path: root/heartbeat/lxc.in

#!@BASH_SHELL@
# Should now conform to guidelines:
# https://github.com/ClusterLabs/resource-agents/blob/master/doc/dev-guides/ra-dev-guide.asc
#
#	LXC (Linux Containers) OCF RA.
#	Used to cluster enable the start, stop and monitoring of a LXC container.
#
# Copyright (c) 2011 AkurIT.com.au, Darren Thompson
#                    All Rights Reserved.
#
# Without limiting the rights of the original copyright holders
# This resource is licensed under GPL version 2
#
# This program is free software; you can redistribute it and/or modify
# it under the terms of version 2 of the GNU General Public License as
# published by the Free Software Foundation.
#
# This program is distributed in the hope that it would be useful, but
# WITHOUT ANY WARRANTY; without even the implied warranty of
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.
#
# Further, this software is distributed without any warranty that it is
# free of the rightful claim of any third person regarding infringement
# or the like.  Any license provided herein, whether implied or
# otherwise, applies only to this software file.  Patent licenses, if
# any, provided herein do not apply to combinations of this program with
# other software, or any other product whatsoever.
#
# You should have received a copy of the GNU General Public License
# along with this program; if not, write the Free Software Foundation,
# Inc., 59 Temple Place - Suite 330, Boston MA 02111-1307, USA.

# OCF instance parameters
#       OCF_RESKEY_container
#       OCF_RESKEY_config
#       OCF_RESKEY_log
#	OCF_RESKEY_use_screen

# Initialization:
: ${OCF_FUNCTIONS_DIR=${OCF_ROOT}/lib/heartbeat}
. ${OCF_FUNCTIONS_DIR}/ocf-shellfuncs

# Defaults
OCF_RESKEY_container_default=""
OCF_RESKEY_config_default=""
OCF_RESKEY_log_default="${HA_RSCTMP}/${OCF_RESOURCE_INSTANCE}.log"
OCF_RESKEY_use_screen_default="false"

: ${OCF_RESKEY_container=${OCF_RESKEY_container_default}}
: ${OCF_RESKEY_config=${OCF_RESKEY_config_default}}
: ${OCF_RESKEY_log=${OCF_RESKEY_log_default}}
: ${OCF_RESKEY_use_screen=${OCF_RESKEY_use_screen_default}}

# Set default TRANS_RES_STATE (temporary file to "flag" if resource was stated but not stopped)
TRANS_RES_STATE="${HA_RSCTMP}/${OCF_RESOURCE_INSTANCE}.state"

meta_data() {
cat <<END
<?xml version="1.0"?>
<!DOCTYPE resource-agent SYSTEM "ra-api-1.dtd">
<resource-agent name="lxc" version="0.1">
<version>1.0</version>
<longdesc lang="en">Allows LXC containers to be managed by the cluster.
Notes for lxc Versions before 1.0.0, where the Container is stopped using kill -PWR instead of lxc-stop:
It is 'assumed' that the 'init' system will do an orderly shudown if presented with a 'kill -PWR' signal.
On a 'sysvinit' this would require the container to have an inittab file containing "p0::powerfail:/sbin/init 0"
</longdesc>
<shortdesc lang="en">Manages LXC containers</shortdesc>

<parameters>
<parameter name="container" required="1" unique="1">
<longdesc lang="en">The unique name for this 'Container Instance' e.g. 'test1'.</longdesc>
<shortdesc lang="en">Container Name</shortdesc>
<content type="string" default="${OCF_RESKEY_container_default}"/>
</parameter>

<parameter name="config" required="1" unique="0">
<longdesc lang="en">Absolute path to the file holding the specific configuration for this container e.g. '/etc/lxc/test1/config'.</longdesc>
<shortdesc lang="en">The LXC config file.</shortdesc>
<content type="string" default="${OCF_RESKEY_config_default}"/>
</parameter>

<parameter name="log" required="0" unique="0">
<longdesc lang="en">Absolute path to the container log file</longdesc>
<shortdesc lang="en">Container log file</shortdesc>
<content type="string" default="${OCF_RESKEY_log_default}"/>
</parameter>

<parameter name="use_screen" required="0" unique="0">
<longdesc lang="en">Provides the option of capturing the 'root console' from the container and showing it on a separate screen. 
To see the screen output run 'screen -r {container name}'
The default value is set to 'false', change to 'true' to activate this option</longdesc>
<shortdesc lang="en">Use 'screen' for container 'root console' output</shortdesc>
<content type="boolean" default="${OCF_RESKEY_use_screen_default}"/>
</parameter>

</parameters>

<actions>
<action name="start"        timeout="10s" />
<action name="stop"         timeout="30s" />
<action name="monitor"      timeout="20s" interval="60s" depth="0"/>
<action name="validate-all" timeout="20s" />
<action name="meta-data"    timeout="5s" />
</actions>
</resource-agent>
END
}


LXC_usage() {
	cat <<END
	usage: $0 {start|stop|monitor|validate-all|meta-data}

	Expects to have a fully populated OCF RA-compliant environment set.
END
}

lxc_version() {
  if have_binary lxc-version ; then
    lxc-version | cut -d' ' -f 3
  else # since LXC 1.0.0 all commands knows about --version
    lxc-info --version
  fi
}

cgroup_mounted() {
# test cgroup_mounted, mount if required
	# Various possible overrides to cgroup mount point.
	# If kernel supplies cgroup mount point, prefer it.
	CGROUP_MOUNT_POINT=/var/run/lxc/cgroup
	CGROUP_MOUNT_NAME=lxc
	CGROUP_MOUNTED=false
	[[ -d /sys/fs/cgroup ]] && CGROUP_MOUNT_POINT=/sys/fs/cgroup CGROUP_MOUNT_NAME=cgroup
	# If cgroup already mounted, use it no matter where it is.
	# If multiple cgroup mounts, prefer the one named lxc if any.
	eval `awk 'BEGIN{P="";N=""}END{print("cgmp="P" cgmn="N)}($3=="cgroup"){N=$1;P=$2;if($1="lxc")exit}' /proc/mounts`
	[[ "$cgmn" && "$cgmp" && -d "$cgmp" ]] && CGROUP_MOUNT_POINT=$cgmp CGROUP_MOUNT_NAME=$cgmn CGROUP_MOUNTED=true
	$CGROUP_MOUNTED || {
		[[ -d $CGROUP_MOUNT_POINT ]] || ocf_run mkdir -p $CGROUP_MOUNT_POINT
		ocf_run mount -t cgroup $CGROUP_MOUNT_NAME $CGROUP_MOUNT_POINT
	}
	echo 1 >${CGROUP_MOUNT_POINT}/notify_on_release
	return 0
}

LXC_start() {
	# put this here as it's so long it gets messy later!!!
	if ocf_is_true $OCF_RESKEY_use_screen; then
		STARTCMD="screen -dmS ${OCF_RESKEY_container} lxc-start -f ${OCF_RESKEY_config} -n ${OCF_RESKEY_container} -o ${OCF_RESKEY_log}"
	else
		STARTCMD="lxc-start -f ${OCF_RESKEY_config} -n ${OCF_RESKEY_container} -o ${OCF_RESKEY_log} -d"
	fi

	LXC_status
	if [ $? -eq $OCF_SUCCESS ]; then
		ocf_log debug "Resource $OCF_RESOURCE_INSTANCE is already running"
		ocf_run touch "${TRANS_RES_STATE}" || exit $OCF_ERR_GENERIC
		return $OCF_SUCCESS
	fi

	cgroup_mounted
	if [ $? -ne 0 ]; then
		ocf_log err "Unable to find cgroup mount"
		exit $OCF_ERR_GENERIC
	fi

	ocf_log info "Starting" ${OCF_RESKEY_container}
	ocf_run ${STARTCMD} || exit $OCF_ERR_GENERIC

	# Spin on status, wait for the cluster manager to time us out if
	# we fail
	while ! LXC_status; do
		ocf_log info "Container ${OCF_RESKEY_container} has not started, waiting"
		sleep 1
	done

	ocf_run touch "${TRANS_RES_STATE}" || exit $OCF_ERR_GENERIC
	return $OCF_SUCCESS
}



LXC_stop() {
	
	LXC_status
	if [ $? -eq $OCF_NOT_RUNNING ]; then
		ocf_log debug "Resource $OCF_RESOURCE_INSTANCE is already stopped"
		ocf_run rm -f $TRANS_RES_STATE
		return $OCF_SUCCESS
	fi

	cgroup_mounted
	if [ $? -ne 0 ]; then
		ocf_log err "Unable to find cgroup mount"
		exit $OCF_ERR_GENERIC
	fi

	if ! ocf_version_cmp "`lxc_version`" 1.0.0 ; then
		# Use lxc-stop if we are newer than 1.0.0
		timeout=$(( ($OCF_RESKEY_CRM_meta_timeout/1000) -5 ))
		ocf_log info "Stopping Container ${OCF_RESKEY_container} using lxc-stop"
		# lxc-stop will return failure even if it reached the timeout and sucessfully hard-stopped the 
		# Container so we check below if the Container is really stopped instead of using || exit $OCF_ERR_GENERIC
		ocf_run lxc-stop -n "${OCF_RESKEY_container}" -t ${timeout}
		
		LXC_status
		if [ $? -eq $OCF_SUCCESS ]; then
			# Try to manually hard-stop if the Container is still running
			ocf_run lxc-stop -n "${OCF_RESKEY_container}" -k || exit $OCF_ERR_GENERIC
		fi
		
	else
		# Use kill -PWR
		# If the container is running "init" and is able to perform and orderly shutdown, then it should be done.
		# It is 'assumed' that the 'init' system will do an orderly shudown if presented with a 'kill -PWR' signal.
		# On a 'sysvinit' this would require the container to have an inittab file containing "p0::powerfail:/sbin/init 0"
		local shutdown_timeout
		local now
		declare -i PID=0
		declare CMD=

