#!/bin/sh

# This is an example CTDB NFS callout script for Ganesha.  It is based
# on the last version of 60.ganesha shipped with CTDB.  As such, it
# does not try to monitor RPC services that were not monitored by
# 60.ganesha - this might be a useful improvement.  It has also not
# been properly tested.

# You should check your version of NFS Ganesha to see if it ships with
# a newer callout.

# To use this:
#
# * Set CTDB_NFS_CALLOUT in your CTDB configuration to point to (a
#   copy of) this script, making sure it is executable.
#
# * Create a new directory alongside the nfs-checks.d directory, for
#   example nfs-checks-ganesha.d.  Install 20.nfs-ganesha.check in
#   this directory.  Symlink to any other check files from
#   nfs-checks.d that should still be used, such as
#   00.portmapper.check.  Set CTDB_NFS_CHECKS_DIR to point to this new
#   directory of check files.
#
# * It is recommended, but not required, to install the grace_period
#   script (usually shipped in a utils package for NFS-Ganesha) to
#   /usr/bin/grace_period

# I (Martin Schwenke) hereby relicense all of my contributions to this
# callout (and, previously, to 60.ganesha) to a license compatible
# with NFS Ganesha (right now this is LGPLv3, but I'm flexible).
# There may be other contributions to be considered for relicensing,
# particularly those in commit 28cbe527d47822f870e8252495ab2a1c8fddd12f.

######################################################################

# Exit on 1st error
set -e

# Filesystem type and mount point for the (typically clustered)
# volume that will contain the NFS-Ganesha state.
state_fs="${CTDB_NFS_STATE_FS_TYPE:-gpfs}"
state_dir="${CTDB_NFS_STATE_MNT}" # No sane default.

# To change the following, edit the default values below.  Do not set
# these - they aren't configuration variables, just hooks for testing.
nfs_exports_file="${CTDB_NFS_EXPORTS_FILE:-/etc/ganesha/ganesha.conf}"
nfs_service="${CTDB_NFS_SERVICE:-nfs-ganesha}"
ganesha_rec_subdir=${CTDB_GANESHA_REC_SUBDIR:-.ganesha}
procfs=${PROCFS_PATH:-/proc}

case "$state_fs" in
gpfs)
	GANRECDIR="/var/lib/nfs/ganesha"
	;;
glusterfs)
	if [ -z "${state_dir}" ]; then
		echo "CTDB_NFS_STATE_MNT not defined for GlusterFS"
		exit 1
	fi
	host=$(hostname)
	NODESTATEDIR="$state_dir/nfs-ganesha/$host"
	GANSTATEDIR="$state_dir/nfs-ganesha/.noderefs"
	NODESTATELN="$GANSTATEDIR/$host"
	;;
esac


##################################################

usage ()
{
	_c=$(basename "$0")
	cat <<EOF
usage: $_c { shutdown | startup }
       $_c { stop | start | check } nfs
       $_c { releaseip | takeip }
       $_c { monitor-list-shares }
EOF
    exit 1
}


##################################################
# Basic service stop and start

basic_stop ()
{
	case "$1" in
	nfs)
		service "$nfs_service" stop
		;;
	*)
		usage
	esac
}

basic_start ()
{
	case "$1" in
	nfs)
		service "$nfs_service" start
		;;
	*)
		usage
	esac
}

##################################################
# "stop" and "start" options for restarting

service_stop ()
{
    case "$1" in
	nfs)
	    basic_stop "nfs"
	    ;;
	nlockmgr)
	    # Do nothing - used by statd-callout
	    :
	    ;;
	*)
	    usage
    esac
}

service_start ()
{
	case "$1" in
	nfs)
		basic_start "nfs"
		;;
	nlockmgr)
		# Do nothing - used by statd-callout
		:
		;;
	*)
		usage
	esac
}

##################################################
# Nitty gritty - monitoring and IP handling

# Check that a symlink exists, create it otherwise.
# Usage: check_ln <TARGET> <LINK>
check_ln ()
{
	if [ ! -L "${2}" ] ; then
		rm -vrf "${2}"
	else
		_t=$(readlink "${2}")
		if [ "$_t" != "${1}" ] ; then
			rm -v "${2}"
		fi
	fi
	# This is not an "else".  It also re-creates the link if it was
	# removed above!
	if [ ! -e "${2}" ]; then
		ln -sfv "${1}" "${2}"
	fi
}

