Setup is two Windows Server 2016 VMs atop of kvm. One machine Primary doing I/O (with sleep 10 to relax load a little) Secondary just being connected. After one week Secondary becomes unresponsive and also partially cannot start new processes. Later on the machine has to be reset.
This should be fixed before tagging 1.0.0
Conjecture is that splitting I/O on the local backing device into 4K pieces (and opening a I/O request for each of them) is suboptimal and causes Windows eventually to get unresponsive.
Setup is two Windows Server 2016 VMs atop of kvm. One machine Primary doing I/O (with sleep 10 to relax load a little) Secondary just being connected. After one week Secondary becomes unresponsive and also partially cannot start new processes. Later on the machine has to be reset.
This should be fixed before tagging 1.0.0
Conjecture is that splitting I/O on the local backing device into 4K pieces (and opening a I/O request for each of them) is suboptimal and causes Windows eventually to get unresponsive.