cluster-agents 1:1.0.4-0ubuntu2 / usr / lib / ocf / resource.d / heartbeat / Raid1

#!/bin/sh
#
#
# License:      GNU General Public License (GPL)
# Support:      linux-ha@lists.linux-ha.org
# 
# Raid1
#      Description: Manages a software Raid1 device on a shared storage medium.
#  Original Author: Eric Z. Ayers (eric.ayers@compgen.com)
# Original Release: 25 Oct 2000
#     RAID patches: http://people.redhat.com/mingo/raid-patches/
# Word to the Wise: http://lwn.net/2000/0810/a/raid-faq.php3
#  Sympathetic Ear: mailto:linux-raid@vger.kernel.org
#
# usage: $0 {start|stop|status|monitor|validate-all|usage|meta-data}
#
#
# EXAMPLE config file /etc/raidtab.md0
# This file must exist on both machines!
#
#  raiddev		    /dev/md0
#  raid-level		    1
#  nr-raid-disks	    2
#  chunk-size		    64k
#  persistent-superblock    1
#  #nr-spare-disks	    0
#    device	    /dev/sda1
#    raid-disk     0
#    device	    /dev/sdb1
#    raid-disk     1
#
# EXAMPLE config file /etc/mdadm.conf (for more info:man mdadm.conf)
#
#  DEVICE /dev/sdb1 /dev/sdc1 
#  ARRAY /dev/md0 UUID=4a865b55:ba27ef8d:29cd5701:6fb42799
#######################################################################
# Initialization:

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

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

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

meta_data() {
	cat <<END
<?xml version="1.0"?>
<!DOCTYPE resource-agent SYSTEM "ra-api-1.dtd">
<resource-agent name="Raid1">
<version>1.0</version>

<longdesc lang="en">
Resource script for RAID1. It manages a software Raid1 device on a shared 
storage medium. 
</longdesc>
<shortdesc lang="en">Manages a software RAID1 device on shared storage</shortdesc>

<parameters>
<parameter name="raidconf" unique="0" required="1">
<longdesc lang="en">
The RAID configuration file. e.g. /etc/raidtab or /etc/mdadm.conf.
</longdesc>
<shortdesc lang="en">RAID config file</shortdesc>
<content type="string" default="" />
</parameter>

<parameter name="raiddev" unique="0" required="1">
<longdesc lang="en">
The block device to use.
</longdesc>
<shortdesc lang="en">block device</shortdesc>
<content type="string" default="" />
</parameter>

<parameter name="homehost" unique="0" required="0">
<longdesc lang="en">
The value for the homehost directive; this is an mdadm feature to
protect RAIDs against being activated by accident. It is recommended to
create RAIDs managed by the cluster with "homehost" set to a special
value, so they are not accidentially auto-assembled by nodes not
supposed to own them.
</longdesc>
<shortdesc lang="en">Homehost for mdadm</shortdesc>
<content type="string" default="" />
</parameter>
</parameters>

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

#
# START: Start up the RAID device
#
raid1_start() {
	raid1_monitor
	rc=$?
	if [ $rc -eq $OCF_SUCCESS ]; then
		# md already online, nothing to do.
		return $OCF_SUCCESS
	fi
	if [ $rc -ne $OCF_NOT_RUNNING ]; then
		# If the array is in a broken state, this agent doesn't
		# know how to repair that.
		ocf_log err "$MDDEV in a broken state; cannot start (rc=$rc)"
		return $OCF_ERR_GENERIC
	fi

	# Insert SCSI module
	$MODPROBE scsi_hostadapter
	if [ $? -ne 0 ] ; then
	    ocf_log warn "Couldn't insert SCSI module."
	fi

	# Insert raid personality module
	$MODPROBE raid1
	if [ $? -ne 0 ] ; then
	# It is not fatal, chance is that we have raid1 builtin...
	    ocf_log warn "Couldn't insert RAID1 module"
	fi
	grep -q "^Personalities.*\[raid1\]" /proc/mdstat 2>/dev/null
	if [ $? -ne 0 ] ; then
	    ocf_log err "We don't have RAID1 support! Exiting"
	    return $OCF_ERR_GENERIC
	fi

	if [ $HAVE_RAIDTOOLS = "true" ]; then
	    # Run raidstart to start up the RAID array
	    $RAIDSTART --configfile $RAIDCONF $MDDEV
	else
	    # Run mdadm
	    $MDADM --assemble $MDDEV --config=$RAIDCONF $MDADM_HOMEHOST
	fi

	raid1_monitor
	if [ $? -eq $OCF_SUCCESS ]; then
	    return $OCF_SUCCESS
	else
	    ocf_log err "Couldn't start RAID for $MDDEV"
	    return $OCF_ERR_GENERIC
	fi
}

#
# STOP: stop the RAID device
#
raid1_stop() {
	# See if the MD device is already cleanly stopped:
	raid1_monitor
	if [ $? -eq $OCF_NOT_RUNNING ]; then
		return $OCF_SUCCESS
	fi

	# Turn off raid
	if [ $HAVE_RAIDTOOLS = "true" ]; then
		$RAIDSTOP --configfile $RAIDCONF $MDDEV
	else
		ocf_log info "Stopping array $MDDEV"
		$MDADM --stop $MDDEV --config=$RAIDCONF --wait-clean -W
	fi
	rc=$?