# Return 'active' if the shared filesystem is accessible.
get_cluster_fs_state ()
{
	case $state_fs in
	gpfs)
		/usr/lpp/mmfs/bin/mmgetstate | awk 'NR == 4 { print $3 }'
		;;
	glusterfs)
		# Since we're past create_ganesha_recdirs(), we're active.
		echo "active"
		;;
	*)
		echo "File system $state_fs not supported"
		exit 1
		;;
	esac
}

create_ganesha_recdirs ()
{
	if ! _mounts=$(mount | grep "$state_fs"); then
		echo "Failed to find mounts of type $state_fs"
		exit 1
	fi
	if [ -z "$_mounts" ]; then
		echo "startup $state_fs not ready"
		exit 0
	fi

	case $state_fs in
	gpfs)
		_mntpt=$(echo "$_mounts" | sort | awk 'NR == 1 {print $3}')
		_link_dst="${_mntpt}/${ganesha_rec_subdir}"
		mkdir -vp "$_link_dst"
		check_ln "$_link_dst" "$GANRECDIR"
		;;
	glusterfs)
		[ -d /var/lib/nfs.backup ] || \
			mv /var/lib/nfs /var/lib/nfs.backup
		check_ln "$NODESTATEDIR" /var/lib/nfs

		mkdir -p "${NODESTATEDIR}/ganesha/v4recov"
		mkdir -p "${NODESTATEDIR}/ganesha/v4old"
		mkdir -p "${NODESTATEDIR}/statd/sm"
		mkdir -p "${NODESTATEDIR}/statd/sm.bak"
		touch "${NODESTATEDIR}/state"
		touch "${NODESTATEDIR}/statd/state"

		mkdir -p "$GANSTATEDIR"
		check_ln "$NODESTATEDIR" "$NODESTATELN"
		for _dir in "${GANSTATEDIR}/"* ; do
			# Handle no directories case
			if [ ! -d "$_dir" ] ; then
				break
			fi

			_node="${_dir##*/}" # basename
			if [ "${_node}" != "${host}" ]; then
				check_ln "${GANSTATEDIR}/${_node}/ganesha" \
					 "${NODESTATEDIR}/ganesha/${_node}"
				check_ln "${GANSTATEDIR}/${_node}/statd" \
					 "${NODESTATEDIR}/statd/${_node}"
			fi
		done
		;;
	esac
}

service_check ()
{
	create_ganesha_recdirs

	# Always succeed if cluster filesystem is not active
	_cluster_fs_state=$(get_cluster_fs_state)
	if [ "$_cluster_fs_state" != "active" ] ; then
		return 0
	fi

	# Check that NFS Ganesha is running, according to PID file
	_pidfile="/var/run/ganesha.pid"
	_ganesha="/usr/bin/ganesha.nfsd"
	if ! { read -r _pid < "$_pidfile" && \
		grep "$_ganesha" "${procfs}/${_pid}/cmdline" ; } >/dev/null 2>&1
	then

		echo "ERROR: NFS Ganesha not running according to PID file"
		return 1
	fi

	return 0
}

#-------------------------------------------------

grace_period()
{
	_arg="$1"

	if [ -x "/usr/bin/grace_period" ]; then
		/usr/bin/grace_period "$_arg"
	else
		dbus-send --print-reply --system --dest=org.ganesha.nfsd \
			/org/ganesha/nfsd/admin org.ganesha.nfsd.admin.grace \
			string:"$_arg"
	fi
}

nfs_releaseip()
{
	_ip="$2"

	# NFS-Ganesha recovery code only processes items matching $_ip
	grace_period "2:${_ip}"
}

nfs_takeip ()
{
	_ip="$2"

	case  $state_fs in
	glusterfs)
		check_ln "$NODESTATEDIR" "${GANSTATEDIR}/${_ip}"
		;;
	esac

	grace_period "5:${_ip}"
}

##################################################
# service init startup and final shutdown

nfs_shutdown ()
{
	basic_stop "nfs"
}

nfs_startup ()
{
	basic_stop "nfs" || true

	create_ganesha_recdirs

	basic_start "nfs"
	_f="${procfs}/sys/net/ipv4/tcp_tw_recycle"
	if [ -f "$_f" ] ; then
		echo 1 >"$_f"
	fi
}

##################################################
# list share directories

nfs_monitor_list_shares ()
{
	grep Path "$nfs_exports_file" |
		cut -f2 -d\" |
		sort -u
}

##################################################

nfs_register ()
{
	cat <<EOF
shutdown
startup
stop
start
check
releaseip
takeip
monitor-list-shares
EOF
}

##################################################

action="$1"
shift

case "$action" in
shutdown)            nfs_shutdown            ;;
startup)             nfs_startup             ;;
stop)                service_stop "$1"       ;;
start)               service_start "$1"      ;;
check)               service_check "$1"      ;;
releaseip)           nfs_releaseip "$@"      ;;
takeip)              nfs_takeip "$@"         ;;
monitor-list-shares) nfs_monitor_list_shares ;;
register)            nfs_register            ;;
monitor-pre|monitor-post|releaseip-pre|takeip-pre|startipreallocate)
	# Not required/implemented
	:
	;;
*)
	usage
esac