Need guide to set up RAID 10 in Freenas 8.3

[indivision]

I've spent a great deal of time scouring articles about this. But, haven't found a clear explanation for how to set up RAID 10 via the GUI in FreeNAS 8.3.

Given 4 disks to work with, this is what I have so far:

Step 1 *** Volume Manager > Volume Name: r0, Select 2 Member Disks, ZFS Filesystem, Mirror Group
Step 2 *** Volume Manager > Extend volume "r0": 2 Member Disks, ZFS Filesystem, ????? Group

In step 2, do I select "mirror" or "stripe" for the second two volumes I am extending to? Or, do I need to do something entirely different to correctly set up RAID 10?

I appreciate any help anyone can offer!

 

[bollar]

Arg. Let me rewrite this...

 

[bollar]

to extend a ZFS mirror, add the same number of drives to a new mirrored vdev. The resulting striped mirror is a RAID 10.

It may not be the clearest, but the docs say:

When extending a volume, ZFS supports the addition of virtual devices (vdevs) to an existing volume (ZFS pool). A vdev can be a single disk, a stripe, a mirror, a RAIDZ1, RAIDZ1, or a RAIDZ3. Once a vdev is created, you can not add more drives to that vdev; however, you can stripe a new vdev (and its disks) with the same type of existing vdev in order to increase the overall size of the ZFS pool. In other words, when you extend a ZFS volume, you are really striping similar vdevs. Here are some examples:
to extend a ZFS stripe, add one or more disks. Since there is no redundancy, you do not have to add the same amount of disks as the existing stripe.
to extend a ZFS mirror, add the same number of drives. The resulting striped mirror is a RAID 10.
to extend a three drive RAIDZ1, add three additional drives. The result is a RAIDZ+0, similar to RAID 50 on a hardware controller.
to extend a RAIDZ2 requires a minimum of four additional drives. The result is a RAIDZ2+0, similar to RAID 60 on a hardware controller.
In the “Volume to extend” section, select the existing volume that you wish to stripe with. This will grey out the Volume name field and select ZFS as the filesystem type. Highlight the required number of additional disk(s), select the same type of RAID used on the existing volume, and click Add Volume.

 

[indivision]

Thank you bollar.

I had read that in the documentation. There are a couple of other factors that kept it confusing:

A) A lot of other explanations of RAID 10 describe it as (assuming 4 drives) making two separately mirrored pairs. Then striping them together in a third command. The flow in the GUI doesn't clarify that the striping step happens.
B) Older versions of FreeNAS did involve more of a 3 step process to create RAID 10. Those tutorials and videos end up in the hands of someone like me, googling around for a solution, reinforcing the concern that a step is being missed.
C) You can extend with a mirrored or striped group. This can reasonably be interpreted to mean that the user is omitting striping altogether if they select mirrored.

Just wanted to clarify so that maybe the docs or GUI can be improved. Or, at least someone might come across this thread and save some time.

For something that I'm setting up to be a reliable, long-term server, I wanted to be sure. Thanks again for clarifying!

 

[bollar]

Yes. The documentation is a wiki and there's always the opportunity to improve it!

 

[dandal88]

Did anyone ever follow-up here? (C) "You can extend with a mirrored or striped group. This can reasonably be interpreted to mean that the user is omitting striping altogether if they select mirrored" is the very same questions I'm wondering about to setup at RAID 10 ... I've searched a bunch on forums and read the documentation, but can't find a clear answer here.

Can anyone who has done this clarify?

 

[indivision]

Did anyone ever follow-up here? (C) "You can extend with a mirrored or striped group. This can reasonably be interpreted to mean that the user is omitting striping altogether if they select mirrored" is the very same questions I'm wondering about to setup at RAID 10 ... I've searched a bunch on forums and read the documentation, but can't find a clear answer here.

Can anyone who has done this clarify?

