diff --git a/heartbeat/galera.in b/heartbeat/galera.in index 6aed3e4b6..b518595cb 100755 --- a/heartbeat/galera.in +++ b/heartbeat/galera.in @@ -1,1097 +1,1106 @@ #!@BASH_SHELL@ # # Copyright (c) 2014 David Vossel # All Rights Reserved. # # 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. # ## # README. # # This agent only supports being configured as a multistate Promoted # resource. # # Unpromoted vs Promoted role: # # During the 'Unpromoted' role, galera instances are in read-only mode and # will not attempt to connect to the cluster. This role exists only as # a means to determine which galera instance is the most up-to-date. The # most up-to-date node will be used to bootstrap a galera cluster that # has no current members. # # The galera instances will only begin to be promoted to the Promoted role # once all the nodes in the 'wsrep_cluster_address' connection address # have entered read-only mode. At that point the node containing the # database that is most current will be promoted to Promoted. Once the first # Promoted instance bootstraps the galera cluster, the other nodes will be # promoted to Promoted as well. # # Example: Create a galera cluster using nodes rhel7-node1 rhel7-node2 rhel7-node3 # # pcs resource create db galera enable_creation=true \ # wsrep_cluster_address="gcomm://rhel7-auto1,rhel7-auto2,rhel7-auto3" meta promoted-max=3 --promoted # # By setting the 'enable_creation' option, the database will be automatically # generated at startup. The meta attribute 'promoted-max=3' means that all 3 # nodes listed in the wsrep_cluster_address list will be allowed to connect # to the galera cluster and perform replication. # # NOTE: If you have more nodes in the pacemaker cluster then you wish # to have in the galera cluster, make sure to use location contraints to prevent # pacemaker from attempting to place a galera instance on a node that is # not in the 'wsrep_cluster_address" list. # ## ####################################################################### # Initialization: : ${OCF_FUNCTIONS_DIR=${OCF_ROOT}/lib/heartbeat} . ${OCF_FUNCTIONS_DIR}/ocf-shellfuncs if [ "$__OCF_ACTION" != "meta-data" ]; then . ${OCF_FUNCTIONS_DIR}/mysql-common.sh NODENAME=$(ocf_attribute_target) fi # It is common for some galera instances to store # check user that can be used to query status # in this file if [ -f "/etc/sysconfig/clustercheck" ]; then . /etc/sysconfig/clustercheck elif [ -f "/etc/default/clustercheck" ]; then . /etc/default/clustercheck fi # Parameter defaults OCF_RESKEY_wsrep_cluster_address_default="" OCF_RESKEY_cluster_host_map_default="" OCF_RESKEY_check_user_default="" OCF_RESKEY_check_passwd_default="" OCF_RESKEY_two_node_mode_default="false" : ${OCF_RESKEY_wsrep_cluster_address=${OCF_RESKEY_wsrep_cluster_address_default}} : ${OCF_RESKEY_cluster_host_map=${OCF_RESKEY_cluster_host_map_default}} : ${OCF_RESKEY_check_user=${OCF_RESKEY_check_user_default}} : ${OCF_RESKEY_check_passwd=${OCF_RESKEY_check_passwd_default}} : ${OCF_RESKEY_two_node_mode=${OCF_RESKEY_two_node_mode_default}} ####################################################################### # Defaults: OCF_RESKEY_check_passwd_use_empty_default=0 : ${OCF_RESKEY_check_passwd_use_empty=${OCF_RESKEY_check_passwd_use_empty_default}} ####################################################################### usage() { cat < 1.0 Resource script for managing galera database. Manages a galera instance Location of the