	if [ $rc -ne 0 ]; then
		ocf_log err "Couldn't stop RAID for $MDDEV (rc=$rc)"
		if [ $HAVE_RAIDTOOLS != "true" ]; then
			ocf_log info "Attempting to mark array $MDDEV readonly"
			$MDADM --readonly $MDDEV --config=$RAIDCONF
			rc=$?
			if [ $rc -ne 0 ]; then
				ocf_log err "Failed to set $MDDEV readonly (rc=$rc)"
			fi
		fi
		return $OCF_ERR_GENERIC
	fi

	raid1_monitor
	if [ $? -eq $OCF_NOT_RUNNING ]; then
		return $OCF_SUCCESS
	fi
	
	ocf_log err "RAID $MDDEV still active after stop command!"
	return $OCF_ERR_GENERIC
}

#
# monitor: a less noisy status
#
raid1_monitor() {
	TRY_READD=0
	# check if the md device exists first
	if [ ! -b $MDDEV ]; then
		ocf_log info "$MDDEV is not a block device"
		return $OCF_NOT_RUNNING
	fi
	if ! grep -e "^$MD[ \t:]" /proc/mdstat >/dev/null ; then
		ocf_log info "$MD not found in /proc/mdstat"
		return $OCF_NOT_RUNNING
	fi
	if [ $HAVE_RAIDTOOLS != "true" ]; then
		$MDADM --detail --test $MDDEV >/dev/null 2>&1 ; rc=$?
		case $rc in
		0)	;;
		1)	ocf_log warn "$MDDEV has at least one failed device."
			TRY_READD=1
			;;
		2)	ocf_log err "$MDDEV has failed."
			return $OCF_ERR_GENERIC
			;;
		4)	ocf_log err "mdadm failed on $MDDEV."
			return $OCF_ERR_GENERIC
			;;
		*)	ocf_log err "mdadm returned an unknown result ($rc)."
			return $OCF_ERR_GENERIC
			;;
		esac
	fi
	if [ "$__OCF_ACTION" = "monitor" -a "$OCF_RESKEY_CRM_meta_interval" != 0 \
		-a $TRY_READD -eq 1 -a $OCF_CHECK_LEVEL -gt 0 ]; then
		ocf_log info "Attempting recovery sequence to re-add devices on $MDDEV:"
		$MDADM $MDDEV --fail detached 
		$MDADM $MDDEV --remove failed 
		$MDADM $MDDEV --re-add missing
		# TODO: At this stage, there's nothing to actually do
		# here. Either this worked or it did not.
	fi

	if ! dd if=$MDDEV count=1 bs=512 of=/dev/null \
	     iflag=direct >/dev/null 2>&1 ; then
		ocf_log err "$MDDEV: I/O error on read"
		return $OCF_ERR_GENERIC
	fi

	return $OCF_SUCCESS
}

#
# STATUS: is the raid device online or offline?
#
raid1_status() {
	# See if the MD device is online
	raid1_monitor
	rc=$?
	if [ $rc -ne $OCF_SUCCESS ]; then
	    echo "stopped"
	else
	    echo "running"
	fi
	return $rc
}

raid1_validate_all() {
    return $OCF_SUCCESS
}
	

if
  ( [ $# -ne 1 ] )
then
  usage
  exit $OCF_ERR_ARGS
fi

case "$1" in
  meta-data)
	meta_data
	exit $OCF_SUCCESS
	;;
  usage) 
	usage
	exit $OCF_SUCCESS
	;;
  *)
	;;
esac

RAIDCONF="$OCF_RESKEY_raidconf"
MDDEV="$OCF_RESKEY_raiddev"

if [ -z "$RAIDCONF" ] ; then
	ocf_log err "Please set OCF_RESKEY_raidconf!"
	exit $OCF_ERR_ARGS
fi

if [ ! -r "$RAIDCONF" ] ; then
	ocf_log err "Configuration file [$RAIDCONF] does not exist, or can not be opend!"
	exit $OCF_ERR_ARGS
fi

if [ -z "$MDDEV" ] ; then
	ocf_log err "Please set OCF_RESKEY_raiddev to the Raid device you want to control!"
	exit $OCF_ERR_ARGS
fi

if [ -e "$MDDEV" -a ! -b "$MDDEV" ] ; then
	ocf_log err "$MDDEV is not a block device!"
	exit $OCF_ERR_ARGS
fi

# strip off the /dev/ prefix to get the name of the MD device
MD=`echo $MDDEV | sed -e 's/\/dev\///'`


HAVE_RAIDTOOLS=false
if have_binary $MDADM >/dev/null 2>&1 ; then
  if [ -n "$OCF_RESKEY_homehost" ]; then
  	MDADM_HOMEHOST="--homehost=${OCF_RESKEY_homehost}"
  else
  	MDADM_HOMEHOST=""
  fi
else
  check_binary $RAIDSTART
  HAVE_RAIDTOOLS=true
fi

# At this stage,
# [ $HAVE_RAIDTOOLS = false ] <=> we have $MDADM,
# otherwise we have raidtools (raidstart and raidstop)

# Look for how we are called 
case "$1" in
  start)
	raid1_start
	;;
  stop)
	raid1_stop
	;;
  status) 
	raid1_status
	;;
  monitor) 
	raid1_monitor
	;;
  validate-all)
	raid1_validate_all
	;;
  *)
	usage
	exit $OCF_ERR_UNIMPLEMENTED 
	;;
esac

exit $?