#!/bin/sh
################################################################################
#
# This file is part of the vhacs project.
# Jerome Martin <jmartin@longphone.fr>
# based on RA by SUSE LINUX Products GmbH, Lars Marowsky-Bree
#
################################################################################
#
#
################################################################################
#
# OCF instance parameters:
#	OCF_RESKEY_drbd_resource
#	OCF_RESKEY_drbdconf
#	OCF_RESKEY_CRM_meta_clone_max
#	OCF_RESKEY_CRM_meta_clone_node_max
#	OCF_RESKEY_master_max
#	OCF_RESKEY_master_node_max
#
################################################################################
# 
# Source common vhacs OCF RA code.
. ${OCF_ROOT}/resource.d/vhacs/.ocf-vhacs
#
################################################################################

rsc_meta_data() {
	cat <<- EOF
		<?xml version="1.0"?>
		<!DOCTYPE resource-agent SYSTEM "ra-api-1.dtd">
		<resource-agent name="drbd">
			<version>1.1</version>
			<longdesc lang="en">
				Master/Slave OCF Resource Agent for DRBD
			</longdesc>
			<shortdesc lang="en">This resource agent manages a Distributed
				Replicated Block Device (DRBD) object as a master/slave
				resource. DRBD is a mechanism for replicating storage; please see the
				documentation for setup details.
			</shortdesc>
			<parameters>
				<parameter name="drbd_resource" unique="1" required="1">
					<longdesc lang="en">
						The name of the drbd resource from the drbd.conf file.
					</longdesc>
					<shortdesc lang="en">drbd resource name</shortdesc>
					<content type="string" default="drbd0" />
				</parameter>
				<parameter name="drbdconf">
					<longdesc lang="en">
						Full path to the drbd.conf file.
					</longdesc>
					<shortdesc lang="en">Path to drbd.conf</shortdesc>
					<content type="string" default="/etc/drbd.conf"/>
				</parameter>
				<parameter name="clone-max" required="1">
					<longdesc lang="en">
						Number of clones of this drbd resource. Do not fiddle with the default.
					</longdesc>
					<shortdesc lang="en">Number of clones</shortdesc>
					<content type="integer" default="2"/>
				</parameter>
				<parameter name="clone-node-max" required="1">
					<longdesc lang="en">
						Clones per node. Do not fiddle with the default.
					</longdesc>
					<shortdesc lang="en">Number of nodes</shortdesc>
					<content type="integer" default="1"/>
				</parameter>
				<parameter name="master-max" required="1">
					<longdesc lang="en">
						Maximum number of active primaries. Do not fiddle with the default.
					</longdesc>
					<shortdesc lang="en">Number of primaries</shortdesc>
					<content type="integer" default="1"/>
				</parameter>
				<parameter name="master-node-max" required="1">
					<longdesc lang="en">
						Maximum number of primaries per node. Do not fiddle with the default.
					</longdesc>
					<shortdesc lang="en">Number of primaries per node</shortdesc>
					<content type="integer" default="1"/>
				</parameter>
			</parameters>
			<actions>
				<action name="start"   timeout="240" />
				<action name="promote"   timeout="90" />
				<action name="demote"   timeout="90" />
				<action name="notify"   timeout="90" />
				<action name="stop"    timeout="100" />
				<action name="monitor" depth="0"  timeout="20" interval="20" start-delay="1m" role="Slave" />
				<action name="monitor" depth="0"  timeout="20" interval="10" start-delay="1m" role="Master" />
				<action name="meta-data"  timeout="5" />
				<action name="validate-all"  timeout="30" />
			</actions>
		</resource-agent>
	EOF
}

rsc_start() {
	# Start the drbd device
	#ocf_lock_RA
	ocf_cmd "${DRBDADM} up ${RESOURCE}" || return ${OCF_ERR_GENERIC}
	#ocf_unlock_RA

	# If there where no errors, return success
	return ${OCF_SUCCESS}
}

rsc_stop() {
	# Clear preference for becoming master
	ocf_log info "rsc_stop trying to delete preference for becoming master"
	ocf_cmd "${CRM_MASTER} -D"

	# Shutdown the drbd device
	ocf_log info "rsc_stop trying to shut down drbd device ${RESOURCE}"
	#ocf_lock_RA
	ocf_cmd "${DRBDADM} down ${RESOURCE}"
	#ocf_unlock_RA

	# We've done all we could to stop that resource, let monitor check it
	return ${OCF_SUCCESS}
}

rsc_monitor() {
	if [ ${DRBD_MINOR} -lt 10 ]; then
		DRBD_MINOR_PADDED=" ${DRBD_MINOR}"
	else
		DRBD_MINOR_PADDED="${DRBD_MINOR}"
	fi
	DRBD_CSTATE=$(awk "/^${DRBD_MINOR_PADDED}:/ {print \$2}" /proc/drbd | sed 's/cs://g' | tr -d '\n')
	DRBD_STATE_LOCAL=$(awk  "/^${DRBD_MINOR_PADDED}:/ {print \$3}" /proc/drbd | sed 's/st://g' | awk -F / '{print $1}' | tr -d '\n')
	DRBD_STATE_REMOTE=$(awk "/^${DRBD_MINOR_PADDED}:/ {print \$3}" /proc/drbd | awk -F / '{print $2}' | tr -d '\n')

	ocf_log info "${RESOURCE} status: local state is ${DRBD_STATE_LOCAL}, remote state is ${DRBD_STATE_REMOTE} and cstate is ${DRBD_CSTATE}"

	# Sanitize the various states, drbdadm is quite annoying; so if it 
	# outputs something which doesn't make sense, translate it into
	# a harmless state:

	case "${DRBD_STATE_LOCAL}" in
		"Not configured"|"Primary"|"Secondary") ;;
		*)	DRBD_STATE_LOCAL="Not configured" ;;
	esac

	case "${DRBD_STATE_REMOTE}" in
		"Primary"|"Secondary"|"Unknown") ;;
		*)	DRBD_STATE_REMOTE="Not configured" ;;
	esac

	case "${DRBD_CSTATE}" in
		cs:Unconfigured|cs:StandAlone|cs:Unconnected|cs:Timeout|cs:BrokenPipe) ;;
		cs:NetworkFailure|cs:WFConnection|cs:WFReportParams|cs:Connected|cs:SkippedSyncS) ;;
		cs:SkippedSyncT|cs:WFBitMapS|cs:WFBitMapT|cs:SyncSource|cs:SyncTarget) ;; 
		cs:PausedSyncS|cs:PausedSyncT) ;;
		*) DRBD_CSTATE="Unconfigured" ;;
	esac

	case ${DRBD_STATE_LOCAL} in
		"Not configured")
			return ${OCF_NOT_RUNNING}
		;;
		"Primary")
			ocf_cmd ${CRM_MASTER} -v 100
			return ${OCF_RUNNING_MASTER}
		;;
		"Secondary")
			ocf_cmd ${CRM_MASTER} -v 100
			return ${OCF_SUCCESS}	
		;;
		*)
	    	ocf_log err "${RESOURCE} monitor: unexpected local state: ${DRBD_STATE_LOCAL}"
			return ${OCF_ERR_GENERIC}
		;;
	esac
}

rsc_promote() {
	# Promote the device to primary
	ocf_log info "rsc_promote trying to promote drbd device ${RESOURCE} to primary"
	ocf_cmd "${DRBDADM} primary ${RESOURCE}" || return ${OCF_ERR_GENERIC}

	# If there where no errors, return success
	return ${OCF_SUCCESS}
}

rsc_demote() {
	# Demote the device to secondary
	ocf_cmd "${DRBDADM} secondary ${RESOURCE}" || return ${OCF_ERR_GENERIC}

