Guo Lian Yun wrote:
# HG changeset patch
# User Guolian Yun <yunguol(a)cn.ibm.com>
# Date 1210745379 25200
# Node ID 6a8318103e42f7e1891011550d3a9f7d7f86839c
# Parent 935b2c485f8857ddb396efd729d5fc7ff7b41614
[TEST] add LXCXML class to support LXC guest
Signed-off-by: Guolian Yun <yunguol(a)cn.ibm.com>
diff -r 935b2c485f88 -r 6a8318103e42 suites/libvirt-cim/lib/XenKvmLib/const.py
--- a/suites/libvirt-cim/lib/XenKvmLib/const.py Tue May 13 22:57:46 2008 -0700
+++ b/suites/libvirt-cim/lib/XenKvmLib/const.py Tue May 13 23:09:39 2008 -0700
@@ -78,3 +78,9 @@ XenFV_default_disk_dev = 'hda'
XenFV_default_disk_dev = 'hda'
XenFV_default_mac = '00:16:3e:5d:c7:9e'
XenFV_default_net_type = 'bridge'
+
+#vxml.LXCXML
+LXC_init_path = os.path.join(_image_dir)
Containers don't boot from an image like Xen and KVM do. You can think
of containers as a way of encapsulating a process or a group of
processes. So the init_path for a container needs to be a process or a
group of processes you can run inside the container.
Something to try would be to create a script and run something like exec
/bin/bash in the script.
+LXC_default_file_src = '/tmp'
+LXC_default_file_tar = '/tmp'
All this does is mount the /tmp dir as /tmp within the container, which
is redundant. I can't think of a good reason to use the filesystem tag
for a test guest. Thoughts?
+LXC_default_tty = '/tmp'
/tmp isn't a valid value for the console. For this, you'd want to
specify /dev/ptmx, which creates a pseudo terminal to use for the console.
--
Kaitlin Rupert
IBM Linux Technology Center
kaitlin(a)linux.vnet.ibm.com