diff --git a/doc/man/Makefile.am b/doc/man/Makefile.am index 53c9975ec..a902252a0 100644 --- a/doc/man/Makefile.am +++ b/doc/man/Makefile.am @@ -1,223 +1,224 @@ # # doc: Linux-HA resource agents # # Copyright (C) 2009 Florian Haas # # This program is free software; you can redistribute it and/or # modify it under the terms of the GNU General Public License # as published by the Free Software Foundation; either version 2 # of the License, or (at your option) any later version. # # This program is distributed in the hope that it will be useful, # but WITHOUT ANY WARRANTY; without even the implied warranty of # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the # GNU General Public License for more details. # # You should have received a copy of the GNU General Public License # along with this program; if not, write to the Free Software # Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA. # MAINTAINERCLEANFILES = Makefile.in EXTRA_DIST = $(doc_DATA) $(REFENTRY_STYLESHEET) \ mkappendix.sh ralist.sh CLEANFILES = $(man_MANS) $(xmlfiles) metadata-*.xml STYLESHEET_PREFIX ?= http://docbook.sourceforge.net/release/xsl/current MANPAGES_STYLESHEET ?= $(STYLESHEET_PREFIX)/manpages/docbook.xsl HTML_STYLESHEET ?= $(STYLESHEET_PREFIX)/xhtml/docbook.xsl FO_STYLESHEET ?= $(STYLESHEET_PREFIX)/fo/docbook.xsl REFENTRY_STYLESHEET ?= ra2refentry.xsl XSLTPROC_OPTIONS ?= --xinclude XSLTPROC_MANPAGES_OPTIONS ?= $(XSLTPROC_OPTIONS) XSLTPROC_HTML_OPTIONS ?= $(XSLTPROC_OPTIONS) XSLTPROC_FO_OPTIONS ?= $(XSLTPROC_OPTIONS) radir = $(top_srcdir)/heartbeat # OCF_ROOT=. is necessary due to a sanity check in ocf-shellfuncs # (which tests whether $OCF_ROOT points to a directory metadata-%.xml: $(radir)/% OCF_ROOT=. OCF_FUNCTIONS_DIR=$(radir) $< meta-data > $@ metadata-IPv6addr.xml: ../../heartbeat/IPv6addr OCF_ROOT=. OCF_FUNCTIONS_DIR=$(radir) $< meta-data > $@ # Please note: we can't name the man pages # ocf:heartbeat:. Believe me, I've tried. It looks like it # works, but then it doesn't. While make can deal correctly with # colons in target names (when properly escaped), it royally messes up # when it is deals with _dependencies_ that contain colons. See Bug # 12126 on savannah.gnu.org. But, maybe it gets fixed soon, it was # first reported in 1995 and added to Savannah in in 2005... if BUILD_DOC man_MANS = ocf_heartbeat_AoEtarget.7 \ ocf_heartbeat_AudibleAlarm.7 \ ocf_heartbeat_ClusterMon.7 \ ocf_heartbeat_CTDB.7 \ ocf_heartbeat_Delay.7 \ ocf_heartbeat_Dummy.7 \ ocf_heartbeat_EvmsSCC.7 \ ocf_heartbeat_Evmsd.7 \ ocf_heartbeat_Filesystem.7 \ ocf_heartbeat_ICP.7 \ ocf_heartbeat_IPaddr.7 \ ocf_heartbeat_IPaddr2.7 \ ocf_heartbeat_IPsrcaddr.7 \ ocf_heartbeat_LVM.7 \ ocf_heartbeat_LVM-activate.7 \ ocf_heartbeat_LinuxSCSI.7 \ ocf_heartbeat_MailTo.7 \ ocf_heartbeat_ManageRAID.7 \ ocf_heartbeat_ManageVE.7 \ ocf_heartbeat_NodeUtilization.7 \ ocf_heartbeat_Pure-FTPd.7 \ ocf_heartbeat_Raid1.7 \ ocf_heartbeat_Route.7 \ ocf_heartbeat_SAPDatabase.7 \ ocf_heartbeat_SAPInstance.7 \ ocf_heartbeat_SendArp.7 \ ocf_heartbeat_ServeRAID.7 \ ocf_heartbeat_SphinxSearchDaemon.7 \ ocf_heartbeat_Squid.7 \ ocf_heartbeat_Stateful.7 \ ocf_heartbeat_SysInfo.7 \ ocf_heartbeat_VIPArip.7 \ ocf_heartbeat_VirtualDomain.7 \ ocf_heartbeat_WAS.7 \ ocf_heartbeat_WAS6.7 \ ocf_heartbeat_WinPopup.7 \ ocf_heartbeat_Xen.7 \ ocf_heartbeat_Xinetd.7 \ ocf_heartbeat_ZFS.7 \ ocf_heartbeat_aliyun-vpc-move-ip.7 \ ocf_heartbeat_anything.7 \ ocf_heartbeat_apache.7 \ ocf_heartbeat_asterisk.7 \ ocf_heartbeat_aws-vpc-move-ip.7 \ ocf_heartbeat_aws-vpc-route53.7 \ ocf_heartbeat_awseip.7 \ ocf_heartbeat_awsvip.7 \ ocf_heartbeat_azure-lb.7 \ ocf_heartbeat_clvm.7 \ ocf_heartbeat_conntrackd.7 \ ocf_heartbeat_crypt.7 \ ocf_heartbeat_db2.7 \ ocf_heartbeat_dhcpd.7 \ ocf_heartbeat_docker.7 \ ocf_heartbeat_dovecot.7 \ ocf_heartbeat_dnsupdate.7 \ ocf_heartbeat_eDir88.7 \ ocf_heartbeat_ethmonitor.7 \ ocf_heartbeat_exportfs.7 \ ocf_heartbeat_fio.7 \ ocf_heartbeat_galera.7 \ ocf_heartbeat_garbd.7 \ ocf_heartbeat_gcp-vpc-move-ip.7 \ ocf_heartbeat_iSCSILogicalUnit.7 \ ocf_heartbeat_iSCSITarget.7 \ ocf_heartbeat_iface-bridge.7 \ ocf_heartbeat_iface-vlan.7 \ ocf_heartbeat_ipsec.7 \ ocf_heartbeat_ids.7 \ ocf_heartbeat_iscsi.7 \ ocf_heartbeat_jboss.7 \ ocf_heartbeat_jira.7 \ ocf_heartbeat_kamailio.7 \ ocf_heartbeat_lvmlockd.7 \ ocf_heartbeat_lxc.7 \ ocf_heartbeat_lxd-info.7 \ ocf_heartbeat_machine-info.7 \ ocf_heartbeat_mariadb.7 \ + ocf_heartbeat_mdraid.7 \ ocf_heartbeat_minio.7 \ ocf_heartbeat_mysql.7 \ ocf_heartbeat_mysql-proxy.7 \ ocf_heartbeat_nagios.7 \ ocf_heartbeat_named.7 \ ocf_heartbeat_nfsnotify.7 \ ocf_heartbeat_nfsserver.7 \ ocf_heartbeat_nginx.7 \ ocf_heartbeat_openstack-info.7 \ ocf_heartbeat_openstack-cinder-volume.7 \ ocf_heartbeat_openstack-floating-ip.7 \ ocf_heartbeat_oraasm.7 \ ocf_heartbeat_oracle.7 \ ocf_heartbeat_oralsnr.7 \ ocf_heartbeat_ovsmonitor.7 \ ocf_heartbeat_pgagent.7 \ ocf_heartbeat_pgsql.7 \ ocf_heartbeat_pingd.7 \ ocf_heartbeat_podman.7 \ ocf_heartbeat_portblock.7 \ ocf_heartbeat_postfix.7 \ ocf_heartbeat_pound.7 \ ocf_heartbeat_proftpd.7 \ ocf_heartbeat_rabbitmq-cluster.7 \ ocf_heartbeat_redis.7 \ ocf_heartbeat_rkt.7 \ ocf_heartbeat_rsyncd.7 \ ocf_heartbeat_rsyslog.7 \ ocf_heartbeat_scsi2reservation.7 \ ocf_heartbeat_sfex.7 \ ocf_heartbeat_slapd.7 \ ocf_heartbeat_sybaseASE.7 \ ocf_heartbeat_sg_persist.7 \ ocf_heartbeat_mpathpersist.7 \ ocf_heartbeat_symlink.7 \ ocf_heartbeat_syslog-ng.7 \ ocf_heartbeat_tomcat.7 \ ocf_heartbeat_varnish.7 \ ocf_heartbeat_vdo-vol.7 \ ocf_heartbeat_vmware.7 \ ocf_heartbeat_vsftpd.7 \ ocf_heartbeat_zabbixserver.7 if USE_IPV6ADDR_AGENT man_MANS += ocf_heartbeat_IPv6addr.7 endif if BUILD_AZURE_EVENTS man_MANS += ocf_heartbeat_azure-events.7 endif if BUILD_GCP_PD_MOVE man_MANS += ocf_heartbeat_gcp-pd-move.7 endif if BUILD_GCP_VPC_MOVE_ROUTE man_MANS += ocf_heartbeat_gcp-vpc-move-route.7 endif if BUILD_GCP_VPC_MOVE_VIP man_MANS += ocf_heartbeat_gcp-vpc-move-vip.7 endif xmlfiles = $(man_MANS:.7=.xml) %.1 %.5 %.7 %.8: %.xml $(XSLTPROC) \ $(XSLTPROC_MANPAGES_OPTIONS) \ $(MANPAGES_STYLESHEET) $< ocf_heartbeat_%.xml: metadata-%.xml $(srcdir)/$(REFENTRY_STYLESHEET) $(XSLTPROC) --novalid \ --stringparam package $(PACKAGE_NAME) \ --stringparam version $(VERSION) \ --output $@ \ $(srcdir)/$(REFENTRY_STYLESHEET) $< ocf_resource_agents.xml: $(xmlfiles) mkappendix.sh ./mkappendix.sh $(xmlfiles) > $@ %.html: %.xml $(XSLTPROC) \ $(XSLTPROC_HTML_OPTIONS) \ --output $@ \ $(HTML_STYLESHEET) $< xml: ocf_resource_agents.xml endif diff --git a/heartbeat/Makefile.am