Inventory report

[kesavkolla]

When we have millions of objects listobject takes lots of time. Is it possible to generate inventory report asynchronosuly? Once inventory report is there the listobject should leverage that to make listing faster.

[guenther-alka]

I cannot say how much it would help as I do not know RustFS internals enough but if your object storage is hd based, you may improve io especially for metadata at storage level below s3.

This is the concept I prefer with my napp-it cluster management tool where I use RustFS in single node, single disk mode ontop ZFS. This means that ZFS does the disk redundancy and bitrot protection and can add compress, realtime dedup and rambased Arc caching. For a use case with millions of objects a ZFS hybrid pool with special vdev may be a key element to place metadata on NVMe while data remains on hd.