MySQL server binary MySQL server binary Location of the MySQL client binary MySQL client binary Configuration file MySQL config Directory containing databases MySQL datadir User running MySQL daemon MySQL user Group running MySQL daemon (for logfile and directory permissions) MySQL group The logfile to be used for mysqld. MySQL log file The pidfile to be used for mysqld. MySQL pid file The socket to be used for mysqld. MySQL socket If the MySQL database does not exist, it will be created Create the database if it does not exist Additional parameters which are passed to the mysqld on startup. (e.g. --skip-external-locking or --skip-grant-tables) Additional parameters to pass to mysqld The galera cluster address. This takes the form of: gcomm://node,node,node Only nodes present in this node list will be allowed to start a galera instance. The galera node names listed in this address are expected to match valid pacemaker node names. If both names need to differ, you must provide a mapping in option cluster_host_map. Galera cluster address A mapping of pacemaker node names to galera node names. To be used when both pacemaker and galera names need to differ, (e.g. when galera names map to IP from a specific network interface) This takes the form of: pcmk1:node.1.galera;pcmk2:node.2.galera;pcmk3:node.3.galera where the galera resource started on node pcmk1 would be named node.1.galera in the wsrep_cluster_address Pacemaker to Galera name mapping Cluster check user. MySQL test user Cluster check user password. Empty passwords are ignored unless the parameter "check_passwd_use_empty" is set to 1. check password Use an empty "check_passwd" password. If this parameter is set to 1, "check_passwd" will be ignored and an empty password is used when calling the "mysql" client binary. check password use empty If running in a 2-node pacemaker cluster, rely on pacemaker quorum to allow automatic recovery even when the other node is unreachable. Use it with caution! (and fencing) Special recovery when running on a 2-node cluster END } get_option_variable() { local key=$1 $MYSQL $MYSQL_OPTIONS_CHECK -e "SHOW VARIABLES like '$key';" | tail -1 } get_status_variable() { local key=$1 $MYSQL $MYSQL_OPTIONS_CHECK -e "show status like '$key';" | tail -1 } set_bootstrap_node() { local node=$(ocf_attribute_target $1) ${HA_SBIN_DIR}/crm_attribute -N $node -l reboot --name "${INSTANCE_ATTR_NAME}-bootstrap" -v "true" } clear_bootstrap_node() { ${HA_SBIN_DIR}/crm_attribute -N $NODENAME -l reboot --name "${INSTANCE_ATTR_NAME}-bootstrap" -D } is_bootstrap() { ${HA_SBIN_DIR}/crm_attribute -N $NODENAME -l reboot --name "${INSTANCE_ATTR_NAME}-bootstrap" --quiet 2>/dev/null } set_no_grastate() { ${HA_SBIN_DIR}/crm_attribute -N $NODENAME -l reboot --name "${INSTANCE_ATTR_NAME}-no-grastate" -v "true" } clear_no_grastate() { ${HA_SBIN_DIR}/crm_attribute -N $NODENAME -l reboot --name "${INSTANCE_ATTR_NAME}-no-grastate" -D } is_no_grastate() { local node=$(ocf_attribute_target $1) ${HA_SBIN_DIR}/crm_attribute -N $node -l reboot --name "${INSTANCE_ATTR_NAME}-no-grastate" --quiet 2>/dev/null } clear_last_commit() { ${HA_SBIN_DIR}/crm_attribute -N $NODENAME -l reboot --name "${INSTANCE_ATTR_NAME}-last-committed" -D } set_last_commit() { ${HA_SBIN_DIR}/crm_attribute -N $NODENAME -l reboot --name "${INSTANCE_ATTR_NAME}-last-committed" -v $1 } get_last_commit() { local node=$(ocf_attribute_target $1) if [ -z "$node" ]; then ${HA_SBIN_DIR}/crm_attribute -N $NODENAME -l reboot --name "${INSTANCE_ATTR_NAME}-last-committed" --quiet 2>/dev/null else ${HA_SBIN_DIR}/crm_attribute -N $node -l reboot --name "${INSTANCE_ATTR_NAME}-last-committed" --quiet 2>/dev/null fi } clear_safe_to_bootstrap() { ${HA_SBIN_DIR}/crm_attribute -N $NODENAME -l reboot --name "${INSTANCE_ATTR_NAME}-safe-to-bootstrap" -D } set_safe_to_bootstrap() { ${HA_SBIN_DIR}/crm_attribute -N $NODENAME -l reboot --name "${INSTANCE_ATTR_NAME}-safe-to-bootstrap" -v $1 } get_safe_to_bootstrap() { local node=$(ocf_attribute_target $1) if [ -z "$node" ]; then ${HA_SBIN_DIR}/crm_attribute -N $NODENAME -l reboot --name "${INSTANCE_ATTR_NAME}-safe-to-bootstrap" --quiet 2>/dev/null else ${HA_SBIN_DIR}/crm_attribute -N $node -l reboot --name "${INSTANCE_ATTR_NAME}-safe-to-bootstrap" --quiet 2>/dev/null fi } wait_for_sync() { local state=$(get_status_variable "wsrep_local_state") ocf_log info "Waiting for database to sync with the cluster. " while [ "$state" != "4" ]; do sleep 1 state=$(get_status_variable "wsrep_local_state") done ocf_log info "Database synced." } is_primary() { cluster_status=$(get_status_variable "wsrep_cluster_status") if [ "$cluster_status" = "Primary" ]; then return 0 fi if [ -z "$cluster_status" ]; then ocf_exit_reason "Unable to retrieve wsrep_cluster_status, verify check_user '$OCF_RESKEY_check_user' has permissions to view status" else ocf_log info "Galera instance wsrep_cluster_status=${cluster_status}" fi return 1 } is_readonly() { local res=$(get_option_variable "read_only") if ! ocf_is_true "$res"; then return 1 fi cluster_status=$(get_status_variable "wsrep_cluster_status") if ! [ "$cluster_status" = "Disconnected" ]; then return 1 fi return 0 } is_two_node_mode_active() { # crm_node or corosync-quorumtool cannot access various corosync # flags when running inside a bundle, so only count the cluster # members ocf_is_true "$OCF_RESKEY_two_node_mode" && crm_mon_no_validation -1X | xmllint --xpath "count(//nodes/node[@type='member'])" - | grep -q -w 2 } is_last_node_in_quorate_partition() { # when a network split occurs in a 2-node cluster, pacemaker # fences the other node and try to retain quorum. So until # the fencing is resolved (and the status of the peer node # is clean), we shouldn't consider ourself quorate. local partition_members=$(${HA_SBIN_DIR}/crm_node -p | wc -w) local quorate=$(${HA_SBIN_DIR}/crm_node -q) local clean_members=$(crm_mon_no_validation -1X | xmllint --xpath 'count(//nodes/node[@type="member" and @unclean="false"])' -) [ "$partition_members" = 1 ] && [ "$quorate" = 1 ] && [ "$clean_members" = 2 ] } master_exists() { if [ "$__OCF_ACTION" = "demote" ]; then # We don't want to detect master instances during demote. # 1. we could be detecting ourselves as being master, which is no longer the case. # 2. we could be detecting other master instances that are in the process of shutting down. # by not detecting other master instances in "demote" we are deferring this check # to the next recurring monitor operation which will be much more accurate return 1 fi # determine if a master instance is already up and is healthy ocf_version_cmp "$OCF_RESKEY_crm_feature_set" "3.1.0" res=$? if [ -z "$OCF_RESKEY_crm_feature_set" ] || [ $res -eq 2 ]; then XMLOPT="--output-as=xml" ocf_version_cmp "$OCF_RESKEY_crm_feature_set" "3.2.0" if [ $? -eq 1 ]; then crm_mon_no_validation -1 $XMLOPT >/dev/null 2>&1 if [ $? -ne 0 ]; then XMLOPT="--as-xml" fi fi else XMLOPT="--as-xml" fi crm_mon_no_validation -1 $XMLOPT | grep -q -i -E "resource.