b/heartbeat/Makefile.am index bbc9590ac..e4423bbae 100644 --- a/heartbeat/Makefile.am +++ b/heartbeat/Makefile.am @@ -1,227 +1,228 @@ # Makefile.am for OCF RAs # # Author: Sun Jing Dong # Copyright (C) 2004 IBM # # This program is free software; you can redistribute it and/or # modify it under the terms of the GNU General Public License # as published by the Free Software Foundation; either version 2 # of the License, or (at your option) any later version. # # This program is distributed in the hope that it will be useful, # but WITHOUT ANY WARRANTY; without even the implied warranty of # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the # GNU General Public License for more details. # # You should have received a copy of the GNU General Public License # along with this program; if not, write to the Free Software # Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA. # MAINTAINERCLEANFILES = Makefile.in EXTRA_DIST = $(ocf_SCRIPTS) $(ocfcommon_DATA) \ $(common_DATA) $(hb_DATA) $(dtd_DATA) \ README AM_CPPFLAGS = -I$(top_srcdir)/include -I$(top_srcdir)/linux-ha halibdir = $(libexecdir)/heartbeat ocfdir = $(OCF_RA_DIR_PREFIX)/heartbeat dtddir = $(datadir)/$(PACKAGE_NAME) dtd_DATA = ra-api-1.dtd metadata.rng if USE_IPV6ADDR_AGENT ocf_PROGRAMS = IPv6addr else ocf_PROGRAMS = endif if IPV6ADDR_COMPATIBLE halib_PROGRAMS = send_ua else halib_PROGRAMS = endif IPv6addr_SOURCES = IPv6addr.c IPv6addr_utils.c send_ua_SOURCES = send_ua.c IPv6addr_utils.c IPv6addr_LDADD = -lplumb $(LIBNETLIBS) send_ua_LDADD = $(LIBNETLIBS) ocf_SCRIPTS = AoEtarget \ AudibleAlarm \ ClusterMon \ CTDB \ Delay \ Dummy \ EvmsSCC \ Evmsd \ Filesystem \ ICP \ IPaddr \ IPaddr2 \ IPsrcaddr \ LVM \ LinuxSCSI \ lvmlockd \ LVM-activate \ MailTo \ ManageRAID \ ManageVE \ NodeUtilization \ Pure-FTPd \ Raid1 \ Route \ SAPDatabase \ SAPInstance \ SendArp \ ServeRAID \ SphinxSearchDaemon \ Squid \ Stateful \ SysInfo \ VIPArip \ VirtualDomain \ WAS \ WAS6 \ WinPopup \ Xen \ Xinetd \ ZFS \ aliyun-vpc-move-ip \ anything \ apache \ asterisk \ aws-vpc-move-ip \ aws-vpc-route53 \ awseip \ awsvip \ azure-lb \ clvm \ conntrackd \ crypt \ db2 \ dhcpd \ dnsupdate \ docker \ dovecot \ eDir88 \ ethmonitor \ exportfs \ fio \ galera \ garbd \ gcp-vpc-move-ip \ iSCSILogicalUnit \ iSCSITarget \ ids \ iface-bridge \ iface-vlan \ ipsec \ iscsi \ jboss \ jira \ kamailio \ lxc \ lxd-info \ machine-info \ mariadb \ + mdraid \ minio \ mysql \ mysql-proxy \ nagios \ named \ nfsnotify \ nfsserver \ nginx \ openstack-cinder-volume \ openstack-floating-ip \ openstack-info \ oraasm \ oracle \ oralsnr \ ovsmonitor \ pgagent \ pgsql \ pingd \ podman \ portblock \ postfix \ pound \ proftpd \ rabbitmq-cluster \ redis \ rkt \ rsyncd \ rsyslog \ scsi2reservation \ sfex \ sg_persist \ mpathpersist \ slapd \ sybaseASE \ symlink \ syslog-ng \ tomcat \ varnish \ vdo-vol \ vmware \ vsftpd \ zabbixserver if BUILD_AZURE_EVENTS ocf_SCRIPTS += azure-events endif if BUILD_GCP_PD_MOVE ocf_SCRIPTS += gcp-pd-move endif if BUILD_GCP_VPC_MOVE_ROUTE ocf_SCRIPTS += gcp-vpc-move-route endif if BUILD_GCP_VPC_MOVE_VIP ocf_SCRIPTS += gcp-vpc-move-vip endif ocfcommondir = $(OCF_LIB_DIR_PREFIX)/heartbeat ocfcommon_DATA = ocf-shellfuncs \ ocf-binaries \ ocf-directories \ ocf-returncodes \ ocf-rarun \ ocf-distro \ apache-conf.sh \ http-mon.sh \ sapdb-nosha.sh \ sapdb.sh \ lvm-clvm.sh \ lvm-plain.sh \ lvm-tag.sh \ ora-common.sh \ mysql-common.sh \ nfsserver-redhat.sh \ findif.sh \ ocf.py # Legacy locations hbdir = $(sysconfdir)/ha.d hb_DATA = shellfuncs check: $(ocf_SCRIPTS:=.check) %.check: % OCF_ROOT=$(abs_srcdir) OCF_FUNCTIONS_DIR=$(abs_srcdir) ./$< meta-data | xmllint --path $(abs_srcdir) --noout --relaxng $(abs_srcdir)/metadata.rng - do_spellcheck = printf '[%s]\n' "$(agent)"; \ OCF_ROOT=$(abs_srcdir) OCF_FUNCTIONS_DIR=$(abs_srcdir) \ ./$(agent) meta-data 2>/dev/null \ | xsltproc $(top_srcdir)/make/extract_text.xsl - \ | aspell pipe list -d en_US --ignore-case \ --home-dir=$(top_srcdir)/make -p spellcheck-ignore \ | sed -n 's|^&\([^:]*\):.*|\1|p'; spellcheck: @$(foreach agent,$(ocf_SCRIPTS), $(do_spellcheck)) diff --git a/heartbeat/mdraid b/heartbeat/mdraid new file mode 100755 index 000000000..19c058afe --- /dev/null +++ b/heartbeat/mdraid @@ -0,0 +1,576 @@ +#!/bin/sh +# +# License: GNU General Public License (GPL) +# Support: users@clusterlabs.org +# +# mdraid (inspired by the Raid1 upstream resource agent) +# +# Description: Manages a Linux software RAID device on a (shared) storage medium. +# Author: Heinz Mauelshagen (heinzm@redhat.com) +# Release: Mar 2020 +# +# usage: $0 {start|stop|monitor|validate-all|usage|meta-data} +# +# EXAMPLE config file /etc/mdadm.conf (for more info: mdadm.conf(5)) +# +# AUTO -all +# ARRAY /dev/md0 UUID=4a865b55:ba27ef8d:29cd5701:6fb42799 +# +####################################################################### +# Initialization: + +: ${OCF_FUNCTIONS_DIR=${OCF_ROOT}/lib/heartbeat} +. ${OCF_FUNCTIONS_DIR}/ocf-shellfuncs + +# Parameter defaults + +OCF_RESKEY_mdadm_conf_default="" +OCF_RESKEY_md_dev_default="" +OCF_RESKEY_force_stop_default="false" +OCF_RESKEY_wait_for_udev_default="true" +OCF_RESKEY_force_clones_default="false" + +: ${OCF_RESKEY_mdadm_conf=${OCF_RESKEY_mdadm_conf_default}} +: ${OCF_RESKEY_md_dev=${OCF_RESKEY_md_dev_default}} +: ${OCF_RESKEY_force_stop=${OCF_RESKEY_force_stop_default}} +: ${OCF_RESKEY_wait_for_udev=${OCF_RESKEY_wait_for_udev_default}} +: ${OCF_RESKEY_force_clones=${OCF_RESKEY_force_clones_default}} + +####################################################################### + +usage() { + cat <<-EOT + usage: $0 {start|stop|monitor|validate-all|usage|meta-data} + EOT +} + +# +# Action: provide meta-data (parameter specifications and descriptive text) +# +meta_data() { + cat < + + +1.0 + + +This resource agent manages Linux software RAID (MD) devices on +a shared storage medium ensuring that non-clustered MD arrays +are prohibited from starting cloned (which would cause data corruption +(e.g., on raid6 arrays) unless forced (see force_clones parameter). +Clustered MD RAID layouts (see below) will be discovered and allowed +cloning by default; no need to set force_clones. + +It uses mdadm(8) to start, stop, and monitor the MD devices. + +Supported clustered (i.e., clonable active-active) arrays are linear, +raid0, and clustered raid1/raid10 (i.e. mdadm(8) created with +--bitmap=clustered). + +Manages Linux software RAID (MD) devices on shared +storage + + + + +The MD RAID configuration file (e.g., /etc/mdadm.conf). + +MD config file + + + + + +MD