FreeNas Not Using All Available Space

[Robert Cuthill]

Hi Guys,
I feel as though I'm missing something simple. I have 2x3tb drives in a FreeNas 9.2 box. The size of the drive is recognized as 2.7tb GPT, but the iSCSI connection in Windows only see's it as 2tb. What things should I check to find out what's wrong? Thanks everyone.

 

[jgreco]

Perhaps you only configured a 2TB virtual disk?

In any case, you really can't allocate much more space without fragmentation becoming a major issue. A ZFS pool used for VM storage will become increasingly fragmented with time, and 60% is about as full as it should be made unless you are particularly masochistic and like slow---block---allocation.

 

[Robert Cuthill]

I've included pics. doesn't the 770gb available mean that the volume was configured for the entire disk?

This drive is connected to a Windows machine for backup storage. Is there a way to avoid fragmentation?

 

[jgreco]

No, fragmentation is an inherent side effect of a copy on write filesystem...

 

[Robert Cuthill]

I see. I'd still like to get to the bottom of why I can't use this available space.

 

[jgreco]

Perhaps you only configured a 2TB virtual disk?

 

[Robert Cuthill]

How can I check or verify if this is the case?

 

[jgreco]

... maybe go into the iSCSI configuration area and take a look? Sharing->Block->Extents->${yourextent}->Device and see how big it thinks it is.

 

[Robert Cuthill]

... maybe go into the iSCSI configuration area and take a look? Sharing->Block->Extents->${yourextent}->Device and see how big it thinks it is.

I'm on 9.2. Under iSCSI>view extends, extend size is set to '0' for auto.

 

[jgreco]

Try actually doing what I said, click Sharing, then Block, then Extents, not "View Extents", and then look in the Device box. It'll say something like "poolname/zvolname (size GiB)".

 

[Robert Cuthill]

Under sharing, there is no 'block' category. I can only assume the gui is a bit different in 9.2.

I appreciate your help, can you please cut down on the attitude?

 

[jgreco]

What specific version of 9.2?

 

[Robert Cuthill]

9.2.1.7

 

[jgreco]

Under sharing, there is no 'block' category.

Crap. I thought that was introduced a long time ago.

I appreciate your help, can you please cut down on the attitude?

Countersuggestion: don't mistake terse for attitude. I actually gave you a recipe that had been checked to do what was needed, because the information is NOT available on the "View Extents" screen, as nonsensical as that may seem. Then I asked you to actually do what I originally asked.

I don't have any 9.2.1.7 laying around. I'll spin up a VM and take a look but it may take a bit.

 

[jgreco]

Kay, so yes, I screwed up, back in 9.2.1.7 it's still sitting over in Services -> iSCSI -> Extents -> ${yourext} -> Device ... but not under View Extents.

Wait. Are you using a file backed extent? Because that's the only time I see a reference to 0 for size being allowed. In which case, the answer becomes, "please look at the size of the file you're using for backing store," I think. If you do that, then there's also the possibility that there's maybe some 32-bit int issue in there which is limiting the size to 2TB, and also the inevitability that there's no fix for this because zvols are the way you're expected to implement iSCSI.

And at this point I'll also reiterate that the fragmentation issue is a significant limiting factor in your options here. At 2TB, you've already created a device that's larger than recommended for your pool size.

However, since you're using 9.2.1.*, I'll note that there is a possibility that you COULD have an option to use your two 3TB drives as iSCSI targets; you'd have no redundancy and it wouldn't be using ZFS, and there'd be no possibility of future FreeNAS upgrades, but you'd have 6TB of space.

 

[Robert Cuthill]

I upgraded to 9.3 and now increasing the extend size is immediately reflected in Windows. Looks like it was a bug. Thanks for your help, jgreco. I'm going to take your advice and avoid extending the size anyway. At least we got to the bottom of the issue.

 

[jgreco]

Strange, but at least a resolution.