[poky] [PATCH 2/2] Add a skeleton for init scripts

[Robert Yang]

Saul suggested that if can't config busybox to make the pidof support the
"-x" option, then we would declare that it doesn't support the "shell script"
as a daemon with this skeleton unless the pidof supports "-x"(Then the user
should  write the init script himself). This may be fine, there are only 3
shell scripts in core-image-minimal(In /bin/, /usr/bin/, /sbin/, /usr/sbin/).

// Robert

On 05/12/2011 11:41 PM, Robert Yang wrote:
> Hi Darren,
>
> I find that the busybox's pidof can't get the pid of the "shell script"
> (opposite to the binary) currently, it doesn't support the "-x" option,
> so it seems that we should patch the busybox's pidof.c, is this acceptable
> or not?
>
> Thank you very much
>
> Robert
>
> On 05/12/2011 01:56 AM, Darren Hart wrote:
>> Hi Robert,
>>
>> Thank you for doing this. This is a great start. As a preface to my
>> comments below, I am going to be incredibly particular about formatting
>> and such as this skeleton will be used by literally thousands of recipes
>> in the future and now is the time to ensure consistent formatting and to
>> eliminate any issues we don't want to see propagated.
>>
>>
>> On 05/11/2011 06:41 AM, Robert Yang wrote:
>>> Add a skeleton for init scripts, the original structure is from
>>> /etc/init.d/skeleton of Ubuntu 10.10, it is in sysvinit_2.87dsf,
>>> so add the COPYRIGHT of sysvinit_2.87dsf, it is GPLv2. I have
>>> modified the original skeleton a lot to make it as easy as possible,
>>> just use posix shell command, and have tested it in core-image-minimal.
>>>
>>> * The skeleton implements the following items:
>>> - start, stop, restart, status and force-reload.
>>> And the force-reload is a alias of restart.
>>> - not implements reload and try-restart, since only a few programs
>>> have such functions, so just leave them as placeholders.
>>>
>>> * Use start-stop-daemon to start and stop the program, it is useful to
>>> deal with the pidfile, check whether there is already a running
>>> program or not, and print relevant messages.
>>>
>>> * The skeleton will run by default, and output the *simulation* message,
>>> here is the testing result:
>>> 1) #./skeleton start (test start)
>>> Starting skeleton ...
>>>
>>> 2) #./skeleton start (test start the already running program)
>>> Starting skeleton ...
>>> /usr/sbin/skeleton already running.
>>> 20426
>>>
>>> It outputs two lines when start again, one line would be better, this
>>> is because of the start-stop-daemon, but I'm not sure whether this is
>>> worth.
>>>
>>> 3) #./skeleton status (test status when running)
>>> skeleton is running.
>>
>> We should use a consistent string for the daemon. You use the full path
>> in #2 above and only the basename here in #3.
>>
>>>
>>> 4) #./skeleton stop (test stop)
>>> Stopped skeleton (pid 20426).
>>
>> I prefer this format (with the pid listed on the same line) to that in
>> #2. #3 should also list the pid. For example, my dovecot initscript reports:
>>
>> $ status dovecot
>> dovecot start/running, process 18529
>>
>> It our case something like the following seems adequate.
>>
>> $ ./skeleton status
>> skeleton is running (18529)
>>
>>>
>>> 5) #./skeleton stop (test stop again)
>>> No skeleton found running; none killed.
>>
>> I'd prefer:
>>
>> skeleton is not running; none killed.
>>
>>>
>>> 6) #./skeleton status (test status when stopped)
>>> skeleton is not running.
>>>
>>> 7) #./skeleton start (test restart when running)
>>> Starting skeleton ...
>>>
>>> #./skeleton restart
>>> Stopped skeleton (pid 20444).
>>> Starting skeleton ...
>>>
>>> 8) #./skeleton stop (test restart when stopped)
>>> Stopped skeleton (pid 20444).
>>>
>>> #./skeleton restart
>>> No skeleton found running; none killed.
>>
>> skeleton is not running; none killed.
>>
>>> Starting skeleton ...
