices and recommendations. Unfortunately, they often have to learn it the very hard way with a catastrophic pool failure.