[v6,4/5] package/docker-engine: proper sysv init file

From: Andrew Webster <awebster@arcx.com>

The packaged redhat init file is not suitable for buildroot.
This adds a customized version for buildroot.

Acked-by: Christian Stewart <christian@paral.in>
Signed-off-by: Andrew Webster <awebster@arcx.com>
Signed-off-by: Christian Stewart <christian@paral.in>
---
 package/docker-engine/docker-engine.mk |   6 +-
 package/docker-engine/docker.init      | 112 +++++++++++++++++++++++++++++++++
 2 files changed, 115 insertions(+), 3 deletions(-)
 create mode 100644 package/docker-engine/docker.init

All,

Acked-by: Christian Stewart <christian@paral.in>
> Signed-off-by: Andrew Webster <awebster@arcx.com>
> Signed-off-by: Christian Stewart <christian@paral.in>
>
>
No idea why it added my signed-off again automatically, I just wanted to
leave the ack, please excuse my random signed-off-by duplicates all over
this series.

Best,
Christian

Hello,

On Mon, 4 Jul 2016 12:28:07 -0700, Christian Stewart wrote:
> All,
> 
> Acked-by: Christian Stewart <christian@paral.in>
> > Signed-off-by: Andrew Webster <awebster@arcx.com>
> > Signed-off-by: Christian Stewart <christian@paral.in>
> >
> >  
> No idea why it added my signed-off again automatically, I just wanted to
> leave the ack, please excuse my random signed-off-by duplicates all over
> this series.

The fact that it added your SoB is actually good. When you carry the
patch from someone else, you need to SoB on his patch, because the
chain of SoB indicates the chain of persons through which the patch has
gone. That's why Peter and I also add our SoB when we apply and merge
patches in Buildroot.

Best regards,

Thomas

Le 04/07/2016 à 21:26, Christian Stewart a écrit :
> From: Andrew Webster <awebster@arcx.com>
> 
> The packaged redhat init file is not suitable for buildroot.
> This adds a customized version for buildroot.
> 
> Acked-by: Christian Stewart <christian@paral.in>
> Signed-off-by: Andrew Webster <awebster@arcx.com>
> Signed-off-by: Christian Stewart <christian@paral.in>

I think patch 4/5 and 5/5 must be squashed with the patch adding docker-engine
package since the redhad init file doesn't work with Buildroot.

> ---
>  package/docker-engine/docker-engine.mk |   6 +-
>  package/docker-engine/docker.init      | 112 +++++++++++++++++++++++++++++++++
>  2 files changed, 115 insertions(+), 3 deletions(-)
>  create mode 100644 package/docker-engine/docker.init
> 
> diff --git a/package/docker-engine/docker-engine.mk b/package/docker-engine/docker-engine.mk
> index 501afbc..4cb896c 100644
> --- a/package/docker-engine/docker-engine.mk
> +++ b/package/docker-engine/docker-engine.mk
> @@ -103,10 +103,10 @@ define DOCKER_ENGINE_INSTALL_INIT_SYSTEMD
>  endef
>  
>  define DOCKER_ENGINE_INSTALL_INIT_SYSV
> -	$(INSTALL) -D -m 755 $(@D)/contrib/init/sysvinit-redhat/docker \
> +	$(INSTALL) -D -m 755 package/docker-engine/docker.init \
>  		$(TARGET_DIR)/etc/init.d/S61docker
> -	$(INSTALL) -D -m 644 $(@D)/contrib/init/sysvinit-redhat/docker.sysconfig \
> -		$(TARGET_DIR)/etc/sysconfig/docker.sysconfig
> +	$(INSTALL) -D -m 644 $(@D)/contrib/init/sysvinit-debian/docker.default \
> +		$(TARGET_DIR)/etc/default/docker
>  endef
>  
>  define DOCKER_ENGINE_USERS
> diff --git a/package/docker-engine/docker.init b/package/docker-engine/docker.init
> new file mode 100644
> index 0000000..a751551
> --- /dev/null
> +++ b/package/docker-engine/docker.init
> @@ -0,0 +1,112 @@
> +#!/bin/sh
> +set -e

Other init script doesn't set exit on error.

