• Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
    Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.

Thinking about moving to ZFS

eg1l

n00b
Joined
Dec 15, 2010
Messages
20
I am researching ZFS, and I believe that this is the best forum out there when it comes to designing huge filesystems for personal use. I myself are intrigued by it's features such as data corruption prevention.
However, I am a total newbie when it comes to ZFS so i'm asking for a little help. I currently have a Norco 4220 running 20 drives with mdadm raid6. My favourite OS is Ubuntu, and Ubuntu has now native support for ZFS.

My question is, how would you set this up? Should I run one raidz2 on all 20 drives, split it into two radiz2's? I would like to have as much space as I can.

If I create a raidz2, how do i add a new harddrive to that raid? Is that one thing that ZFS is not capable of? This is not such a big deal now as I have filled all my trays, but there might be that day when a expander chassis comes knocking on my door. :)
 
Better to do two raidz2 vdevs striped together. e.g. something like 'zpool create foo raidz2 N N N N N N N N N N raidz2 N N N N N N N N N N N'
 
I would do two 10 disk raidz2 groups, and create one zpool.

If you need to add more disks afterwards, then you should add a whole group to the zpool. You should not one disk, and later add another disk. Save up 10 disks, and create one raidz2 and add the raidz2 group to the zpool.

If you expand the zpool with a group of disks, then you can rebalance the data. Because all data is on the old disks. When you add new disks, they are empty. The old data is not automatically moved to the new disks. To rebalance data between old and new disks, you can copy off data to another server and then copy back. Or, which is easier: create a new directory and then move the data to the new directory.
 
Why do you want two raids instead of one? If one of the raids go down, the data in the pool would be lost, wouldn't it?

Also, do I have to create a filesystem on top of the pool, or could I simply store all my data directly on the pool, eg. /tank?
 
You misunderstand. What we are suggesting is basically raid60. You would need to have 3 drives fail in one vdev to lose the pool.
 
Also, do I have to create a filesystem on top of the pool, or could I simply store all my data directly on the pool, eg. /tank?

You can, but I would suggest creating several zpool folders. I have 3.
/tank/important/ for important files, it gets a snapshot every night that last for 2 weeks, once a week that last for a month, and once a month that last for 2 years.
/tank/non_crittical/ for Downloads and movies, it gets a snap once a day and lasts 4 weeks.
/tank/crashplan/ for crashplan that does not get snapped at all.

Once you share the folders in windows you can create folders within windows. Also, I don't share everything the same. Before, I just put everything on /tank; then ran out of space when my snaps were saving all my downloaded files that I had deleted. I learned the hard way.
 
You can, but I would suggest creating several zpool folders. I have 3.
/tank/important/ for important files, it gets a snapshot every night that last for 2 weeks, once a week that last for a month, and once a month that last for 2 years.
/tank/non_crittical/ for Downloads and movies, it gets a snap once a day and lasts 4 weeks.
/tank/crashplan/ for crashplan that does not get snapped at all.

Once you share the folders in windows you can create folders within windows. Also, I don't share everything the same. Before, I just put everything on /tank; then ran out of space when my snaps were saving all my downloaded files that I had deleted. I learned the hard way.

I would definitely do that and store my important files in a separate zpool folder. How much space does a snapshot require (depends on how much data it is a snapshot of)?
And, would I be better off with two raidz's as far as performance is concerned?
 
One zpool is made of groups of disks. I create two groups, each group consisting of 10 disks in raidz2. These two groups I merge and create one big zpool.
 
And, would I be better off with two raidz's as far as performance is concerned?


Yes, generally speaking!
Rebuild time after a failure would be less too!

However, if you mean RAIDZ (rather than RAIDZ2), then a double disk failure in either vdev would mean total loss of pool (the risk of this happening is greatest during a rebuild after a drive failure - the risk is that a second drive fails during the rebuild) - RAIDZ2 is safer!
The "drawback" is that two 10 drive RAIDZ2's means losing 4 drives' capacity to redundancy.

You could also go with two pools, though you might lose out a little on IOPs (that depends on data location and the type of IO though) - the upside is that losing a pool means only losing half your data (if you can really call that an upside - I suppose it's better than total loss though)
 
You don't get total loss of all your data if you lose too many disks with your ZFS raid, you just lose some files.
 
You don't get total loss of all your data if you lose too many disks with your ZFS raid, you just lose some files.

If you have N+x parity disks, and you lose x+1 disks you are screwed if you don't have a backup.
 
More specifically, what I have seen in this situation is that ZFS will not let you access data on the pool, so whether all data is lost or not doesn't really matter.
 
I would definitely do that and store my important files in a separate zpool folder. How much space does a snapshot require (depends on how much data it is a snapshot of)?
And, would I be better off with two raidz's as far as performance is concerned?

Looks like my smallest Snaps are around 38K for 1.3 GB, That means there was no movement of Data since the snap was taken. They start to increase in size when you modify or delete files.
 
One guy had like 10.000 snapshots and they consumed 1.4GB or so. Only the changes where snapshoted.
 
Back
Top