>>>
>>> * Have used syslogd to test it in a real world(With both
>>> core-image-minimal and core-image-sato)
>>>
>>> * TODO:
>>> - Move the function status_of_proc to /etc/init.d/functions ?
>>
>>
>> It's not a bad idea. If so, that has to be part of the base images, not
>> the skeleton.
>>
>>
>>> - Fix start-stop-daemon to get an one line output when start a
>>> already runing program?
>>
>>
>> I wouldn't bother. I don't see anything wrong with two lines.
>>
>>
>>>
>>> Signed-off-by: Robert Yang<liezhi.yang at windriver.com>
>>> ---
>>> .../recipes-skeleton/service/service/COPYRIGHT | 19 ++
>>> .../recipes-skeleton/service/service/skeleton | 188 ++++++++++++++++++++
>>> .../recipes-skeleton/service/service_0.1.bb | 20 ++
>>> 3 files changed, 227 insertions(+), 0 deletions(-)
>>> create mode 100644 meta-skeleton/recipes-skeleton/service/service/COPYRIGHT
>>> create mode 100644 meta-skeleton/recipes-skeleton/service/service/skeleton
>>> create mode 100644 meta-skeleton/recipes-skeleton/service/service_0.1.bb
>>>
>>> diff --git a/meta-skeleton/recipes-skeleton/service/service/COPYRIGHT
>>> b/meta-skeleton/recipes-skeleton/service/service/COPYRIGHT
>>> new file mode 100644
>>> index 0000000..36703d9
>>> --- /dev/null
>>> +++ b/meta-skeleton/recipes-skeleton/service/service/COPYRIGHT
>>> @@ -0,0 +1,19 @@
>>> +Sysvinit is Copyright (C) 1991-2004 Miquel van Smoorenburg
>>> +
>>> + This program is free software; you can redistribute it and/or modify
>>> + it under the terms of the GNU General Public License as published by
>>> + the Free Software Foundation; either version 2 of the License, or
>>> + (at your option) any later version.
>>> +
>>> + This program is distributed in the hope that it will be useful,
>>> + but WITHOUT ANY WARRANTY; without even the implied warranty of
>>> + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
>>> + GNU General Public License for more details.
>>> +
>>> + You should have received a copy of the GNU General Public License
>>> + along with this program; if not, write to the Free Software
>>> + Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
>>> +
>>> +On Debian GNU/Linux systems, the complete text of the GNU General
>>> +Public License can be found in `/usr/share/common-licenses/GPL'.
>>
>>
>> You don't need to keep this last bit of where to find it on Debian
>> systems - we aren't a Debian system.
>>
>>
>>> +
>>> diff --git a/meta-skeleton/recipes-skeleton/service/service/skeleton
>>> b/meta-skeleton/recipes-skeleton/service/service/skeleton
>>> new file mode 100644
>>> index 0000000..58804e4
>>> --- /dev/null
>>> +++ b/meta-skeleton/recipes-skeleton/service/service/skeleton
>>> @@ -0,0 +1,188 @@
>>> +#! /bin/sh
>>> +### BEGIN INIT INFO
>>> +# Provides: skeleton
>>> +# Required-Start: $local_fs
>>> +# Should-Start:
>>> +# Required-Stop: $local_fs
>>> +# Should-Stop:
>>> +# Default-Start: 2 3 4 5
>>> +# Default-Stop: 0 1 6
>>> +# Short-Description: Example initscript
>>> +# Description: This file should be used to construct scripts to be
>>> +# placed in /etc/init.d.
>>> +### END INIT INFO
>>> +
>>> +# Common steps to convert this skeleton into a real init script
>>> +# 1) cp skeleton<the_real_name>
>>> +# 2) Set DESC and NAME
>>> +# 3) Check whether the daemon app is /usr/sbin/$NAME, if not, set it.
>>> +# 4) Set DAEMON_ARGS if there is any
>>> +# 5) Remove the useless code
>>> +
>>> +# PATH should only include /usr/* if it runs after the mountnfs.sh script
>>> +PATH=/sbin:/usr/sbin:/bin:/usr/bin
>>
>>
>> Paths like usr and etc need to be rewritten at install. I wonder if
>> there might not be a better way to represent them in the skeleton.