I am by no means an expert. But, this is my understanding:

In Step 2 (from my initial post), you select "mirror". The act of "extending" sets up striping between the two mirrored groups. So, creating a mirrored pair of 2 drives and extending that group to another mirrored pair of 2 drives would create a RAID 10 group.

I couldn't confirm this with absolute certainty. But, I set this configuration up around the time this thread started. The file size and behavior of the group is what I expected from RAID 10. I even replaced a failed drive in the group since then. So, I'm pretty sure it worked. :)

 

[dandal88]

Great, thanks for the update & quick response -- I also did this and is what I surmised from other forums -- good to hear that you were able to replace failed drive .... makes me think you did something right!

 

[paleoN]

Can anyone who has done this clarify?

I couldn't confirm this with absolute certainty. But, I set this configuration up around the time this thread started. The file size and behavior of the group is what I expected from RAID 10.

Allow me to state it plainly, "ZFS stripes all top-level vdevs together.

 

[Joseph Day]

You said ZFS Stripes all top-level vdevs together. So what does a RAID 10 look like in FreeNAS 9 ? I see two mirror volumes but it is not absolutely clear that they are striped together.

 

[cyberjock]

A RAID10 would look like 2 mirrored vdevs. :)

 

[elangley]

I had the same issue with the documentation in trying to create a "RAID 10" with ZFS.

After creating a Mirrored vDev with 2 1.5TB drives I extended it by added 2 more 1.5TB drives and selected Striped. The result was a 4TB volume, not what I was expecting!

From http://doc.freenas.org/index.php/Volumes I read this:
to extend a ZFS mirror, add the same number of drives. The resulting striped mirror is a RAID 10.

The "add the same number of drives" is the part that is not clear.

>My mistake was in creating the Volume and extending it instead of adding additional (mirrored) drives during Volume creation. <

Now in Volume Status there are two mirrors, mirror-0 and mirror-1. I guess it is assumed that these are striped because it is not stated.

The size is now correctly reported as 2.7TB.

~eric

 

[cyberjock]

Vdevs are always striped, so it is assumes because of how multiple vdevs interact.

 

[aran kaspar]

Welcome to FreeNAS 9.3 Gentlemen...
Looks very much like RAID-10 to me.
But someone please tell me... what are the main differences between this and RaidZ2?
If 2 drives fail and they both happen to be in the same mirror... have I lost the entire volume?
e.g. (ada0 and ada4) in screenshot

Yes I know I'm degraded back there...

 

[gpsguy]

Yes, that is correct.

With RAIDz2 the liability is reduced to any 2 drives.

If 2 drives fail and they both happen to be in the same mirror... have I lost the entire volume? e.g. (ada0 and ada4) in screenshot

 

[aran kaspar]

So the software developer people are telling me to use 10 instead of 5... for SQL.
I know there is some sort of write whole thing with 5 but I just can't see it being worth the possible complete loss.
Especially when I'm using 4 SSDs.
Unless there were 2 seperate backups that were consistently solid.

Anyone, feel free to chime in... Would love to hear your opinions.

 

[indivision]

Do you not have a back-up system in place?

 

[cyberjock]

Databases like SQl are very much a "random read/random write" workload, so more vdevs are better. So yes, RAID10 would be a better idea than RAID5.

If you aren't backing up SQL databases regularly, expect to be unemployed when it goes sideways. Databases are a whole different beast from anything else, and backups are absolutely 500% mandatory. I say 500% because you *really* need good regular backups!

 

[aran kaspar]

Thanks for the input guys.
I've learned that r10 is gonna be best here and yes yes yes backups are tri-fold. The first three rules of storage.

These are r/w intensive SSDs... Hoping I don't have a failure in the next few years but if so I'll have some spares handy and hope that we never have to recoup from backup. I'm just afraid of shit getting outta-sync. Roughly 450 employees, eek...

 

[indivision]

I think that the odds are in favor of you not needing to go to back-up. But, for greater peace of mind, you could do test runs of restoring data from your back-ups and document the process before you put the system into live use.