> +
> +### BEGIN INIT INFO
> +# Provides:           docker
> +# Required-Start:     $syslog $remote_fs
> +# Required-Stop:      $syslog $remote_fs
> +# Should-Start:       cgroupfs-mount cgroup-lite
> +# Should-Stop:        cgroupfs-mount cgroup-lite
> +# Default-Start:      2 3 4 5
> +# Default-Stop:       0 1 6
> +# Short-Description:  Create lightweight, portable, self-sufficient containers.
> +# Description:
> +#  Docker is an open-source project to easily create lightweight, portable,
> +#  self-sufficient containers from any application. The same container that a
> +#  developer builds and tests on a laptop can run at scale, in production, on
> +#  VMs, bare metal, OpenStack clusters, public clouds and more.
> +### END INIT INFO

run level are not used if Busybox init is used, so we can remove this comment.

> +
> +export PATH=/sbin:/bin:/usr/sbin:/usr/bin:/usr/local/sbin:/usr/local/bin

Why are you adding /usr/local/bin to the PATH here ?

By default the PATH is: /usr/local/sbin:/sbin:/bin:/usr/sbin:/usr/bin

http://www.tldp.org/HOWTO/Path-4.html

> +
> +BASE=docker
> +
> +# modify these in /etc/default/$BASE (/etc/default/docker)
> +DOCKERD=/usr/bin/dockerd
> +# This is the pid file managed by docker itself
> +DOCKER_PIDFILE=/var/run/$BASE.pid
> +# This is the pid file created/managed by start-stop-daemon
> +DOCKER_SSD_PIDFILE=/var/run/$BASE-ssd.pid
> +DOCKER_LOGFILE=/var/log/$BASE.log
> +DOCKER_OPTS=

DOCKER_OPTS can be removed

> +DOCKER_DESC="Docker"
> +
> +if [ -f /etc/default/$BASE ]; then
> +       . /etc/default/$BASE
> +fi
> +
> +# Check docker is present
> +if [ ! -x $DOCKERD ]; then
> +       echo "$DOCKERD not present or not executable"
> +       exit 1
> +fi
> +
> +cgroupfs_mount() {
> +       # see also https://github.com/tianon/cgroupfs-mount/blob/master/cgroupfs-mount
                                                                   ^
I think the comment should use a hast instead of master branch, the content of
cgroupfs-mount may change over the time.

> +       if grep -v '^#' /etc/fstab | grep -q cgroup \
> +               || [ ! -e /proc/cgroups ] \
> +               || [ ! -d /sys/fs/cgroup ]; then
> +               return
> +       fi
> +       if ! mountpoint -q /sys/fs/cgroup; then
> +               mount -t tmpfs -o uid=0,gid=0,mode=0755 cgroup /sys/fs/cgroup
> +       fi
> +       (
> +               cd /sys/fs/cgroup
> +               for sys in $(awk '!/^#/ { if ($4 == 1) print $1 }' /proc/cgroups); do
> +                       mkdir -p $sys
> +                       if ! mountpoint -q $sys; then
> +                               if ! mount -n -t cgroup -o $sys cgroup $sys; then
> +                                       rmdir $sys || true
> +                               fi
> +                       fi
> +               done
> +       )
> +}
> +
> +case "$1" in
> +       start)
> +               cgroupfs_mount
> +
> +               ulimit -n 1048576
> +               if [ "$BASH" ]; then
> +                       ulimit -u 1048576
> +               else
> +                       ulimit -p 1048576
> +               fi

The init shell is not bash, so the test on $BASH is never true.

> +
> +               echo "Starting $DOCKER_DESC: $BASE"
> +               start-stop-daemon --start --background \
> +                       --exec "$DOCKERD" \
> +                       --pidfile "$DOCKER_SSD_PIDFILE" \
> +                       --make-pidfile \
> +                       -- \
> +                               -p "$DOCKER_PIDFILE" \
> +                               $DOCKER_OPTS \
> +                                       >> "$DOCKER_LOGFILE" 2>&1
> +               echo $?
> +               ;;
> +
> +       stop)
> +               echo "Stopping $DOCKER_DESC: $BASE"
> +               start-stop-daemon --stop --pidfile "$DOCKER_SSD_PIDFILE" --retry 10
> +               echo $?
> +               ;;
> +
> +       restart)
> +               docker_pid=`cat "$DOCKER_SSD_PIDFILE" 2>/dev/null`
> +               [ -n "$docker_pid" ] \
> +                       && ps -p $docker_pid > /dev/null 2>&1 \

Please, squash the next patch here.
Indeed ps -p is not supported by busybox, if you really need it then you can
select procps-ng package.

> +                       && $0 stop
> +               $0 start
> +               ;;
> +
> +       force-reload)
> +               $0 restart
> +               ;;
> +
> +       *)
> +               echo "Usage: service docker {start|stop|restart}"
> +               exit 1
> +               ;;
> +esac
>