*id=\"${INSTANCE_ATTR_NAME}\".*role=\"(Promoted|Master)\".*active=\"true\".*orphaned=\"false\".*failed=\"false\"" return $? } clear_master_score() { local node=$(ocf_attribute_target $1) if [ -z "$node" ]; then ocf_promotion_score -D else ocf_promotion_score -D -N $node fi } set_master_score() { local node=$(ocf_attribute_target $1) if [ -z "$node" ]; then ocf_promotion_score -v 100 else ocf_promotion_score -N $node -v 100 fi } promote_everyone() { for node in $(echo "$OCF_RESKEY_wsrep_cluster_address" | sed 's/gcomm:\/\///g' | tr -d ' ' | tr -s ',' ' '); do local pcmk_node=$(galera_to_pcmk_name $node) if [ -z "$pcmk_node" ]; then ocf_log err "Could not determine pacemaker node from galera name <${node}>." return else node=$pcmk_node fi set_master_score $node done } greater_than_equal_long() { # there are values we need to compare in this script # that are too large for shell -gt to process echo | awk -v n1="$1" -v n2="$2" '{if (n1>=n2) printf ("true"); else printf ("false");}' | grep -q "true" } galera_to_pcmk_name() { local galera=$1 if [ -z "$OCF_RESKEY_cluster_host_map" ]; then echo $galera else echo "$OCF_RESKEY_cluster_host_map" | tr ';' '\n' | tr -d ' ' | sed 's/:/ /' | awk -F' ' '$2=="'"$galera"'" {print $1;exit}' fi } pcmk_to_galera_name() { local pcmk=$1 if [ -z "$OCF_RESKEY_cluster_host_map" ]; then echo $pcmk else echo "$OCF_RESKEY_cluster_host_map" | tr ';' '\n' | tr -d ' ' | sed 's/:/ /' | awk -F' ' '$1=="'"$pcmk"'" {print $2;exit}' fi } detect_first_master() { local best_commit=0 local last_commit=0 local missing_nodes=0 local nodes="" local nodes_recovered="" local all_nodes local best_node_gcomm local best_node local safe_to_bootstrap all_nodes=$(echo "$OCF_RESKEY_wsrep_cluster_address" | sed 's/gcomm:\/\///g' | tr -d ' ' | tr -s ',' ' ') best_node_gcomm=$(echo "$all_nodes" | sed 's/^.* \(.*\)$/\1/') best_node=$(galera_to_pcmk_name $best_node_gcomm) if [ -z "$best_node" ]; then ocf_log err "Could not determine initial best node from galera name <${best_node_gcomm}>." return fi # avoid selecting a recovered node as bootstrap if possible for node in $all_nodes; do local pcmk_node=$(galera_to_pcmk_name $node) if [ -z "$pcmk_node" ]; then ocf_log err "Could not determine pacemaker node from galera name <${node}>." return else node=$pcmk_node fi if is_no_grastate $node; then nodes_recovered="$nodes_recovered $node" else nodes="$nodes $node" fi done for node in $nodes_recovered $nodes; do # On clean shutdown, galera sets the last stopped node as 'safe to bootstrap', # so use this hint when we can safe_to_bootstrap=$(get_safe_to_bootstrap $node) # Special case for 2-node clusters: during a network split, rely on # pacemaker's quorum to check whether we can restart galera if [ "$safe_to_bootstrap" != "1" ] && [ "$node" = "$NODENAME" ] && is_two_node_mode_active; then is_last_node_in_quorate_partition if [ $? -eq 0 ]; then ocf_log warn "Survived a split in a 2-node cluster, considering ourselves safe to bootstrap" safe_to_bootstrap=1 fi fi if [ "$safe_to_bootstrap" = "1" ]; then # Galera marked the node as safe to boostrap during shutdown. Let's just # pick it as our bootstrap node. ocf_log info "Node <${node}> is marked as safe to bootstrap." best_node=$node # We don't need to wait for the other nodes to report state in this case missing_nodes=0 break fi last_commit=$(get_last_commit $node) if [ -z "$last_commit" ]; then ocf_log info "Waiting on node <${node}> to report database status before Master instances can start." missing_nodes=1 continue fi # this means -1, or that no commit has occured yet. if [ "$last_commit" = "18446744073709551615" ]; then last_commit="0" fi greater_than_equal_long "$last_commit" "$best_commit" if [ $? -eq 0 ]; then best_node=$(ocf_attribute_target $node) best_commit=$last_commit fi done if [ $missing_nodes -eq 1 ]; then return fi ocf_log info "Promoting $best_node to be our bootstrap node" set_bootstrap_node $best_node set_master_score $best_node } detect_safe_to_bootstrap() { local safe_to_bootstrap="" local uuid="" local seqno="" if [ -f ${OCF_RESKEY_datadir}/grastate.dat ]; then ocf_log info "attempting to read safe_to_bootstrap flag from ${OCF_RESKEY_datadir}/grastate.dat" safe_to_bootstrap=$(sed -n 's/^safe_to_bootstrap:\s*\(.*\)$/\1/p' < ${OCF_RESKEY_datadir}/grastate.dat) uuid=$(sed -n 's/^uuid:\s*\(.*\)$/\1/p' < ${OCF_RESKEY_datadir}/grastate.dat) seqno=$(sed -n 's/^seqno:\s*\(.*\)$/\1/p' < ${OCF_RESKEY_datadir}/grastate.dat) fi if [ -z "$uuid" ] || \ [ "$uuid" = "00000000-0000-0000-0000-000000000000" ]; then clear_safe_to_bootstrap return fi if [ "$safe_to_bootstrap" = "1" ]; then if [ -z "$seqno" ] || [ "$seqno" = "-1" ]; then clear_safe_to_bootstrap return fi fi if [ "$safe_to_bootstrap" = "1" ] || [ "$safe_to_bootstrap" = "0" ]; then set_safe_to_bootstrap $safe_to_bootstrap else clear_safe_to_bootstrap fi } detect_last_commit() { local last_commit local recover_args="--defaults-file=$OCF_RESKEY_config \ --pid-file=$OCF_RESKEY_pid \ --socket=$OCF_RESKEY_socket \ --datadir=$OCF_RESKEY_datadir" local recovery_file_regex='s/.*WSREP\:.*position\s*recovery.*--log_error='\''\([^'\'']*\)'\''.*/\1/p' local recovered_position_regex='s/.*WSREP\:\s*[R|r]ecovered\s*position.*\:\(.*\)\s*$/\1/p' # codership/galera#354 # Some ungraceful shutdowns can leave an empty gvwstate.dat on # disk. This will prevent galera to join the cluster if it is # configured to attempt PC recovery. Removing that file makes the # node fall back to the normal, unoptimized joining process. if [ -f ${OCF_RESKEY_datadir}/gvwstate.dat ] && \ [ ! -s ${OCF_RESKEY_datadir}/gvwstate.dat ]; then ocf_log warn "empty ${OCF_RESKEY_datadir}/gvwstate.dat detected, removing it to prevent PC recovery failure at next restart" rm -f ${OCF_RESKEY_datadir}/gvwstate.dat fi ocf_log info "attempting to detect last commit version by reading ${OCF_RESKEY_datadir}/grastate.dat" last_commit="$(cat ${OCF_RESKEY_datadir}/grastate.dat | sed -n 's/^seqno.\s*\(.*\)\s*$/\1/p')" if [ -z "$last_commit" ] || [ "$last_commit" = "-1" ]; then local tmp=$(mktemp) chown $OCF_RESKEY_user:$OCF_RESKEY_group $tmp # if we pass here because grastate.dat doesn't exist, # try not to bootstrap from this node if possible if [ ! -f ${OCF_RESKEY_datadir}/grastate.dat ]; then set_no_grastate fi ocf_log info "now attempting to detect last commit version using 'mysqld_safe --wsrep-recover'" $SU - $OCF_RESKEY_user -s /bin/sh -c \ "${OCF_RESKEY_binary} $recover_args --wsrep-recover --log-error=$tmp 2>/dev/null" last_commit="$(cat $tmp | sed -n $recovered_position_regex | tail -1)" if [ -z "$last_commit" ]; then # Galera uses InnoDB's 2pc transactions internally. If # server was stopped in the middle of a replication, the # recovery may find a "prepared" XA transaction in the # redo log, and mysql won't recover automatically local recovery_file="$(cat $tmp | sed -n $recovery_file_regex)" if [ -e $recovery_file ]; then cat $recovery_file | grep -q -E '\[ERROR\]\s+Found\s+[0-9]+\s+prepared\s+transactions!' 