Datasets created during initial configuration wizard?

[SwampRabbit]

I recently have been testing with 9.3 and I have a question about the initial configuration wizard.

While following the wizard, I completed theVolume Creation Wizard, named the pool and selected "Automatic - Pick reasonable defaults for available drives".

Completed the rest of the wizard, not setting any shares, or anything special like that.

I just wanted to see what the "Automatic" option would determine is the best setup for my drive layout.

It worked perfectly, set up the pool with 2 vdevs of 6 drives each.

Where I am confused is that it also created a dataset in each vdev, the datasets are named the same as the vdevs.

My questions are: why are they created/what are these for?
Are they created for use as the system dataset?

I attempted to delete them, just to see if I could. But couldn't.

I do not remember any datasets being automatically created when I created a pool before using the Volume Manager in 9.2.1.8. I checked the FreeNAS User Guide for 9.3 and it doesn't mention these datasets at all.

Thank you in advance

 

[danb35]

I'm strongly suspecting that you're misunderstanding what you're seeing, as FreeNAS doesn't expose individual vdevs in most of the web GUI. I'd bet that what you're seeing is the root dataset of the pool, which is normal and can't be deleted (every zpool contains a dataset of the same name as the pool; any files or other datasets are created within that dataset). That dataset was hidden in the web GUI in 9.2.x and prior, though it still existed. Can you post a screen shot of what you're seeing? Edit: The output of 'zpool status' in code tags would also be helpful.

 

[SwampRabbit]

I could be misunderstanding what I am seeing, I am not at home to check right now, but will later this evening.

On the screen it was:

Vol1
>Vol1

Vol2
>Vol2

I believe that when I clicked it and the option icon buttons at the bottom of the screen changed to the dataset ones (i.e. delete dataset). It was 2 days ago and I just wasn't able to post until today.

Also I am running the latest 9.3 stable, which I think was posted on the 9th?

 

[SwampRabbit]

Ok after going back to the manual I found in section 8.1.8:

If you click Storage ‣ Volumes ‣ View Volumes, you can view and further configure existing ZFS pools, datasets, and zvols. The example shown in Figure 8.1k demonstrates one ZFS pool (volume1) with two datasets (the one automatically created with the pool, volume1, and dataset1) and one zvol (zvol1).

So what I am seeing must be the dataset that is created automatically with the pool, the "root dataset of the pool" as you explained it.

Like you, said this dataset was hidden in the web GUI in 9.2.x and prior, so that adds to my confusion.

So my questions are now... what was the reason for showing it now?
Should we be doing something that was not done in 9.2.x and prior because this dataset is shown?

My assumption is to not mess with this dataset so as to not mess up the pool some how on accident.

 

[danb35]

I don't know why the pool dataset was hidden in 9.2.x and earlier, nor why iX decided to stop hiding it now. Also note that you don't have two vdevs, you have two pools. This may or may not have been what you intended.

 

[cyberjock]

It wasn't hidden. It was mixed in with the volume listing. Too many people were confused as to how to read the info (it was all there if you understood what you were looking at). This setup, hopefully, makes more sense to people. :)

 

[SwampRabbit]

@danb35 - I did not intend to have two pools, but I did select "Automatic" at the Volume Creation Wizard portion, so this must be what the wizard thought was best for 12 drives in my situation.

@cyberjock - understood

For some reason I think it did not show this dataset after using Volume Manager, just the wizard.

When I get home I will do it through the Volume Creation Wizard and the regular Volume Manager and see if it comes out the same again.

I have only messed with 9.2.x and 9.3, so my experience is limited, but I noticed this was different... so just wanted to understand it better. :)

 

[SwampRabbit]

Ok, I removed the pool I had and ran the wizard again, named the pool "bob" and created a CIFS share this time too. Attached is what shows up under "Volume Status".
Using the Volume Manager I replicated the same thing, selecting 6 drives across, and two rows down. Which gave me that same thing.

This is two 6 drive RAIDZ2 vdevs correct?

A dataset named "bob" popped up under the volume "bob".
So this is the root dataset of the pool like danb35 said, which I perfectly understand now.

I really like the new manual for 9.3, but maybe a suggestion for section 8.1.8 to make it a little more clear for silly people like me who don't know about the root pool dataset.

The example shown in Figure 8.1k demonstrates one ZFS pool (volume1) with two datasets (the one automatically created with the pool, volume1, and dataset1) and one zvol (zvol1).

changed to something like:
The example shown in Figure 8.1k demonstrates one ZFS pool (volume1) with two datasets: volume1 (the root dataset automatically created with the pool), dataset1, and one zvol (zvol1).

Thank you both for helping me understand.

 

[danb35]

What you have now is a single zpool consisting of two RAIDZ2 vdevs. Make sure you know which disk numbers (adaN) map to which physical disks, and which of those maps to which vdev--it doesn't look like the the auto-configuration used contiguous blocks of disks for the vdevs (i.e., ada0-ada5 for one, and ada6-ada11 for the other). This isn't a problem at all for the system, but might cause some confusion when it comes to replacing disks.

 

[SwampRabbit]

@danb35 I had noticed that and wasn't sure if it was an issue or not. I have two IBM M1115s in the system at the moment, I assumed that the wizard chose this configuration based on some reason.

And thanks again for all your explanations and observations, really appreciate it.

But you are very right, knowing which disks are where in terms of physical layout is important!
They are in drive sleds for quick removal and I already have a serial number, disk number, and slot position map. I know from past experience how important something like this can be.

I have just completed a fresh install on new 16GB flash drive and am about to mirror it, so no pool is set up at the moment.
I am pondering if I want to configure it through Volume Manager myself and put the disks in order of how they are in the chassis to simplify things.

A question would be, does the wizard chose the disks based on a specific reason to increase performance or redundancy?