array block device to use (e.g., /dev/md0 or /dev/md/3). +With shared access to the array's storage, this should +preferably be a clustered raid1 or raid10 array created +with --bitmap=clustered, assuming its resource will +be cloned (i.e., active-active access). + +Be sure to disable auto-assembly for the resource-managed arrays! + +MD block device + + + + + +If processes or kernel threads are using the array, it cannot be +stopped. We will try to stop processes, first by sending TERM and +then, if that doesn't help in $PROC_CLEANUP_TIME seconds, using KILL. +The lsof(8) program is required to get the list of array users. +Of course, the kernel threads cannot be stopped this way. +If the processes are critical for data integrity, then set this +parameter to false. Note that in that case the stop operation +will fail and the node will be fenced. + +force stop processes using the array + + + + + +Wait until udevd creates a device in the start operation. On a +normally loaded host this should happen quickly, but you may be +unlucky. If you are not using udev set this to "no". + +wait_for_udev + + + + + +Activating the same, non-clustered MD RAID array (i.e. single-host +raid1/4/5/6/10) on multiple nodes at the same time will result in +data corruption and thus is forbidden by default. + +A safe example could be an (exotic) array that is only named identically +across all nodes, but is in fact based on distinct (non-shared) storage. + +Only set this to "true" if you know what you are doing! + +force ability to run as a clone + + + + + + + + + + + + + +END +} + +####################################################################### + +# ocf-shellfunc ocf_is_clone() fails with meta attribute clone-max < 2. +# Checking for defined meta_clone_max reskey is sufficient until fixed. +resource_is_cloned() { + [ -z "$OCF_RESKEY_CRM_meta_clone_max" ] && return 1 || return 0; +} + +raid_validate_all() { + if [ -z "$mdadm_conf" ] ; then + ocf_exit_reason "Please set OCF_RESKEY_mdadm_conf" + return $OCF_ERR_CONFIGURED + fi + if [ ! -r "$mdadm_conf" ] ; then + ocf_exit_reason "Configuration file [$mdadm_conf] does not exist, or cannot be opened" + return $OCF_ERR_ARGS + fi + if [ -z "$md_dev" ] ; then + ocf_exit_reason "Please set OCF_RESKEY_md_dev to the MD RAID array block device you want to control" + return $OCF_ERR_CONFIGURED + fi + case "$md_dev" in + /dev/*) ;; + *) ocf_exit_reason "Bogus MD RAID array block device name (\"$md_dev\")" + return $OCF_ERR_ARGS;; + esac + if ocf_is_true $wait_for_udev && ! have_binary udevadm && [ "$__OCF_ACTION" = "start" ]; then + ocf_exit_reason "either install udevadm or set udev to false" + return $OCF_ERR_INSTALLED + fi + if ocf_is_true $force_stop && ! have_binary lsof; then + ocf_exit_reason "Please install lsof(8) or set force_stop to false." + return $OCF_ERR_INSTALLED + fi + if ! have_binary $MDADM; then + ocf_exit_reason "Please install mdadm(8)!" + return $OCF_ERR_INSTALLED + fi + if ! have_binary blkid; then + ocf_exit_reason "Please install blkid(8). We need it to list MD array UUIDs!" + return $OCF_ERR_INSTALLED + fi + if [ `echo $md_dev|wc -w` -gt 1 ]; then + ocf_exit_reason "Only one MD array supported" + return $OCF_ERR_CONFIGURED + fi + + return $OCF_SUCCESS +} + +# Remove ':' or '-' from uuid string to be able to compare between MD and blkid format. +uuid_flat() { + echo $1|sed 's/[-:]//g' +} + +# Global variable for devices by MD uuid. +devs="" + +# Get array uuid from mdadm_conf based on $md_dev. +get_array_uuid_by_mddev() { + local array_uuid + + array_uuid="`grep $md_dev $mdadm_conf`" + if [ -z "$array_uuid" ] + then + ocf_exit_reason "Entry for $MMDEV does not exist in $mdadm_conf!" + return $OCF_ERR_CONFIGURED + fi + + array_uuid=$(echo $array_uuid | sed 's/^.*UUID=//;s/ .*$//') + if [ -z "$array_uuid" ] + then + ocf_exit_reason "Bogus entry for $MMDEV in $mdadm_conf!" + return $OCF_ERR_CONFIGURED + fi + + echo `uuid_flat $array_uuid` + + return $OCF_SUCCESS +} + +# Use blkid to get to the subset of raid members by array uuid. +list_devices_for_mddev() { + local array_uuid blkid_array_uuid dev line rc + + array_uuid=`get_array_uuid_by_mddev` + rc=$? + if [ $rc -ne $OCF_SUCCESS ]; then + ocf_exit_reason "Failed to get UUID of $md_dev from $mdadm_conf" + return $rc + fi + + blkid | grep linux_raid_member 2>/dev/null | while read line + do + dev=`echo $line | sed 's/: .*//'` + blkid_array_uuid=$(echo $line | sed 's/^.* UUID="//;s/" .*$//') + [ "`uuid_flat $blkid_array_uuid`" = "$array_uuid" ] && echo $dev + done +} + +# Check for linear or raid0 array; presumes defined global devs() array. +array_is_linear_or_raid0() { + local c=0 d + + for d in $devs + do + $MDADM -E $d 2>&1 | $EGREP -i "raid level : (raid0|linear)" >/dev/null 2>&1 + [ $? -eq 0 ] && let c=c+1 + done + + [ $c -eq `echo $devs|wc -w` ] && return 0 || return 1 +} + +# Return true for clustered RAID relying on all component devices reporting clustered type; +# presumes defined global devs variable with the component devices of the array. +array_is_clustered_raid() { + local c=0 d dev_count=`echo $devs|wc -w` s + # Check based on specific "intent bitmap : clustered" output once + # available in mdadm output or fall back to "Cluster Name" defined + # presuming bitmap is clustered if so. + local strs="clustered cluster.name" + + for d in $devs + do + for s in $strs + do + $MDADM -E $d 2>&1 | grep -i "$s" >/dev/null 2>&1 + if [ $? -eq 0 ]; then + let c=c+1 + break + fi + done + done + + [ $c -eq $dev_count ] && return 0 || return 1 +} + +# Check for all clustered types (linear, raid0, and clustered raid1/raid10). +is_clustered_raid() { + array_is_clustered_raid || array_is_linear_or_raid0 +} + +md_assemble() { + local rc + + $MDADM --assemble $md_dev --config="$mdadm_conf" + rc=$? + [ $rc -eq 0 ] && ocf_is_true $wait_for_udev && udevadm settle --exit-if-exists=$md_dev + + return $rc +} + +# Try setting an MD array to readonly. +mark_readonly() { + local rc + + $MDADM --readonly $md_dev --config="$mdadm_conf" + rc=$? + [ $rc -ne 0 ] && ocf_exit_reason "Failed to set $md_dev readonly (rc=$rc)" + + return $rc +} + +# Try stopping an MD array in case its block device is nonexistent for some reason. +mknod_raid_stop() { + local rc n tmp_block_file + + # first create a block device file, then try to stop the array + n=`echo $1 | sed 's/[^0-9]*//'` + if ! ocf_is_decimal "$n"; then + ocf_log warn "could not get the minor device number from $1" + return 1 + fi + tmp_block_file="$HA_RSCTMP/${OCF_RESOURCE_INSTANCE}-`basename $1`" + rm -f $tmp_block_file + ocf_log info "block device file $1 missing, creating one in order to stop the array" + mknod $tmp_block_file b 9 $n + $MDADM --stop $tmp_block_file --config="$mdadm_conf" + rc=$? + rm -f $tmp_block_file + return $rc +} + +# Stop an MD array. +raid_stop_one() { + if [ -b "$1" ]; then + $MDADM --stop $1 --config="$mdadm_conf" && return + else + # newer mdadm releases can stop arrays when given the + # basename; try that first + $MDADM --stop `basename $1` --config="$mdadm_conf" && return + # otherwise create a block device file + mknod_raid_stop $1 + fi +} + +# Functions show/stop any resource holding processes. +get_users_pids() { + ocf_log debug "running lsof to list $md_dev users..." + ocf_run -warn 'lsof $md_dev | tail -n +2 | $AWK "{print $2}" | sort -u' +} + +stop_raid_users() { + local pids=`get_users_pids $md_dev` + + if [ -z "$pids" ]; then + ocf_log warn "lsof reported no users holding arrays" + return 2 + else + ocf_stop_processes TERM $PROC_CLEANUP_TIME $pids + fi +} + +showusers() { + local disk=`basename $md_dev` + + ocf_log info "running lsof to list $disk users..." + ocf_run -warn lsof $md_dev + + if [ -d /sys/block/$disk/holders ]; then + ocf_log info "ls -l /sys/block/$disk/holders" + ocf_run -warn ls -l /sys/block/$disk/holders + fi +} + +####################################################################### + +# +# Action: START up the MD RAID array. +# +raid_start() { + local rc + + if resource_is_cloned && ! is_clustered_raid; then + if ocf_is_true "$OCF_RESKEY_force_clones"; then + ocf_log warn "Forced cloned starting non-clustered $md_dev which may lead to data corruption!" + else + ocf_exit_reason "Rejecting start: non-clustered MD RAID array $md_dev is NOT safe to run cloned" + exit $OCF_ERR_CONFIGURED + fi + fi + + raid_monitor + rc=$? + # md array already online, nothing to do. + [ $rc -eq $OCF_SUCCESS ] && return $rc + + 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_exit_reason "MD RAID array $md_dev in a broken state; cannot start (rc=$rc)" + return $OCF_ERR_GENERIC + fi + + md_assemble + rc=$? + if [ $rc -ne 0 ]; then + ocf_exit_reason "Failed to assemble MD RAID array $md_dev (rc=$rc, is $mdadm_conf up-to-date?)" + return $OCF_ERR_GENERIC + fi + + raid_monitor + [ $? -eq $OCF_SUCCESS ] && return $OCF_SUCCESS + + ocf_exit_reason "Couldn't start MD RAID array $md_dev (rc=$rc)" + + return $OCF_ERR_GENERIC +} + +# +# Action: STOP the MD RAID array +# +raid_stop() { + local rc + + # See if the MD device is already cleanly stopped: + raid_monitor + [ $? -eq $OCF_NOT_RUNNING ] && return $OCF_SUCCESS + + # Turn off raid + if ! raid_stop_one $md_dev; then + if ocf_is_true $force_stop; then + stop_raid_users + case $? in + 2) false;; + *) raid_stop_one $md_dev;; + esac + else + false + fi + fi + rc=$? + + if [ $rc -ne 0 ]; then + ocf_log warn "Couldn't stop MD RAID array $md_dev (rc=$rc)" + showusers $md_dev + mark_readonly $md_dev + return $OCF_ERR_GENERIC + fi + + raid_monitor + rc=$? + [ $rc -eq $OCF_NOT_RUNNING ] && return $OCF_SUCCESS + + ocf_exit_reason "MD RAID array $md_dev still active after stop command (rc=$rc)" + return $OCF_ERR_GENERIC +} + +# +# Action: monitor the MD RAID array. +# +raid_monitor() { + local TRY_READD=0 md rc pbsize + + # check if the md device exists first + # but not if we are in the stop operation + # device existence is important only for the running arrays + if [ "$__OCF_ACTION" != "stop" ]; then + if [ -h "$md_dev" ]; then + md=$(ls $md_dev -l | $AWK -F'/' '{print $NF}') + elif [ -b "$md_dev" ]; then + md=${md_dev#/dev/} + else + ocf_log info "$md_dev is not a block device" + return $OCF_NOT_RUNNING + fi + fi + + if ! grep -e "^$md[ \t:]" /proc/mdstat >/dev/null ; then + ocf_log info "$md not found in /proc/mdstat" + return $OCF_NOT_RUNNING + fi + + $MDADM --detail --test $md_dev >/dev/null 2>&1 + rc=$? + case $rc in + 0) ;; + 1) ocf_log warn "$md_dev has at least one failed device." + TRY_READD=1;; + 2) ocf_exit_reason "$md_dev has failed." + return $OCF_ERR_GENERIC;; + 4) + if [ "$__OCF_ACTION" = "stop" ] ; then + # There may be a transient invalid device after + # we stop MD due to uevent processing, the + # original device is stopped though. + return $OCF_NOT_RUNNING + else + ocf_exit_reason "mdadm failed on $md_dev." + return $OCF_ERR_GENERIC + fi;; + *) ocf_exit_reason "mdadm returned an unknown result ($rc)." + return $OCF_ERR_GENERIC;; + esac + + if ! array_is_linear_or_raid0; then + if [ "$__OCF_ACTION" = "monitor" ] && [ "$OCF_RESKEY_CRM_meta_interval" != 0 \ + ] && [ $TRY_READD -eq 1 ] && [ $OCF_CHECK_LEVEL -gt 0 ]; then + ocf_log info "Attempting recovery sequence to re-add devices on MD RAID array $md_dev:" + $MDADM $md_dev --fail detached + $MDADM $md_dev --remove failed + $MDADM $md_dev --re-add missing + # TODO: At this stage, there's nothing to actually do + # here. Either this worked or it did not. + fi + fi + + pbsize=`(blockdev --getpbsz $md_dev || stat -c "%o" $md_dev) 2>/dev/null` + if [ -z "$pbsize" ]; then + ocf_log warn "both blockdev and stat were unable to get the block size (will use 4k)" + pbsize=4096 # try with 4k + fi + if ! dd if=$md_dev count=1 bs=$pbsize of=/dev/null iflag=direct >/dev/null 2>&1; then + ocf_exit_reason "$md_dev: I/O error on read" + return $OCF_ERR_GENERIC + fi + + [ "$__OCF_ACTION" = "monitor" ] && ocf_log info "monitoring...($md_dev)" + + return $OCF_SUCCESS +} + +if [ $# -ne 1 ]; then + usage + exit $OCF_ERR_ARGS +fi + +# Process actions which are independant from validation +case "$1" in +meta-data) meta_data + exit $OCF_SUCCESS;; +usage) usage + exit $OCF_SUCCESS;; +*) ;; +esac + +# Define global variables used in ^ functions. +mdadm_conf="${OCF_RESKEY_mdadm_conf}" +md_dev="${OCF_RESKEY_md_dev}" +force_stop="${OCF_RESKEY_force_stop}" +wait_for_udev="${OCF_RESKEY_wait_for_udev}" + +# Validate all parameters and check for mandatory binaries present +raid_validate_all +rc=$? +[ $rc -ne $OCF_SUCCESS ] && exit $rc +# raid_validate_all already processed and result checked. +[ "$1" = "validate-all" ] && return ${OCF_SUCCESS} + +# Required by start|stop|monitor processed below +devs=(`list_devices_for_mddev`) +if [ `echo $devs|wc -l` -eq 0 ]; then + ocf_exit_reason "No component device(s) found for MD RAID array $md_dev" + exit $OCF_ERR_GENERIC +fi + +case "$1" in +start) raid_start;; +stop) raid_stop;; +monitor) raid_monitor;; +*) usage + exit $OCF_ERR_UNIMPLEMENTED;; +esac +rc=$? + +ocf_log debug "${OCF_RESOURCE_INSTANCE} $__OCF_ACTION : $rc" +exit $rc