>>
>>
>>> +
>>> +DESC="skeleton"
>>> +NAME="skeleton"
>>> +DAEMON=/usr/sbin/$NAME
>>> +DAEMON_ARGS=""
>>> +PIDFILE=/var/run/$NAME.pid
>>> +SCRIPTNAME=/etc/init.d/$NAME
>>> +
>>> +#### Simulation begin
>>> +# Simulate the script running, this is only used for printing the
>>> +# information by default, it does nothing and should be removed from
>>> +# the real init script
>>
>>
>> It should be removed from the skeleton as well. I suggest writing a test
>> script which calls "skeleton start; skeleton stop" etc. Especially since
>> you're not testing the actual script by running this, and you duplicate
>> code. So if you update it in only one spot, this testing will not catch it.
>>
>>
>>> +fake_do_stop () {
>>> + if [ -f $PIDFILE ]; then
>>> + echo "Stopped $NAME (pid `cat $PIDFILE`)."
>>> + rm -f $PIDFILE
>>> + else
>>> + echo "No $NAME found running; none killed."
>>> + fi
>>> +}
>>> +
>>> +if [ "$NAME" = "skeleton" ]; then
>>> + case $1 in
>>> + start)
>>> + echo "Starting $DESC ..."
>>> + if [ -f $PIDFILE ]; then
>>> + echo "$DAEMON already running."
>>> + cat $PIDFILE
>>> + else
>>> + echo $$>$PIDFILE
>>> + fi
>>> + ;;
>>> + stop)
>>> + fake_do_stop
>>> + ;;
>>> + status)
>>> + if [ -f $PIDFILE ]; then
>>> + echo "$NAME is running."
>>> + else
>>> + echo "$NAME is not running."
>>> + fi
>>> + ;;
>>> + restart|force-reload)
>>> + fake_do_stop
>>> + echo "Starting $DESC ..."
>>> + echo $$>$PIDFILE
>>> + ;;
>>> + *)
>>> + echo "Usage: $SCRIPTNAME {start|stop|status|restart|force-reload}">&2
>>> + exit 3
>>> + ;;
>>> + esac
>>> +exit 0
>>> +fi
>>> +#### Simulation end
>>> +
>>> +# Exit if the package is not installed
>>> +[ -x "$DAEMON" ] || exit 0
>>> +
>>> +# Read configuration variable file if it is present
>>> +[ -r /etc/default/$NAME ]&& . /etc/default/$NAME
>>> +
>>> +#
>>> +# Function that starts the daemon/service
>>> +#
>>> +do_start()
>>> +{
>>> + start-stop-daemon -S --pidfile $PIDFILE --exec $DAEMON -- $DAEMON_ARGS
>>> + # Add code here, if necessary, that waits for the process to be ready
>>> + # to handle requests from services started subsequently which depend
>>> + # on this one. As a last resort, sleep for some time.
>>> +}
>>> +
>>> +#
>>> +# Function that stops the daemon/service
>>> +#
>>> +do_stop()
>>> +{
>>> + start-stop-daemon -K -v --pidfile $PIDFILE --name $NAME
>>> + # Wait for children to finish too if this is a daemon that forks
>>> + # and if the daemon is only ever run from this initscript.
>>> + # If the above conditions are not satisfied then add some other code
>>> + # that waits for the process to drop all resources that could be
>>> + # needed by services started subsequently. A last resort is to
>>> + # sleep for some time.
>>> + #
>>> + #start-stop-daemon -K --oknodo --exec $DAEMON
>>
>>
>> What does this command do?
>>
>>
>>> +
>>> + # Many daemons don't delete their pidfiles when they exit.
>>> + rm -f $PIDFILE
>>
>>
>> Should STDERR be redirected to /dev/null in case the file does not exist?
>>
>>
>>> +}
>>> +
>>> +#
>>> +# Function that sends a SIGHUP to the daemon/service
>>> +#
>>> +#do_reload() {
>>> + #
>>> + # If the daemon can reload its configuration without
>>> + # restarting (for example, when it is sent a SIGHUP),
>>> + # then implement that here.