		# This should work for traditional 'sysvinit' and 'upstart'
		lxc-ps --name "${OCF_RESKEY_container}" -- -C init -o pid,comm |while read CN PID CMD ;do
			[ $PID -gt 1 ] || continue
			[ "$CMD" = "init" ] || continue
			ocf_log info "Sending \"OS shut down\" instruction to" ${OCF_RESKEY_container} "as it was found to be using \"sysV init\" or \"upstart\""
			kill -PWR $PID
		done
		# This should work for containers using 'systemd' instead of 'init'
		lxc-ps --name "${OCF_RESKEY_container}" -- -C systemd -o pid,comm |while read CN PID CMD ;do
			[ $PID -gt 1 ] || continue
			[ "$CMD" = "systemd" ] || continue
			ocf_log info "Sending \"OS shut down\" instruction to" ${OCF_RESKEY_container} "as it was found to be using \"systemd\""
			kill -PWR $PID
		done
		
		# The "shutdown_timeout" we use here is the operation
		# timeout specified in the CIB, minus 5 seconds
		now=$(date +%s)
		shutdown_timeout=$(( $now + ($OCF_RESKEY_CRM_meta_timeout/1000) -5 ))
		# Loop on status until we reach $shutdown_timeout
		while [ $now -lt $shutdown_timeout ]; do
			LXC_status
			status=$?
			case $status in
			"$OCF_NOT_RUNNING")
				ocf_run rm -f $TRANS_RES_STATE
				return $OCF_SUCCESS
				;;
			"$OCF_SUCCESS")
				# Container is still running, keep waiting (until
				# shutdown_timeout expires)
				sleep 1
				;;
			*)
				# Something went wrong. Bail out and
				# resort to forced stop (destroy).
				break;
			esac
			now=$(date +%s)
		done

		# If the container is still running, it will be stopped now. regardless of state!
		ocf_run lxc-stop -n ${OCF_RESKEY_container} || exit $OCF_ERR_GENERIC
	fi
	
	ocf_log info "Container" ${OCF_RESKEY_container} "stopped"
	ocf_run rm -f $TRANS_RES_STATE

	return $OCF_SUCCESS
}

LXC_status() {
	# run lxc-info with -s option for LXC-0.7.5 or later
	local lxc_info_opt="-s"
	ocf_version_cmp "`lxc_version`" 0.7.5 && lxc_info_opt=""
	S=`lxc-info $lxc_info_opt -n ${OCF_RESKEY_container}`
	ocf_log debug "State of ${OCF_RESKEY_container}: $S"
	if [[ "${S##* }" = "RUNNING" ]] ; then 
		return $OCF_SUCCESS
	fi
	return $OCF_NOT_RUNNING
}

LXC_monitor() {
	LXC_status && return $OCF_SUCCESS
	if [ -f $TRANS_RES_STATE ]; then
		ocf_log err "${OCF_RESKEY_container} is not running, but state file ${TRANS_RES_STATE} exists."
		exit $OCF_ERR_GENERIC
	fi
	return $OCF_NOT_RUNNING
}


LXC_validate() {
	# Quick check that all required attributes are set
	if [ -z "${OCF_RESKEY_container}" ]; then
		ocf_log err "LXC container name not set!"
		exit $OCF_ERR_CONFIGURED
	fi
	if [ -z "${OCF_RESKEY_config}" ]; then
		ocf_log err "LXC configuration filename name not set!"
		exit $OCF_ERR_CONFIGURED
	fi

	# Tests that apply only to non-probes
	if ! ocf_is_probe; then
		if ! [ -f "${OCF_RESKEY_config}" ]; then
			ocf_log err "LXC configuration file \"${OCF_RESKEY_config}\" missing or not found!"
			exit $OCF_ERR_INSTALLED
		fi

		if ocf_is_true $OCF_RESKEY_use_screen; then
			check_binary screen
		fi

	    check_binary lxc-start
	    check_binary lxc-stop
	    if ocf_version_cmp "`lxc_version`" 1.0.0 ; then
	        check_binary lxc-ps
	    fi
	    check_binary lxc-info
	fi
	return $OCF_SUCCESS
}

if [ $# -ne 1 ]; then
  LXC_usage
  exit $OCF_ERR_ARGS
fi

case $__OCF_ACTION in
    meta-data)	meta_data
	exit $OCF_SUCCESS
	;;
    usage|help)	LXC_usage
	exit $OCF_SUCCESS
	;;
esac

# Everything except usage and meta-data must pass the validate test
LXC_validate

case $__OCF_ACTION in
start)		LXC_start;;
stop)		LXC_stop;;
status)	LXC_status;;
monitor)	LXC_monitor;;
validate-all)	;;
*)		LXC_usage
		ocf_log err "$0 was called with unsupported arguments: $*"
		exit $OCF_ERR_UNIMPLEMENTED
		;;
esac
rc=$?
ocf_log debug "${OCF_RESOURCE_INSTANCE} $__OCF_ACTION : $rc"
exit $rc