2>/dev/null if [ $? -eq 0 ]; then # we can only rollback the transaction, but that's OK # since the DB will get resynchronized anyway ocf_log warn "local node <${NODENAME}> was not shutdown properly. Rollback stuck transaction with --tc-heuristic-recover" $SU - $OCF_RESKEY_user -s /bin/sh -c \ "${OCF_RESKEY_binary} $recover_args --wsrep-recover \ --tc-heuristic-recover=rollback --log-error=$tmp 2>/dev/null" last_commit="$(cat $tmp | sed -n $recovered_position_regex | tail -1)" if [ ! -z "$last_commit" ]; then ocf_log warn "State recovered. force SST at next restart for full resynchronization" rm -f ${OCF_RESKEY_datadir}/grastate.dat # try not to bootstrap from this node if possible set_no_grastate fi fi fi fi rm -f $tmp fi if [ ! -z "$last_commit" ]; then ocf_log info "Last commit version found: $last_commit" set_last_commit $last_commit return $OCF_SUCCESS else ocf_exit_reason "Unable to detect last known write sequence number" clear_last_commit return $OCF_ERR_GENERIC fi } # For galera, promote is really start galera_promote() { local rc local extra_opts local bootstrap local safe_to_bootstrap master_exists if [ $? -eq 0 ]; then # join without bootstrapping extra_opts="--wsrep-cluster-address=${OCF_RESKEY_wsrep_cluster_address}" else bootstrap=$(is_bootstrap) if ocf_is_true $bootstrap; then # The best node for bootstrapping wasn't cleanly shutdown. Allow # bootstrapping anyways if [ "$(get_safe_to_bootstrap)" = "0" ]; then sed -ie 's/^\(safe_to_bootstrap:\) 0/\1 1/' ${OCF_RESKEY_datadir}/grastate.dat ocf_log info "safe_to_bootstrap in ${OCF_RESKEY_datadir}/grastate.dat set to 1 on node ${NODENAME}" fi ocf_log info "Node <${NODENAME}> is bootstrapping the cluster" extra_opts="--wsrep-cluster-address=gcomm://" else # We are being promoted without having the bootstrap # attribute in the CIB, which means we are supposed to # join a cluster; however if we end up here, there is no # Master remaining right now, which means there is no # cluster to join anymore. So force a demotion, and and # let the RA decide later which node should be the next # bootstrap node. ocf_log warn "There is no running cluster to join, demoting ourself" clear_master_score return $OCF_SUCCESS fi fi galera_monitor if [ $? -eq $OCF_RUNNING_MASTER ]; then if ocf_is_true $bootstrap; then promote_everyone clear_bootstrap_node ocf_log info "boostrap node already up, promoting the rest of the galera instances." fi clear_safe_to_bootstrap clear_last_commit return $OCF_SUCCESS fi # last commit/safe_to_bootstrap flag are no longer relevant once promoted clear_last_commit clear_safe_to_bootstrap mysql_common_prepare_dirs mysql_common_start "$extra_opts" rc=$? if [ $rc != $OCF_SUCCESS ]; then return $rc fi + # At this point, the mysql pidfile is created on disk and the + # mysql server is reacheable via its UNIX socket. If we are a + # joiner, SST transfers (rsync) have finished, but an IST may + # still be requested or ongoing + galera_monitor rc=$? if [ $rc != $OCF_SUCCESS -a $rc != $OCF_RUNNING_MASTER ]; then ocf_exit_reason "Failed initial monitor action" return $rc fi is_readonly if [ $? -eq 0 ]; then ocf_exit_reason "Failure. Master instance started in read-only mode, check configuration." return $OCF_ERR_GENERIC fi - is_primary - if [ $? -ne 0 ]; then - ocf_exit_reason "Failure. Master instance started, but is not in Primary mode." - return $OCF_ERR_GENERIC - fi - if ocf_is_true $bootstrap; then promote_everyone clear_bootstrap_node # clear attribute no-grastate. if last shutdown was # not clean, we cannot be extra-cautious by requesting a SST # since this is the bootstrap node clear_no_grastate ocf_log info "Bootstrap complete, promoting the rest of the galera instances." else # if this is not the bootstrap node, make sure this instance # syncs with the rest of the cluster before promotion returns. wait_for_sync # sync is done, clear info about last startup clear_no_grastate fi ocf_log info "Galera started" return $OCF_SUCCESS } galera_demote() { mysql_common_stop rc=$? if [ $rc -ne $OCF_SUCCESS ] && [ $rc -ne $OCF_NOT_RUNNING ]; then ocf_exit_reason "Failed to stop Master galera instance during demotion to Master" return $rc fi # if this node was previously a bootstrap node, that is no longer the case. clear_bootstrap_node clear_last_commit clear_no_grastate clear_safe_to_bootstrap # Clear master score here rather than letting pacemaker do so once # demote finishes. This way a promote cannot take place right # after this demote even if pacemaker is requested to do so. It # will first have to run a start/monitor op, to reprobe the state # of the other galera nodes and act accordingly. clear_master_score # record last commit for next promotion detect_safe_to_bootstrap detect_last_commit rc=$? return $rc } galera_start() { local rc local galera_node galera_node=$(pcmk_to_galera_name $NODENAME) if [ -z "$galera_node" ]; then ocf_exit_reason "Could not determine galera name from pacemaker node <${NODENAME}>." return $OCF_ERR_CONFIGURED fi echo $OCF_RESKEY_wsrep_cluster_address | grep -q -F $galera_node if [ $? -ne 0 ]; then ocf_exit_reason "local node <${NODENAME}> (galera node <${galera_node}>) must be a member of the wsrep_cluster_address <${OCF_RESKEY_wsrep_cluster_address}> to start this galera instance" return $OCF_ERR_CONFIGURED fi galera_monitor if [ $? -eq $OCF_RUNNING_MASTER ]; then ocf_exit_reason "master galera instance started outside of the cluster's control" return $OCF_ERR_GENERIC fi mysql_common_prepare_dirs detect_safe_to_bootstrap detect_last_commit rc=$? if [ $rc -ne $OCF_SUCCESS ]; then return $rc fi master_exists if [ $? -eq 0 ]; then ocf_log info "Master instances are already up, setting master score so this instance will join galera cluster." set_master_score $NODENAME else clear_master_score detect_first_master fi return $OCF_SUCCESS } galera_monitor() { local rc local galera_node local status_loglevel="err" # Set loglevel to info during probe if ocf_is_probe; then status_loglevel="info" fi mysql_common_status $status_loglevel rc=$? if [ $rc -eq $OCF_NOT_RUNNING ]; then last_commit=$(get_last_commit $node) if [ -n "$last_commit" ]; then # if last commit is set, this instance is considered started in slave mode rc=$OCF_SUCCESS master_exists if [ $? -ne 0 ]; then detect_first_master else # a master instance exists and is healthy, promote this # local read only instance # so it can join the master galera cluster. set_master_score fi fi return $rc elif [ $rc -ne $OCF_SUCCESS ]; then return $rc fi # if we make it here, mysql is running. Check cluster status now. galera_node=$(pcmk_to_galera_name $NODENAME) if [ -z "$galera_node" ]; then ocf_exit_reason "Could not determine galera name from pacemaker node <${NODENAME}>." return $OCF_ERR_CONFIGURED fi