Why's the dataset bigger than the volume?

Status
Not open for further replies.

willnx

Dabbler
Joined
Aug 11, 2013
Messages
49
I never really noticed this before, but isn't this a bit odd?

FOPCZhB.png


vol1 is made up of 6x4TB HDDs in a raidz2, has 10GB reserved space configured, and dedupe disabled.

The only thing I can think of as to why a dataset below the volume would report a "size" that's bigger than that of the volume is because it's not showing hardware size?
But more so, a variable size that takes compression into consideration?
I'm not really convinced with that idea; I'd image with 1.36x compression that a variable size would be greater than 1.4% bigger...​

Or maybe a "windows share type" doesn't have as much overhead?
This also doesn't seem unlikely for many reasons...​

What are your thoughts/ ideas?
 

cyberjock

Inactive Account
Joined
Mar 25, 2012
Messages
19,525
If I'm not mistaken the size of a dataset is a function of what your reservations are (on all volumes and datasets, not just the one you are interested in), your actual pool used, actual pool free, as well as used and free space on the datasets.

Not cut and dry, but you'll probably recognize that available + used = size. So that's most likely where the 14.3TB is coming from.
 

willnx

Dabbler
Joined
Aug 11, 2013
Messages
49
Ok, so this seems like a bug with the number - somehow Available isn't being calculated properly. I'll poke around a bit with this, and check to see if there's already something on file about it (note: two things to do, not just one and assume it's a bug)

If I'm not mistaken the size of a dataset is a function of what your reservations are (on all volumes and datasets, not just the one you are interested in)

This generated another question in my brain, to which my google/search-foo was unable to find an answer.

Does a dataset respect a reserve space set at the volume level? If so, what's the use-case of setting a reserve space on the dataset level?​
 

cyberjock

Inactive Account
Joined
Mar 25, 2012
Messages
19,525
Technically it's not a bug. It technically has to do with how free space is calculated for pools and datasets. Datasets are actually their own filesystem, so the boundaries make things look non-obvious. There are plans in the works to have this make a little more logical sense to the common reader, but you will still have to understand where the boundaries are and how it all works. ;)

To make things worse things like quotas, snapshots and reservations do not show up on that particular list, but they are there and they do take up space.

In essence, that gives you only the broadest info, and only if you understand it all. If you don't then it looks totally jacked up. There's been tons of threads on this topic and a bug ticket already exists to make this a little more "real" for casual readers.
 

willnx

Dabbler
Joined
Aug 11, 2013
Messages
49
I get what you're saying, and thanks for taking time to respond.
 
Status
Not open for further replies.
Top