Package: libvirt0 Version: 0.8.1-2 Severity: normal
Setup: libvirt with qemu, network managed by libvirt. Problem: When starting libvirt VMs when no working DNS is available, virsh is unable to connect to libvirt and simply hangs without saying anything. Adding the IPs configured for the VMs and corresponding (fake) hostnames to /etc/hosts and then restating libvirt fixes this issue. It looks like this is because the iptables calls made by libvirt to set up VM routing do not use the -n option. So, if all DNS servers for one (or more) VM IPs are unreachable, iptables waits a long time until reaching a timeout, and in this time libvirt is completely unusable. I propose to use "iptables -n" every time iptables is called by libvirt. It should not have any negative effects and would save a lot of trouble and debugging in case of DNS fuckups. --derf -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