echo $OCF_RESKEY_wsrep_cluster_address | grep -q -F $galera_node if [ $? -ne 0 ]; then ocf_exit_reason "local node <${NODENAME}> (galera node <${galera_node}>) is started, but is not a member of the wsrep_cluster_address <${OCF_RESKEY_wsrep_cluster_address}>" return $OCF_ERR_GENERIC fi is_primary if [ $? -eq 0 ]; then if ocf_is_probe; then # restore master score during probe # if we detect this is a master instance set_master_score fi rc=$OCF_RUNNING_MASTER else - ocf_exit_reason "local node <${NODENAME}> is started, but not in primary mode. Unknown state." - rc=$OCF_ERR_GENERIC + # It seems that with recent galera (26.4+), a joiner that is + # connected to a Primary component and is preparing its IST + # request might still temporarily report its state as + # Non-Primary. Do not fail in this case as the promote + # operation will loop until the IST finishes or the promote + # times out. + if [ "$__OCF_ACTION" = "promote" ] && ! ocf_is_true $(is_bootstrap); then + ocf_log info "local node <${NODENAME}> is receiving a State Transfer." + else + ocf_exit_reason "local node <${NODENAME}> is started, but not in primary mode. Unknown state." + rc=$OCF_ERR_GENERIC + fi fi return $rc } galera_stop() { local rc # make sure the process is stopped mysql_common_stop rc=$1 clear_safe_to_bootstrap clear_last_commit clear_master_score clear_bootstrap_node clear_no_grastate return $rc } galera_validate() { if [ "$OCF_CHECK_LEVEL" -eq 10 ]; then if ! ocf_is_ms; then ocf_exit_reason "Galera must be configured as a multistate Master/Slave resource." return $OCF_ERR_CONFIGURED fi fi if [ -z "$OCF_RESKEY_wsrep_cluster_address" ]; then ocf_exit_reason "Galera must be configured with a wsrep_cluster_address value." return $OCF_ERR_CONFIGURED fi mysql_common_validate } case "$1" in meta-data) meta_data exit $OCF_SUCCESS;; usage|help) usage exit $OCF_SUCCESS;; esac [ "$__OCF_ACTION" = "start" ] && OCF_CHECK_LEVEL=10 galera_validate rc=$? LSB_STATUS_STOPPED=3 if [ $rc -ne 0 ]; then case "$1" in stop) exit $OCF_SUCCESS;; monitor) exit $OCF_NOT_RUNNING;; status) exit $LSB_STATUS_STOPPED;; *) exit $rc;; esac fi if [ -z "${OCF_RESKEY_check_passwd}" ]; then # This value is automatically sourced from /etc/sysconfig/checkcluster if available OCF_RESKEY_check_passwd=${MYSQL_PASSWORD} fi if [ -z "${OCF_RESKEY_check_user}" ]; then # This value is automatically sourced from /etc/sysconfig/checkcluster if available OCF_RESKEY_check_user=${MYSQL_USERNAME} fi : ${OCF_RESKEY_check_user="root"} MYSQL_OPTIONS_CHECK="-nNE --user=${OCF_RESKEY_check_user}" if ocf_is_true "${OCF_RESKEY_check_passwd_use_empty}"; then MYSQL_OPTIONS_CHECK="$MYSQL_OPTIONS_CHECK --password=" elif [ -n "${OCF_RESKEY_check_passwd}" ]; then MYSQL_OPTIONS_CHECK="$MYSQL_OPTIONS_CHECK --password=${OCF_RESKEY_check_passwd}" fi # This value is automatically sourced from /etc/sysconfig/checkcluster if available if [ -n "${MYSQL_HOST}" ]; then MYSQL_OPTIONS_CHECK="$MYSQL_OPTIONS_CHECK -h ${MYSQL_HOST}" fi # This value is automatically sourced from /etc/sysconfig/checkcluster if available if [ -n "${MYSQL_PORT}" ]; then MYSQL_OPTIONS_CHECK="$MYSQL_OPTIONS_CHECK -P ${MYSQL_PORT}" fi # What kind of method was invoked? case "$1" in start) galera_start;; stop) galera_stop;; status) mysql_common_status err;; monitor) galera_monitor;; promote) galera_promote;; demote) galera_demote;; validate-all) exit $OCF_SUCCESS;; *) usage exit $OCF_ERR_UNIMPLEMENTED;; esac # vi:sw=4:ts=4:et: