diff options
Diffstat (limited to 'heartbeat/ManageRAID.in')
| -rw-r--r-- | heartbeat/ManageRAID.in | 391 |
1 files changed, 391 insertions, 0 deletions
diff --git a/heartbeat/ManageRAID.in b/heartbeat/ManageRAID.in new file mode 100644 index 0000000..bf5c745 --- /dev/null +++ b/heartbeat/ManageRAID.in @@ -0,0 +1,391 @@ +#!@BASH_SHELL@ +# +# Name ManageRAID +# Author Matthias Dahl, m.dahl@designassembly.de +# License GPL version 2 +# +# (c) 2006 The Design Assembly GmbH. +# +# +# WARNING WARNING WARNING WARNING WARNING WARNING WARNING WARNING WARNING WARNING +# +# This resource agent is most likely function complete but not error free. Please +# consider it BETA quality for the moment until it has proven itself stable... +# +# USE AT YOUR OWN RISK. +# +# WARNING WARNING WARNING WARNING WARNING WARNING WARNING WARNING WARNING WARNING +# +# +# partly based on/inspired by original Heartbeat2 OCF resource agents +# +# Description +# +# Manages starting, mounting, unmounting, stopping and monitoring of RAID devices +# which are preconfigured in /etc/conf.d/HB-ManageRAID. +# +# +# Created 11. Sep 2006 +# Updated 18. Sep 2006 +# +# rev. 1.00.2 +# +# Changelog +# +# 18/Sep/06 1.00.1 more cleanup +# 12/Sep/06 1.00.1 add more functionality +# add sanity check for config parameters +# general cleanup all over the place +# 11/Sep/06 1.00.0 it's alive... muahaha... ALIVE... :-) +# +# +# TODO +# +# - check if at least one disk out of PREFIX_LOCALDISKS is still active +# in RAID otherwise consider RAID broken and stop it. +# +# The reason behind this: consider a RAID-1 which contains iSCSI devices +# shared over Ethernet which get dynamically added/removed to/from the RAID. +# Once all local disks have failed and only those iSCSI disks remain, the RAID +# should really stop to prevent bad performance and possible data loss. +# + +### +: ${OCF_FUNCTIONS_DIR=${OCF_ROOT}/lib/heartbeat} +. ${OCF_FUNCTIONS_DIR}/ocf-shellfuncs + +# Parameter defaults + +OCF_RESKEY_raidname_default="" + +: ${OCF_RESKEY_raidname=${OCF_RESKEY_raidname_default}} + +### + +# required utilities + +# required files/devices +RAID_MDSTAT=/proc/mdstat + +# +# check_file() +# +check_file () +{ + if [[ ! -e $1 ]]; then + ocf_log err "setup problem: file $1 does not exist." + exit $OCF_ERR_GENERIC + fi +} + +# +# usage() +# +usage() +{ + cat <<-EOT + usage: $0 {start|stop|status|monitor|validate-all|usage|meta-data} + EOT +} + +# +# meta_data() +# +meta_data() +{ + cat <<END +<?xml version="1.0"?> +<!DOCTYPE resource-agent SYSTEM "ra-api-1.dtd"> +<resource-agent name="ManageRAID" version="1.00.2"> + <version>1.0</version> + + <longdesc lang="en"> + Manages starting, stopping and monitoring of RAID devices which + are preconfigured in /etc/conf.d/HB-ManageRAID. + </longdesc> + + <shortdesc lang="en">Manages RAID devices</shortdesc> + + <parameters> + <parameter name="raidname" unique="0" required="1"> + <longdesc lang="en"> + Name (case sensitive) of RAID to manage. (preconfigured in /etc/conf.d/HB-ManageRAID) + </longdesc> + <shortdesc lang="en">RAID name</shortdesc> + <content type="string" default="${OCF_RESKEY_raidname_default}" /> + </parameter> + </parameters> + + <actions> + <action name="start" timeout="75s" /> + <action name="stop" timeout="75s" /> + <action name="status" depth="0" timeout="10s" interval="10s" /> + <action name="monitor" depth="0" timeout="10s" interval="10s" /> + <action name="validate-all" timeout="5s" /> + <action name="meta-data" timeout="5s" /> + </actions> +</resource-agent> +END +} + +# +# start_raid() +# +start_raid() +{ + declare -i retcode + + status_raid + retcode=$? + if [[ $retcode == $OCF_SUCCESS ]]; then + return $OCF_SUCCESS + elif [[ $retcode != $OCF_NOT_RUNNING ]]; then + return $retcode + fi + + for ldev in "${RAID_LOCALDISKS[@]}"; do + if [[ ! -b $ldev ]]; then + ocf_log err "$ldev is not a (local) block device." + return $OCF_ERR_ARGS + fi + done + + $MDADM -A $RAID_DEVPATH -a yes -u ${!RAID_UUID} "${RAID_LOCALDISKS[@]}" &> /dev/null + if [[ $? != 0 ]]; then + ocf_log err "starting ${!RAID_DEV} with ${RAID_LOCALDISKS[*]} failed." + return $OCF_ERR_GENERIC + fi + + $MOUNT -o ${!RAID_MOUNTOPTIONS} $RAID_DEVPATH ${!RAID_MOUNTPOINT} &> /dev/null + if [[ $? != 0 ]]; then + $MDADM -S $RAID_DEVPATH &> /dev/null + + if [[ $? != 0 ]]; then + ocf_log err "mounting ${!RAID_DEV} to ${!RAID_MOUNTPOINT} failed as well as stopping the RAID itself." + else + ocf_log err "mounting ${!RAID_DEV} to ${!RAID_MOUNTPOINT} failed. RAID stopped again." + fi + + return $OCF_ERR_GENERIC + fi + + return $OCF_SUCCESS +} + +# +# stop_raid() +# +stop_raid() +{ + status_raid + if [[ $? == $OCF_NOT_RUNNING ]]; then + return $OCF_SUCCESS + fi + + $UMOUNT ${!RAID_MOUNTPOINT} &> /dev/null + if [[ $? != 0 ]]; then + ocf_log err "unmounting ${!RAID_MOUNTPOINT} failed. not stopping ${!RAID_DEV}!" + return $OCF_ERR_GENERIC + fi + + $MDADM -S $RAID_DEVPATH &> /dev/null + if [[ $? != 0 ]]; then + ocf_log err "stopping RAID ${!RAID_DEV} failed." + return $OCF_ERR_GENERIC + fi + + return $OCF_SUCCESS +} + +# +# status_raid() +# +status_raid() +{ + declare -i retcode_raidcheck + declare -i retcode_uuidcheck + + $CAT $RAID_MDSTAT | $GREP -e "${!RAID_DEV}[\ ]*:[\ ]*active" &> /dev/null + if [ $? -ne 0 ]; then + return $OCF_NOT_RUNNING + fi + + if [ ! -e $RAID_DEVPATH ]; then + return $OCF_ERR_GENERIC + fi + + $MDADM --detail -t $RAID_DEVPATH &> /dev/null + retcode_raidcheck=$? + $MDADM --detail -t $RAID_DEVPATH | $GREP -qEe "^[\ ]*UUID[\ ]*:[\ ]*${!RAID_UUID}" &> /dev/null + retcode_uuidcheck=$? + + if [ $retcode_raidcheck -gt 3 ]; then + ocf_log err "mdadm returned error code $retcode_raidcheck while checking ${!RAID_DEV}." + return $OCF_ERR_GENERIC + elif [ $retcode_raidcheck -eq 3 ]; then + ocf_log err "${!RAID_DEV} has failed." + return $OCF_ERR_GENERIC + elif [ $retcode_raidcheck -lt 3 ] && [ $retcode_uuidcheck != 0 ]; then + ocf_log err "active RAID ${!RAID_DEV} and configured UUID (!$RAID_UUID) do not match." + return $OCF_ERR_GENERIC + fi + + $MOUNT | $GREP -e "$RAID_DEVPATH on ${!RAID_MOUNTPOINT}" &> /dev/null + if [[ $? != 0 ]]; then + ocf_log err "${!RAID_DEV} seems to be no longer mounted at ${!RAID_MOUNTPOINT}" + return $OCF_ERR_GENERIC + fi + + return $OCF_SUCCESS +} + +# +# validate_all_raid() +# +validate_all_raid() +{ + # + # since all parameters are checked every time ManageRAID is + # invoked, there not much more to check... + # + # status_raid should cover the rest. + # + declare -i retcode + + status_ve + retcode=$? + + if [[ $retcode != $OCF_SUCCESS && $retcode != $OCF_NOT_RUNNING ]]; then + return $retcode + fi + + 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 + +## required configuration +# +[ -f /etc/conf.d/HB-ManageRAID ] || { + ocf_log err "/etc/conf.d/HB-ManageRAID missing" + exit $OCF_ERR_INSTALLED +} +. /etc/conf.d/HB-ManageRAID +# +## + +# +# check relevant environment variables for sanity and security +# + +declare -i retcode_test +declare -i retcode_grep + +$TEST -z "$OCF_RESKEY_raidname" +retcode_test=$? +echo "$OCF_RESKEY_raidname" | $GREP -qEe "^[[:alnum:]\_]+$" +retcode_grep=$? + +if [[ $retcode_test != 1 || $retcode_grep != 0 ]]; then + ocf_log err "OCF_RESKEY_raidname not set or invalid." + exit $OCF_ERR_ARGS +fi + +RAID_UUID=${OCF_RESKEY_raidname}_UUID + +echo ${!RAID_UUID} | $GREP -qEe "^[[:alnum:]]{8}:[[:alnum:]]{8}:[[:alnum:]]{8}:[[:alnum:]]{8}$" +if [[ $? != 0 ]]; then + ocf_log err "${OCF_RESKEY_raidname}_UUID is invalid." + exit $OCF_ERR_ARGS +fi + +RAID_DEV=${OCF_RESKEY_raidname}_DEV + +echo ${!RAID_DEV} | $GREP -qEe "^md[0-9]+$" +if [[ $? != 0 ]]; then + ocf_log err "${OCF_RESKEY_raidname}_DEV is invalid." + exit $OCF_ERR_ARGS +fi + +RAID_DEVPATH=/dev/${!RAID_DEV/md/md\/} +RAID_MOUNTPOINT=${OCF_RESKEY_raidname}_MOUNTPOINT + +echo ${!RAID_MOUNTPOINT} | $GREP -qEe "^[[:alnum:]\/\_\"\ ]+$" +if [[ $? != 0 ]]; then + ocf_log err "${OCF_RESKEY_raidname}_MOUNTPOINT is invalid." + exit $OCF_ERR_ARGS +fi + +RAID_MOUNTOPTIONS=${OCF_RESKEY_raidname}_MOUNTOPTIONS + +echo ${!RAID_MOUNTOPTIONS} | $GREP -qEe "^[[:alpha:]\,]+$" +if [[ $? != 0 ]]; then + ocf_log err "${OCF_RESKEY_raidname}_MOUNTOPTIONS is invalid." + exit $OCF_ERR_ARGS +fi + +RAID_LOCALDISKS=${OCF_RESKEY_raidname}_LOCALDISKS[@] +RAID_LOCALDISKS=( "${!RAID_LOCALDISKS}" ) + +if [ ${#RAID_LOCALDISKS[@]} -lt 1 ]; then + ocf_log err "you have to specify at least one local disk." + exit $OCF_ERR_ARGS +fi + +# +# check that all relevant utilities are available +# +check_binary $MDADM +check_binary $MOUNT +check_binary $UMOUNT +check_binary $GREP +check_binary $CAT +check_binary $TEST +check_binary echo + + +# +# check that all relevant devices are available +# +check_file $RAID_MDSTAT + +# +# finally... let's see what we are ordered to do :-) +# +case "$1" in + start) + start_raid + ;; + stop) + stop_raid + ;; + status|monitor) + status_raid + ;; + validate-all) + validate_all_raid + ;; + *) + usage + exit $OCF_ERR_UNIMPLEMENTED + ;; +esac + +exit $? + |