So, I’ve been using 8x16TB EXOS X16 and two of my disk failed and i did RMA as my drive was under warranty. However, the drives i received are 1x16TB EXOS X18 and 1x16TB EXOS X24.
Now, 6 of my drives are X16 family and uses the same firmware but then the RMAed drives are X18 and X24 which uses different firmware. Moreover, this X24 is 4Kn and the others are 512.
If i replace the drives, will it cause any problem in terms of performance or stability?
In any NAS, does all the drive has to be same firmware and model?
I am using the 18TB X18 model and I experimented with the 512e vs 4Kn option and didn’t experience any issue when rebuilding the pool.
You can update how the drive should handle the data 512e vs 4Kn but it will delete all the data present on the disk.
While using the 4Kn mode when other disk in the pool are using 512, there is in theory no underlying issues. However, I came across a video on a conference across the topic on ZFS which addressed this concern and there might be some subtle corner cases that might affect how the blocks are being allocated on the pool across multiple drives.
My “opinion” is 4K (4096 bytes). Why do I say this? Because 4K is the native sector size of an Advance Format Drive and transfers would be faster for files larger than this 4K bytes size.
512e (512 bytes emulated) is for compatibility with older hardware. It writes eight (8) 512 byte blocks to a 4096 byte block. If you change a single 512 byte entry, the rest of the 4096 bytes must be read and then write the entire modified 4096 bytes.
Now, if you have bazillions of tiny files, 512e would be better from a space use perspective.
And I do mean very small files. You can Google it for an explanation on used space compared to 512 or 4096.
Advice, just plug it in. Let TrueNAS handle it for you. but I suspect you are just trying to understand the mechanics of it all.
This is not a problem with TrueNAS, since it defaults to an “ashift” of 12 (4 KiB) for new pools. The smallest write to disk is 4 KiB, no matter if the disk is low-level formatted as 512e or 4Kn.
Let me expound on this a little. A drive manufacturer can lie. WHAT! But they can advertise a drive having say “4TB” (which should mean 4,000,000,000,000 bytes of data) and actually having a tiny be less than the full amount. Creating a VDEV is not a problem. It is when you replace a drive with a different drive. It doesn’t matter the manufacturer, the replacement drive must be equal or larger than the old drive.
TrueNAS should be adding back a small unused partition when a pool is created. We use to have a partition called “SWAP”, by default was 2GB. This allowed 2GB of variance. The new partition, well I have not heard much about it recently but it will be probably a few MB in size and that would allow for this kind of issue to be solved. People have had it, it sucks. This new change (whenever it is added) will help. Heck, it may already be included. I should look into this for my own knowledge.