Hello,

I'm using libvirt to deploy a series of 7 KVM (in qcow2 format) sequentially. The base image of the qcow2 is an ubuntu server of around 1.6GB.

The environment where i am doing this is a Live USB Ubuntu with a persistence file (so that changes made remain).


So, the problem:

* If the persistence file (i.e. free disk space in the live ubuntu) is up to around 1.6GB, the qemu process of launching the first one of the KVM gets stuck, and a few seconds later an Ubuntu message pops up saying there is no disk space left.

* If the persistence file is a bit bigger (100 MB more), it can launch the 7 KVM machines without any disk space problems.


This would suggest that during the deployment of one virtual machine, there is a short period of time where a big amount of disk space (similar to the size of the base image) is used (even if i am using qcow2 vms) and afterwards released, so it can be used to launch the second vm, released, used for the third, and so on.

If this is correct, i would like to know who is using that space, since my intention would be to, if possible, redirect this temporary disk usage to a RAM filesystem such as tmpfs, so that i don't need such a huge persistence file in the USB.


I'm aware that this might be related to qemu-kvm rather than to libvirt itself, but i can't find information about it and i would appreciate if anyone could point me in the right direction. More details or clarification can be provided if needed.

Best regards,

Jorge