Quoting Daniel P. Berrange (berrange@redhat.com):
On Mon, May 16, 2011 at 09:47:02AM -0500, Serge Hallyn wrote:
This allows upstart scripts to detect that they are in a container and modify their behavior accordingly.
In this way, the same ubuntu maverick or natty image with the 'lxcguest' package installed can be booted on bare metal, in kvm, or as a libvirt or liblxc container.
Signed-off-by: Chuck Short <zulcss@ubuntu.com> Signed-off-by: Serge Hallyn <serge.hallyn@ubuntu.com> --- src/lxc/lxc_container.c | 1 + 1 files changed, 1 insertions(+), 0 deletions(-)
diff --git a/src/lxc/lxc_container.c b/src/lxc/lxc_container.c index 9ae93b5..a70aeeb 100644 --- a/src/lxc/lxc_container.c +++ b/src/lxc/lxc_container.c @@ -112,6 +112,7 @@ static virCommandPtr lxcContainerBuildInitCmd(virDomainDefPtr vmDef)
virCommandAddEnvString(cmd, "PATH=/bin:/sbin"); virCommandAddEnvString(cmd, "TERM=linux"); + virCommandAddEnvString(cmd, "container=libvirt"); virCommandAddEnvPair(cmd, "LIBVIRT_LXC_UUID", uuidstr); virCommandAddEnvPair(cmd, "LIBVIRT_LXC_NAME", vmDef->name);
I've no objection to this, but IMHO the value should be
container=libvirt-lxc
So long as it is guaranteed that henceforth LIBVIRT_LXC_UUID=x will always be passed to init, that suffices. But I worry about it changing in 6 or 12 months to a slightly different name, or not always being passed. Do you think there is any chance of that happening? thanks, -serge