>>> + #
>>> + #start-stop-daemon -K --signal 1 --pidfile $PIDFILE --name $NAME
>>> +#}
>>> +
>>> +#
>>> +# Function that shows the daemon/service status
>>> +#
>>> +status_of_proc () {
>>> + pidof $NAME>/dev/null 2>&1
>>> + case $? in
>>> + 0)
>>> + echo "$NAME is running."
>>> + return 0
>>> + ;;
>>> + 127)
>>> + echo "Can't get the status of $NAME (no pidof found)"
>>> + ;;
>>> + *)
>>> + echo "$NAME is not running.">&2
>>> + return 1
>>> + ;;
>>> + esac
>>> +}
>>> +
>>> +case "$1" in
>>> + start)
>>> + echo "Starting $DESC ..."
>>> + # Don't need check the return value, start-stop-daemon would
>>> + # print the failed reason automatically.
>>> + do_start
>>> + ;;
>>> + stop)
>>> + do_stop
>>> + ;;
>>> + status)
>>> + status_of_proc&& exit 0 || exit $?
>>> + ;;
>>> + #reload|force-reload)
>>
>>
>> We don't want to have force-reload in two locations, even commented out.
>> Going to lead to confusion.
>>
>>
>>> + #
>>> + # If do_reload() is not implemented then leave this commented out
>>> + # and leave 'force-reload' as an alias for 'restart'.
>>> + #
>>> + #echo "Reloading $DESC ..."
>>> + #do_reload
>>> + #;;
>>> + restart|force-reload)
>>> + #
>>> + # If the "reload" option is implemented then remove the
>>> + # 'force-reload' alias.
>>> + #
>>> + do_stop
>>> + echo "Starting $DESC ..."
>>> + do_start
>>> + ;;
>>> + #try-restart)
>>> + # If do_try_restart() is not implemented then leave this commented out
>>> + #echo "Trying to restart $DESC ..."
>>
>>
>> This needs a description of how it should be implemented. i.e. how is
>> this different from restart?
>>
>>
>>> + #do_try_restart()
>>> + #;;
>>> + *)
>>> + echo "Usage: $SCRIPTNAME {start|stop|status|restart|force-reload}">&2
>>> + exit 3
>>> + ;;
>>> +esac
>>> +
>>> diff --git a/meta-skeleton/recipes-skeleton/service/service_0.1.bb
>>> b/meta-skeleton/recipes-skeleton/service/service_0.1.bb
>>> new file mode 100644
>>> index 0000000..88a7a59
>>> --- /dev/null
>>> +++ b/meta-skeleton/recipes-skeleton/service/service_0.1.bb
>>> @@ -0,0 +1,20 @@
>>> +DESCRIPTION = "The canonical example of init scripts"
>>> +SECTION = "base"
>>> +LICENSE = "GPLv2"
>>> +LIC_FILES_CHKSUM =
>>> "file://${WORKDIR}/COPYRIGHT;md5=3900421dc55b9e70428bc522557a66d4"
>>> +RDEPENDS_${PN} = "initscripts"
>>> +PR = "r0"
>>> +
>>> +SRC_URI = "file://skeleton \
>>> + file://COPYRIGHT \
>>> + "
>>> +
>>> +CONFFILES_${PN} += "${sysconfdir}/init.d/skeleton"
>>> +
>>> +PACKAGE_ARCH = "all"
>>> +
>>> +do_install () {
>>> + install -d ${D}/${sysconfdir}/init.d
>>> + install -m 0755 ${WORKDIR}/skeleton ${D}/${sysconfdir}/init.d/
>>> +}
>>> +
>>
>> This installs a file that uses directories like /usr which should be
>> replaced with the appropriate setting. For example, in mediatomb (and
>> others) we do this like this:
>>
>> cat ${WORKDIR}/config.xml | \
>> sed -e 's,/etc,${sysconfdir},g' \
>> -e 's,/usr/sbin,${sbindir},g' \
>> -e 's,/var,${localstatedir},g' \
>> -e 's,/usr/bin,${bindir},g' \
>> -e 's,/usr,${prefix},g'> ${D}${sysconfdir}/mediatomb/config.xml
>>