2010-10-12 11:47 PM
We are running ontap 7.3.3 and our vmware environment has a number of NFS datastores.
these volumes are no guarantee, 20 odd % snapshot reserve with dedupe enabled.
Here is the strange part - the space breakdown shows xGB virtual disks, xGB swap files, xGB snapshots, xGB other VM files, xGB other, xGB free space -- the weird thing is the xGB of other, what is other? I have browsed the datastore and there is nothing relating to this used space, which infact consumes a large amount of the datastore itself.
I have at one time seen this disappear completely within 1-5 minutes, this is a lot of data to be purged at in a small space of time, around 300GB or more.
The only thing I can see that occurred at that time was that I migrated a powered off vm to another datastore to free up some space on the datastore to allow other vms to continue operating.
I have attached a screenshot of this quirk to this thread.
If anyone can explain this it would be much appreciated.
2010-10-13 10:49 AM
"The only thing I can see that occurred at that time was that I migrated a powered off vm to another datastore to free up some space on the datastore to allow other vms to continue operating."
That pretty much explains it. Most filesystems, VMware datastores included, will tag deleted blocks. At some point, not in real time, a garbage collection or "segment cleansing" process comes along and actually clears the space. 300GB in 1-5 minutes? How long does it take to validate and update pointers in a table? How big was the VM you moved? I think you'll find your answer there.