It appears that the virtualisation host reboots the guest because it fails to grow the QCOW2 volume, so rather a QEMU issue than a cobbler issue:

from tail -f /var/log/libvirt/qemu/mycobblerserver.log, during the critical event:

qemu-kvm: /builddir/build/BUILD/qemu-kvm-0.12.1.2/block/qcow2.c:694: qcow2_co_writev: Assertion `(cluster_offset & 511) == 0' failed.
2015-03-25 10:54:14.452+0000: shutting down
2015-03-25 10:54:23.506+0000: starting up

From: "Joost Ringoot" <joost.ringoot@meteo.be>
To: "cobbler mailing list" <cobbler@lists.fedorahosted.org>
Sent: Wednesday, 25 March, 2015 11:44:45
Subject: Re: [cobbler] server crashes when trying reposync of centos6_x64_updates

Meanwhile I found that there is no crashdump, because the host is a QEMU virtualized machine, the virtualisation engine takes over this memory dump, qemu did dump cores however.