For my first TrueNAS setup I want to have a raidz1 with a total of four disks.
My initial plan was to start with only two disks and then at a later point in time expand it when necessary. But right now I’m trying to find out why TrueNAS is not supporting a raidz1 setup with only two disks? Are there any mayor downsides, when I create a pool via CLI with only two disks?
What size disks are you planning to use? Raid-Z1 is not recommended for large drives, (over 4 TB?), due to resilver times and the pool being in danger of losing another disk to failure.
ZFS Raid expansion is working but the values in the GUI would appear wrong. You have to use the CLI to see actual free space, etc.
@maltepk
It is more like over 1TB / 2TB.
The rest of the issue is the chance of a un-correctable read error on another disk during resilver. Not necessarily another failed disk. Any file impacted by an un-correctable read can not be re-silvered properly, and will have to be restored from backups.
The issue is more theoretical and statistical math, because of how the un-correctable error rates are defined for consumer grade disks. Enterprise quality disks have slightly higher values, but still have limits.
Here is a Western Digital Red Plus 4TB’s value:
Reliability/Data Integrity
Non-recoverable errors per bits read - <1 in 10^14
Here is a Western Digital Red Pro 4TB’s value:
Reliability/Data Integrity
Non-recoverable errors per bits read - <1 in 10^15
As you can see, “Pro”, (or supposedly higher quality drives), have higher values listed.
Does it make a difference?
Most of us don’t know, yet.
Do some / many people exceed that using RAID-5 / RAID-Z1?
Yes.
Do many do so without problems?
Yes.
It’s all about your comfort level. I run my media server’s media pool as a ZFS stripe, no data redundancy. Have had to restore perhaps a dozen or 2 files over the last 10 years. But, that is the beauty of ZFS. It will tell you which files are lost, so you can restore them from backups.
I’m planning with 4x18TB at the end. Is that a problem with raidz1?
I don’t want two mirrors resulting in 18TB less storage capacity.
See my prior post. We posted close together, so you may not have seen it yet.
In the end this is ALL about what level of risk you are prepared to take with your data, and that will depend on how irreplaceable it is, what the impact would be if you lost it and whether you have on-site or off-site backups of the most important parts of it.
Personally I would consider that RAIDZ1 on a 72TB array is a bit too risky for my own taste. I configured my 20TB array (5x 4TB) as a RAIDZ1 but later thought that RAIDZ2 would have been a better choice and now I am stuck with it - and in the end I paid the same as 2x4TB for a lifetime purchase of 500GB of cloud storage in order to mitigate this risk and mitigate risks of e.g. a fire destroying the NAS.
My knowledge about zfs and raidz is really too little to do a proper risk evaluation…
My data mainly consists of video/photo files. I’m fine losing a few files, but would not be too fine losing a whole disk. My most important data (e.g. finished video projects) is backed up on a external hdd at a different location.
I will use Toshiba Enterprice Capacity MG Series.
The problem isn’t loosing a disk, it is losing the entire pool.
This may help
BASICS
iX Systems pool layout whitepaper
As @SmallBarky says, with ZFS you either get a handful of corrupted files (small risk) or you get a dead pool (also small risk). You never lose a single disk’s worth of data.
Only you can evaluate your own risks. I would be torn between the cost (50%) of RAIDZ2 on 4 drives, and the risks of a 2nd drive dying. I have 5 drives, so RAIDZ2 would be only 40% but I have less space, yet I still wish I had done RAIDZ2 instead of RAIDZ1.