	# If there where no errors, return success
	return ${OCF_SUCCESS}
}

rsc_notify() {
	local n_type="${OCF_RESKEY_CRM_meta_notify_type}"
	local n_op="${OCF_RESKEY_CRM_meta_notify_operation}"
	set -- ${OCF_RESKEY_CRM_meta_notify_active_resource}
	local n_active="$#"
	set -- ${OCF_RESKEY_CRM_meta_notify_stop_resource}
	local n_stop="$#"
	set -- ${OCF_RESKEY_CRM_meta_notify_start_resource}
	local n_start="$#"

	ocf_log info "${RESOURCE} notify: ${n_type} for ${n_op} - counts: active ${n_active} - starting ${n_start} - stopping ${n_stop}"
	
	case ${n_type} in
	pre)
		case ${n_op} in
		promote) # TODO:
			 # Resist promotion of the other side in case we
			 # are already primary - though the CRM should
			 # not even attempt that.
			;;
		esac
		;;
	post)	# TODO: Entire case statement which follows redundant?
		case ${n_op} in
		start)
			if [ "${n_active}" -eq 2 ]; then
				# The other side is running, so we ought
				# to connect and wait for that.
				${DRBDADM} connect ${RESOURCE}
				${DRBDADM} wait_connect ${RESOURCE}
				# TODO: If this can cause a hang if the
				# other side isn't connected or goes
				# away during that, maybe just sleep
				# here for 5-10s or take out the entire
				# case statement
			fi
			;;
		stop)
			# TODO BUG: disconnect seems to force
			# non-primary mode?!?
			#### ${DRBDADM} disconnect $RESOURCE

			# TODO: If we are secondary, do we need to do
			# anything about a stopped primary in case we
			# had an outdated flag...?
			;;
		esac

		;;
	esac
	
	return ${OCF_SUCCESS}
}

rsc_validate_all () {
	# First check the configuration file
	if [ -n "${DRBDCONF}" ] && [ ! -f "${DRBDCONF}" ]; then
	    ocf_log err "Configuration file does not exist: ${DRBDCONF}"
		return ${OCF_ERR_INSTALLED}
	fi

	# Check the resource name, it should appear in DRBDCONF
	if [ -z "${RESOURCE}" ]; then
	    ocf_log err "No resource name specified!"
	    return ${OCF_ERR_ARGS}
	fi
	
	if ${DRBDADM} dump ${RESOURCE} 2>/dev/null ; then
	    :
	else
	    ocf_log err "Invalid configuration file ${DRBDCONF}"
	    return ${OCF_ERR_CONFIGURED}
	fi
	
	if [ "${OCF_RESKEY_CRM_meta-clone-max}" -ne 2 ] \
	 || [ "${OCF_RESKEY_CRM-meta-clone-node_max}" -ne 1 ] \
	 || [ "${OCF_RESKEY_master-node-max}" -ne 1 ] \
	 || [ "${OCF_RESKEY_master-max}" -ne 1 ] ; then
		ocf_log err "Clone options misconfigured."
		exit ${OCF_ERR_CONFIGURED}
	fi
	
	return ${OCF_SUCCESS}
}

rsc_setup() {
	CRM_MASTER="${HA_SBIN_DIR}/crm_master -l reboot "
	RESOURCE="${OCF_RESKEY_drbd_resource}"
	CLONE_NO="${OCF_RESKEY_CRM_meta_clone}"
	DRBDCONF="${OCF_RESKEY_drbdconf:=/etc/drbd.conf}"
	DRBDADM="/sbin/drbdadm -c ${DRBDCONF}"

	if [ ! -f "${DRBDCONF}" ]; then
		return ${OCF_ERR_INSTALLED}
	fi

	# determine the drbd device number from the config file
	# this works with vhacs-generated drbd config files,
	# with only one single resource defined and device parameter
	# within 4 lines of node definition start

	DRBD_MAJOR=147
	DRBD_MINOR=$(grep -A 4 "on $(uname -n)" ${DRBDCONF} | awk '/device/ {print $2}' | \
	tr -d ';' | sed 's/[a-zA-Z/]*\([0-9]*\)/\1/' | tr -d '\n')

	[ -z "${DRBD_MINOR}" ] && return ${OCF_ERR_GENERIC}

	ocf_log info "drbd device is ${DRBD_MAJOR}:${DRBD_MINOR}"

	return ${OCF_SUCCESS}
}

################################################################################
#
# Call the main ocf RA logic
#
################################################################################
ocf_main ${1}