I have two nodes for EQL 6210XS and there are bind a name "EQL Pool". Total capacity almost 36TB.
Now when i create a volume 8TB for the Hyper-v cluster shared volume and create some VM in CSV .All Hyper-V HOST connection to the EQL Volume via ISCSI MPIO.
In the hyper-v failover cluster manager show our CSV have 500GB free space , but the EQL Group Manager show have 5TB free space.
It seems that Hyper-V HOST cann't correct identification the volume free space after deduplication.
Is there any suggestion to exclude this problem?
This is actually not a problem at all. The short version is that block storage isn't the best place to monitor free space. Only in recent times is storage notified about file deletions. And that's not supported by every OS or configuration. This is known as UNMAP AKA Space Reclaim With EQL if you replicate a volume then UNMAP is no longer supported.
When you use a HyperVisor that brings out another interesting affect as you noticed. Storage showing LESS in-use space vs. OS. When you care a virtual disk, say of 50GB, the hypervisor doesn't actually write 50GB to storage. In fact at creation time it's a tiny fraction. Only as files are written will the in-use space grow. Since that's all that block storage can record. writes. The OS though deducts the full 50GB at creation time from the allocation table it maintains. That table isn't shared with the storage device.
They two will never fully line up. Even with UNMAP in place. The OS knows what blocks can be re-used so even when the array reports 100% in use, go by what the OS says. It is the only authority on the subject.
Here's the long explanation in case you had more questions.
With Hyper-V 2012 UNMAP on CSVs is only supported with VHDX virtual disks, not the older VHD.
Social Media and Community Professional#IWork4DellGet Support on Twitter - @dellcarespro