[Lxc-users] [PATCH] multiple mods to lxc-clone
Serge Hallyn
serge.hallyn at canonical.com
Mon Aug 29 22:08:47 UTC 2011
Thanks, Ramez. It looks good to me. My only comment would be that
if the rootfs copy fails (either rsync or lvm clone), and you've
frozen the original container, then you need to unfreeze the original
container before erroring out.
-serge
Quoting Ramez Hanna (rhanna at informatiq.org):
> * allow cloning of non-snapshot lvm devices
> * if no -s then create a copy of the lvm block device and copy data
> from the orig to the new container device
> * first take a snapshot, then use this snapshot to copy data,
> remove snapshot after done
> * if orig container is running freeze it while copying
> * in case lvm block device, the container is only frozen during
> creation of snapshot ~1 sec
> * use rsync -ax insted of cp -a
> * in case copying a live contrainer it won't copy runtine mounted
> files such as /proc, /sys and some /dev
> * new opts
> * fstype: type of fs for the newly created lvm device in case of
> non-snapshot lvm
> * lvprefix: prefix for new lvm device name.
> * do not delete the lines lxc.mount by default
> * check is fstab exists then copy it
> * only modify lines that contain "lxc.mount =", debian template
> seems to not have that line but uses lxc.mount. lines which get
> screwed
>
>
> Signed-off-by: InformatiQ <rhanna at informatiq.org>
> ---
> src/lxc/lxc-clone.in | 98 ++++++++++++++++++++++++++++++++++++++------------
> 1 files changed, 75 insertions(+), 23 deletions(-)
> mode change 100644 => 100755 src/lxc/lxc-clone.in
>
> diff --git a/src/lxc/lxc-clone.in b/src/lxc/lxc-clone.in
> old mode 100644
> new mode 100755
> index 91944a0..d42160b
> --- a/src/lxc/lxc-clone.in
> +++ b/src/lxc/lxc-clone.in
> @@ -1,4 +1,4 @@
> -#!/bin/bash
> +#!/bin/bash
>
> #
> # lxc: linux Container library
> @@ -22,7 +22,7 @@
> # Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
>
> usage() {
> - echo "usage: lxc-clone -o <orig> -n <new> [-s] [-h] [-L fssize]
> [-v vgname]"
> + echo "usage: lxc-clone -o <orig> -n <new> [-s] [-h] [-L fssize]
> [-v vgname] [-p lxc_lv_prefix] [-t fstype]"
> }
>
> help() {
> @@ -36,15 +36,19 @@ help() {
> echo "-s : make the new rootfs a snapshot of the original"
> echo "fssize : size if creating a new fs. By default, 2G"
> echo "vgname : lvm volume group name, lxc by default"
> + echo "lvprefix" : lvm volume name prefix, none by default, e.g.
> --lvprefix=lxc_ then new lxc lv name will be lxc_newname"
> + echo "fstype" : new container file system type, ext3 by
> default (only works for non-snapshot lvm)"
> }
>
> -shortoptions='ho:n:sL:v:'
> -longoptions='help,orig:,name:,snapshot,fssize,vgname'
> +shortoptions='ho:n:sL:v:p:t:'
> +longoptions='help,orig:,name:,snapshot,fssize:,vgname:,lvprefix:,fstype:'
> lxc_path=/var/lib/lxc
> bindir=/usr/bin
> snapshot=no
> lxc_size=2G
> lxc_vg=lxc
> +lxc_lv_prefix=""
> +fstype=ext3
>
> getopt=$(getopt -o $shortoptions --longoptions $longoptions -- "$@")
> if [ $? != 0 ]; then
> @@ -63,6 +67,7 @@ while true; do
> -s|--snapshot)
> shift
> snapshot=yes
> + snapshot_opt="-s"
> ;;
> -o|--orig)
> shift
> @@ -84,6 +89,11 @@ while true; do
> lxc_new=$1
> shift
> ;;
> + -p|--lvprefix)
> + shift
> + lxc_lv_prefix=$1
> + shift
> + ;;
> --)
> shift
> break;;
> @@ -141,50 +151,92 @@ trap "${bindir}/lxc-destroy -n $lxc_new; echo
> aborted; exit 1" SIGHUP SIGINT SIG
>
> mkdir -p $lxc_path/$lxc_new
>
> +hostname=$lxc_new
> +
> echo "Tweaking configuration"
> cp $lxc_path/$lxc_orig/config $lxc_path/$lxc_new/config
> sed -i '/lxc.utsname/d' $lxc_path/$lxc_new/config
> echo "lxc.utsname = $hostname" >> $lxc_path/$lxc_new/config
>
> -sed -i '/lxc.mount/d' $lxc_path/$lxc_new/config
> -echo "lxc.mount = $lxc_path/$lxc_new/fstab" >> $lxc_path/$lxc_new/config
> +grep "lxc.mount =" $lxc_path/$lxc_new/config >/dev/null 2>&1 && { sed
> -i '/lxc.mount =/d' $lxc_path/$lxc_new/config; echo "lxc.mount =
> $lxc_path/$lxc_new/fstab" >> $lxc_path/$lxc_new/config; }
>
> -cp $lxc_path/$lxc_orig/fstab $lxc_path/$lxc_new/fstab
> -sed -i "s@$lxc_path/$lxc_orig@$lxc_path/$lxc_new@" $lxc_path/$lxc_new/fstab
> +if [ -e $lxc_path/$lxc_orig/fstab ];then
> + cp $lxc_path/$lxc_orig/fstab $lxc_path/$lxc_new/fstab
> + sed -i "s@$lxc_path/$lxc_orig@$lxc_path/$lxc_new@" $lxc_path/$lxc_new/fstab
> +fi
>
> echo "Copying rootfs..."
