[Swan-dev] disk usage on test machine

[Paul Wouters]

On Sun, 31 Aug 2014, D. Hugh Redelmeier wrote:

> So "results" is serious (larger than libreswan itself by a factor of 4).

It has lots of results? So not entirely strange.

> But pool seems to be much large (21G):
> 828744	nic.qcow2
> 1261000	north.qcow2
> 2206732	swanfedorabase.qcow2
> 2303472	swanfedorabase.img
> 2542280	road.qcow2
> 5819596	east.qcow2
> 5946572	west.qcow2
>
> Is there a simple way to redo the qcow2 files?  I think that every write
> in the VM to these filesystems adds a block and old ones are not recycled.
> "qemu-img info" doesn't seem to tell me stats about this.

See testing/libvirt/install.sh

basically:

for host in nic north road east west
do
 	rm $POOLSPACE/$host.qcow2
 	qemu-img create -F qcow2 -f qcow2 -b $POOLSPACE/swan"$OSTYPE"base.qcow2 $POOLSPACE/$hostname.qcow2
done

although you'll end with the old swanbase you created a long time ago,
so runnig yum update will immediately deviate from those quickly. So
perhaps a better method is to just delete the entire $POOLSPACE and
run testing/libvirt/install.sh so it will create an updated
swanfedorabase.img before cloning it into 5 VMs.

Paul