[Qemu-discuss] vm get stuck during snapshot commit with qemu-img

[Hans-Juergen Rux]

Morning,


We have the following problem and we have absolut no idea how to fix it.



We are using CentOS 7.2 / 7.3 with KVM (qemu 2.6). Everything works really 
great with VM image files (qcow2) under about ~700/800 GB. It is fast and works 
fine, interruptions of the VM during snapshot commit into base file of max 1 
second.



But when we build virtual servers with disk images >1-2 TB or more, and commit 
snapshots from a snapshot chain into the basefile, we have interruptions in the 
virtual machines of 30-60 seconds. Ping to the VMs is interrupted, the RDP 
sessions are lost. The time inside VM jumps.



It seems that the timeout happens at the end of the commit proccess.



The same on different HW, diferrent disk subsystems, different Windows or Linux 
VMs. So regardless of whether the drive on VM operating system has a paritition 
on ist virtual disks or the qemu guest-agent is installed.



Any ideas?



Thanks



Hans-Juergen