since we work with python2 only, mentioning it in all shebangs make
the commands from git work even when python3 is set as default.
This also fixes one test where command being ran is 'virt-xml' through
subprocess.Popen().
While at that, add '-tt' where possible in order to make everyone use
same indentation characters.
Signed-off-by: Martin Kletzander <mkletzan@redhat.com>
Add "-c" as short option of "--connect" for virt-xml.
We could use either
virt-xml -c lxc:///
or
virt-xml --connect lxc:///
This will be more convenient if we operate
non-default hypervisor.
Signed-off-by: Chen Hanxiao <chenhanxiao@cn.fujitsu.com>
Add a new field "pool-source-name" in the createpool UI that is used
only with gluster fs pools.
Signed-off-by: Giuseppe Scrivano <gscrivan@redhat.com>
When events were successfully registered, we skip the VM listing on
every tick, and instead trigger a manual refresh whenever a VM event
is received. Not as efficient as it should be, but saves us a lot of
API calls.
This path hits the qemu monitor, and leads to the virt-manager UI being
locked up when performing asynchronous tasks like creating a snapshot.
So make it opt in for people that want it.
This patch will enable configuring idmap.
It could be used as enable user namespace
for LXC containers.
Signed-off-by: Chen Hanxiao <chenhanxiao@cn.fujitsu.com>
This means if we are passed an unmanaged path, we try to create a
storage pool for the parent directory.
We skip directories like /dev where doing this might be problematic.
This makes things much friendlier to use for remote connections, and
means we can always rely on having libvirt's storage APIs to use
for format probing.
We totally break CLI compat here, but the previous tool wasn't sustainable.
Instead, repurpose the tool as strictly converting external formats
like ovf/vmx to native libvirt XML, and launch the guest.
So we drop vmx/virt-image output, and virt-image input, and a slew of
command line options. I don't think anyone was depending on this in a
scripted fashion, so in practice I don't think anyone will care.
Add much more comprehensive unit tests while we are at it.
Martin Kletzander