> rootfs=$lxc_path/$lxc_new/rootfs
> # First figure out if the old is a device. For now we only support
> # lvm devices.
> mounted=0
> +#is container running
> +lxc-info -s -n $lxc_orig|grep RUNNING >/dev/null 2>&1
> +if [ $? -ne 0 ]; then
> + container_running=True
> +fi
> sed -i '/lxc.rootfs/d' $lxc_path/$lxc_new/config
> oldroot=`grep lxc.rootfs $lxc_path/$lxc_orig/config | awk -F= '{ print $2 '}`
> if [ -b $oldroot ]; then
> # this is a device. If we don't want to snapshot, then mkfs, mount
> # and rsync. Trivial but not yet implemented
> - if [ $snapshot == "no" ]; then
> - echo "non-snapshot and non-lvm clone of block device not yet implemented"
> - exit 1
> - fi
> + #if [ $snapshot == "no" ]; then
> + # echo "non-snapshot and non-lvm clone of block device not yet implemented"
> + # exit 1
> + #fi
> lvdisplay $oldroot > /dev/null 2>&1
> if [ $? -ne 0 ]; then
> - echo "non-snapshot and non-lvm clone of block device not yet implemented"
> - exit 1
> - fi
> + lvm=False
> + echo "non-lvm block device cloning not yet implemented"
> + exit 1
> + else
> + lvm=TRUE
> + fi
> # ok, create a snapshot of the lvm device
> - lvcreate -s -L $lxc_size -n $lxc_new /dev/$lxc_vg/$lxc_orig || exit 1
> - echo "lxc.rootfs = /dev/$lxc_vg/$lxc_new" >> $lxc_path/$lxc_new/config
> - # and mount it so we can tweak it
> - mkdir -p $lxc_path/$lxc_new/rootfs
> - mount /dev/$lxc_vg/$lxc_new $rootfs || { echo "failed to mount new
> rootfs"; exit 1; }
> - mounted=1
> + if [ $container_running == "True" ]; then
> + lxc-freeze -n $lxc_orig
> + fi
> + lvcreate -s -L $lxc_size -n ${lxc_lv_prefix}${lxc_new}_snapshot
> $oldroot || exit 1
> + if [ $container_running == "True" ]; then
> + lxc-unfreeze -n $lxc_orig
> + fi
> + if [ $snapshot == "no" ]; then
> + #mount snapshot
> + mkdir -p ${rootfs}_snapshot
> + mount /dev/$lxc_vg/${lxc_lv_prefix}${lxc_new}_snapshot
> ${rootfs}_snapshot || { echo "failed to mount new rootfs_snapshot";
> exit 1; }
> + #create a new lv
> + lvcreate -L $lxc_size $lxc_vg -n ${lxc_lv_prefix}$lxc_new
> + echo "lxc.rootfs = /dev/$lxc_vg/${lxc_lv_prefix}$lxc_new" >>
> $lxc_path/$lxc_new/config
> + # and mount it so we can tweak it
> + mkdir -p $lxc_path/$lxc_new/rootfs
> + mkfs -t $fstype /dev/$lxc_vg/${lxc_lv_prefix}$lxc_new
> + mount /dev/$lxc_vg/${lxc_lv_prefix}$lxc_new $rootfs || {
> echo "failed to mount new rootfs"; exit 1; }
> + mounted=1
> + rsync -ax ${rootfs}_snapshot/ ${rootfs}/ || { echo "copy
> of data to new lv failed"; exit 1; }
> + umount ${rootfs}_snapshot || { echo "failed to unmount
> new rootfs_snapshot"; exit 1; }
> + rm -rf ${rootfs}_snapshot
> + lvremove -f $lxc_vg/${lxc_lv_prefix}$lxc_new || echo
> "failed to remove the snapshot"
> + else
> + lvrename $lxc_vg/${lxc_lv_prefix}}${lxc_new}_snapshot
> $lxc_vg/${lxc_lv_prefix}$lxc_new
> + echo "lxc.rootfs = /dev/$lxc_vg/${lxc_lv_prefix}$lxc_new"
> >> $lxc_path/$lxc_new/config
> + # and mount it so we can tweak it
> + mkdir -p $lxc_path/$lxc_new/rootfs
> + mount /dev/$lxc_vg/${lxc_lv_prefix}$lxc_new $rootfs || {
> echo "failed to mount new rootfs"; exit 1; }
> + mounted=1
> + fi
> +
> else
> - cp -a $lxc_path/$lxc_orig/rootfs $lxc_path/$lxc_new/rootfs || return 1
> + if [ $container_running == True ];then
> + lxc-freeze -n $lxc_orig
> + fi
> + rsync -ax $lxc_path/$lxc_orig/rootfs $lxc_path/$lxc_new/rootfs || return 1
> + if [ $container_running == True ];then
> + lxc-unfreeze -n $lxc_orig
> + fi
> echo "lxc.rootfs = $rootfs" >> $lxc_path/$lxc_new/config
> fi
>
> echo "Updating rootfs..."
> -hostname=$lxc_new
>
> # so you can 'ssh $hostname.' or 'ssh $hostname.local'
> if [ -f $rootfs/etc/dhcp/dhclient.conf ]; then
> --
> 1.7.6
>
> ------------------------------------------------------------------------------
> EMC VNX: the world's simplest storage, starting under $10K
> The only unified storage solution that offers unified management
> Up to 160% more powerful than alternatives and 25% more efficient.
> Guaranteed. http://p.sf.net/sfu/emc-vnx-dev2dev
> _______________________________________________
> Lxc-users mailing list
> Lxc-users at lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/lxc-users
More information about the lxc-users
mailing list