We’re (still) on 3.9.2-57.
I have just noticed that all of my snapshots show zero bytes. They’re getting created per daily per schedule but they show empty.
Why? How to fix?
@MRC-MBU Hello again.
This may well just be down to you running with quotas disabled. But we do also still have some work to do in our space accounting; especially in snapshots and clones. Plus there is the possibility that if the snapshots target subvolume have underdone no change then they can actually take up very little space as they are simply a very small amount of metadata pointing to the very same data chunks as the last snapshot.
If it’s down to no data change in associated share (btrfs subvolume) between snapshots then no fix, but more likely it’s down to quotas disabled. And believe you disabled quotas for either performance or to a bug in that area. We don’t enforce quota limits yet anyway.
A command of interest may be ‘btrfs fi usage’; for further investigation on this. Again a newer underlying btrfs in the form of our v4 offering may help here but not so much on our size calculations actually as we have done little work recently on that front. Apologies for that. It is on the cards but we must tackle some more pressing developmental related issues first unfortunately.
Hope that helps.
2 Likes
Ok, I can see now that the GUI can lie to you. The snapshots are not zero bytes.
When I log in on the CLI I can see and access the snapshots and they have data in them. I was able to recover what I needed in this way.
1 Like
@MRC-MBU Glad you got this sorted. Yes we definitely have some work to do on refining our snapshot accounting. This work will hopefully be part of the next testing channel. But it also depends on the quotas and that in turn can have unwanted performance impacts. Bit by bit.
Cheers.