Re: [libvirt] PATCH: Bash auto-completion for virsh

2008-07-15 Thread David Lutterkort
On Mon, 2008-07-14 at 11:57 +0100, Daniel P. Berrange wrote: > Everyone[1] seems to be doing funky bash auto-completion for commands these > days, so I thought I'd make a stab at doing something for virsh. I guess the completions I did previously[1] never were committed then - I'm using them quite

Re: [libvirt] auto shutdown for qemu / kvm

2008-07-15 Thread Matthias Pfafferodt
Hello Daniel, Am Dienstag, 15. Juli 2008 21:17:04 schrieb Daniel P. Berrange: > On Sat, Jul 12, 2008 at 10:25:01PM +0200, Matthias Pfafferodt wrote: > > Hello, > > > > there is the possibility to autostart a (qemu/kvm) domain. If libvirtd is > > stopped this domain is killed and the filesystem is

Re: [libvirt] static address assignment via dnsmasq

2008-07-15 Thread Daniel P. Berrange
On Mon, Jul 14, 2008 at 10:49:39AM -0700, Kurt Sussman wrote: > I want my VMs to be assigned static addresses by hostname. I would > prefer to set up a generic way to configure dnsmasq, since I don't think > this is the last thing I'll want to change about address assignment. > > I saw this patch

Re: [libvirt] auto shutdown for qemu / kvm

2008-07-15 Thread Daniel P. Berrange
On Sat, Jul 12, 2008 at 10:25:01PM +0200, Matthias Pfafferodt wrote: > Hello, > > there is the possibility to autostart a (qemu/kvm) domain. If libvirtd is > stopped this domain is killed and the filesystem is in a unclean state. I > miss the possible of an 'autostop' option that will shutdown a

Re: [libvirt] PATCH: 5/5: Preserve live XML config across restarts

2008-07-15 Thread Stefan de Konink
Daniel P. Berrange schreef: I'm not entirely happy with this approach though - I'd ideally like to avoid having communications between the libvirtd & the controller process if practical, Yesterday I was reading into the Xen approach that is taken on this, with respect to defining and undefinin

Re: [libvirt] PATCH: 5/5: Preserve live XML config across restarts

2008-07-15 Thread Daniel P. Berrange
Although the LXC driver can cope with libvirtd restarts, there is one small problem relating to the XML format. The XML format as stored in /etc/libvirtd/lxc/NAME.xml is the canonical 'inactive' configuration data. There are a few pieces of data which are filled in at runtime, such as the current /

Re: [libvirt] PATCH: 3/5: Introduce a container controller process

2008-07-15 Thread Daniel P. Berrange
Every LXC container has 2 processes, the leader of the actual container and a helper process to do the I/O forwarding. The LXC driver is written to allow libvirtd to be restarted without needing to shutdown the active containers. For this it uses a PID file in /var/run/libvirt/lxc/NAME.pid The driv

Re: [libvirt] PATCH: 2/5: Re-arrange methods acros LXC source files

2008-07-15 Thread Daniel P. Berrange
The lxc_driver.c file contains quite a large amount of code, serving two reasonably well separated purposes. First there is the direct implemntation of each of the libvirt driver APIs. Second there is the code to spawn a container and a controller for forwarding I/O to/from the PTYs. This patch att

Re: [libvirt] PATCH: 1/5: Removing state from lxc_vm_t

2008-07-15 Thread Daniel P. Berrange
This is same as the patch I sent yesterday... This patch does some simple re-factoring of the way the TTYs and control socket are handled to reduce the amount of state stored in the lxc_vm_t structure, in preparation for the switchover to the generic domain handling APIs. lxc_conf.c |1

[libvirt] PATCH: 0/5: Refactoring LXC & switching to generic domain APIs

2008-07-15 Thread Daniel P. Berrange
The following series of 5 patches perform some general re-factoring of the LXC driver, the relationship between libvirtd & the processes it spawns, and finally switches LXC over to the generic domain XML APIs. This is just basically 1 day's work, so its a little rough around the edges in places, bu