Weird behaviour of mkfs.xfs

[Christoph Hellwig]

On Sat, Mar 26, 2016 at 03:46:19PM +0100, Emmanuel Florac wrote:
> Actually I was too impatient; it finally ended avec 30 minutes of
> burning bits to the flash. I don't understand the behaviour, though.
> I'm used to mkfs.xfs making its magic extremely quickly, even on
> humongous devices. Here it's a very fast array of only 3.2 TB...

Trey doing a mkfs.xfs -K, without that it diascards the whole device.
I've seen some NVMe device misbehaving under discard storms, up to
the point of resetting the controller..