diff --git a/heartbeat/CTDB.in b/heartbeat/CTDB.in index 46f56cfac..b4af66bc1 100755 --- a/heartbeat/CTDB.in +++ b/heartbeat/CTDB.in @@ -1,996 +1,996 @@ #!@BASH_SHELL@ # # OCF Resource Agent for managing CTDB # # Copyright (c) 2009-2010 Novell Inc., Tim Serong # 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. # # # OVERVIEW # # When run by itself, CTDB can handle IP failover and includes scripts # to manage various services (Samba, Winbind, HTTP, etc.). When run as # a resource in a Pacemaker cluster, this additional functionality # should not be used; instead one should define separate resources for # CTDB, Samba, Winbind, IP addresses, etc. # # As of 2010-11-17, there is no separate OCF Samba or Winbind RA, so # it is still possible to configure CTDB so that it manages these # resources itself. In future, once Samba and Winbind RAs are # available, this ability will be deprecated and ultimately removed. # # This RA intentionally provides no ability to configure CTDB such that # it manages IP failover, HTTP, NFS, etc. # # # TODO: # - ctdb_stop doesn't really support multiple independent CTDB instances, # unless they're running from distinct ctdbd binaries (it uses pkill # $OCF_RESKEY_ctdbd_binary if "ctdb stop" doesn't work, which it might # not under heavy load - this will kill all ctdbd instances on the # system). OTOH, running multiple CTDB instances per node is, well, # AFAIK, completely crazy. Can't run more than one in a vanilla CTDB # cluster, with the CTDB init script. So it might be nice to address # this for complete semantic correctness of the RA, but shouldn't # actually cause any trouble in real life. # - As much as possible, get rid of auto config generation # - Especially smb.conf # - Verify timeouts are sane # - Monitor differentiate between error and not running? # - Do we need to verify globally unique setting? # - Should set CTDB_NODES to ${HA_RSCTMP}/ctdb (generated based on # current nodes) # - Look at enabling set_ctdb_variables() if necessary. # - Probably possible for sysconfig file to not be restored if # CTDB dies unexpectedly. # ####################################################################### # Initialization: : ${OCF_FUNCTIONS_DIR=${OCF_ROOT}/lib/heartbeat} . ${OCF_FUNCTIONS_DIR}/ocf-shellfuncs ####################################################################### # Default parameter values: # Some distro's ctdb package stores the persistent db in /var/lib/ctdb, # others store in /var/ctdb. This attempts to detect the correct default # directory. var_prefix="/var/lib/ctdb" if [ ! -d "$var_prefix" ] && [ -d "/var/ctdb" ]; then var_prefix="/var/ctdb" fi run_prefix="/run" if [ ! -d "$var_prefix" ] && [ -d "/var/run" ]; then var_prefix="/var/run" fi # Parameter defaults OCF_RESKEY_ctdb_recovery_lock_default="" OCF_RESKEY_ctdb_manages_samba_default="no" OCF_RESKEY_ctdb_manages_winbind_default="no" OCF_RESKEY_ctdb_service_smb_default="" OCF_RESKEY_ctdb_service_nmb_default="" OCF_RESKEY_ctdb_service_winbind_default="" OCF_RESKEY_ctdb_samba_skip_share_check_default="yes" OCF_RESKEY_ctdb_monitor_free_memory_default="100" OCF_RESKEY_ctdb_start_as_disabled_default="no" : ${OCF_RESKEY_ctdb_recovery_lock=${OCF_RESKEY_ctdb_recovery_lock_default}} : ${OCF_RESKEY_ctdb_manages_samba=${OCF_RESKEY_ctdb_manages_samba_default}} : ${OCF_RESKEY_ctdb_manages_winbind=${OCF_RESKEY_ctdb_manages_winbind_default}} : ${OCF_RESKEY_ctdb_service_smb=${OCF_RESKEY_ctdb_service_smb_default}} : ${OCF_RESKEY_ctdb_service_nmb=${OCF_RESKEY_ctdb_service_nmb_default}} : ${OCF_RESKEY_ctdb_service_winbind=${OCF_RESKEY_ctdb_service_winbind_default}} : ${OCF_RESKEY_ctdb_samba_skip_share_check=${OCF_RESKEY_ctdb_samba_skip_share_check_default}} : ${OCF_RESKEY_ctdb_monitor_free_memory=${OCF_RESKEY_ctdb_monitor_free_memory_default}} : ${OCF_RESKEY_ctdb_start_as_disabled=${OCF_RESKEY_ctdb_start_as_disabled_default}} OCF_RESKEY_ctdb_config_dir_default="/etc/ctdb" OCF_RESKEY_ctdb_binary_default="/usr/bin/ctdb" OCF_RESKEY_ctdbd_binary_default="/usr/sbin/ctdbd" OCF_RESKEY_ctdb_dbdir_default="${var_prefix}" OCF_RESKEY_ctdb_logfile_default="/var/log/ctdb/log.ctdb" OCF_RESKEY_ctdb_rundir_default="${run_prefix}/ctdb" OCF_RESKEY_ctdb_timeout_default="10" : ${OCF_RESKEY_ctdb_config_dir=${OCF_RESKEY_ctdb_config_dir_default}} : ${OCF_RESKEY_ctdb_binary=${OCF_RESKEY_ctdb_binary_default}} : ${OCF_RESKEY_ctdbd_binary=${OCF_RESKEY_ctdbd_binary_default}} : ${OCF_RESKEY_ctdb_dbdir=${OCF_RESKEY_ctdb_dbdir_default}} : ${OCF_RESKEY_ctdb_logfile=${OCF_RESKEY_ctdb_logfile_default}} : ${OCF_RESKEY_ctdb_rundir=${OCF_RESKEY_ctdb_rundir_default}} : ${OCF_RESKEY_ctdb_timeout=${OCF_RESKEY_ctdb_timeout_default}} OCF_RESKEY_ctdb_socket_default="${OCF_RESKEY_ctdb_rundir}/ctdbd.socket" OCF_RESKEY_ctdb_debuglevel_default="2" OCF_RESKEY_ctdb_max_open_files_default="" : ${OCF_RESKEY_ctdb_socket=${OCF_RESKEY_ctdb_socket_default}} : ${OCF_RESKEY_ctdb_debuglevel=${OCF_RESKEY_ctdb_debuglevel_default}} : ${OCF_RESKEY_ctdb_max_open_files=${OCF_RESKEY_ctdb_max_open_files_default}} OCF_RESKEY_smb_conf_default="/etc/samba/smb.conf" OCF_RESKEY_smb_private_dir_default="" OCF_RESKEY_smb_passdb_backend_default="tdbsam" OCF_RESKEY_smb_idmap_backend_default="tdb2" OCF_RESKEY_smb_fileid_algorithm_default="" : ${OCF_RESKEY_smb_conf=${OCF_RESKEY_smb_conf_default}} : ${OCF_RESKEY_smb_private_dir=${OCF_RESKEY_smb_private_dir_default}} : ${OCF_RESKEY_smb_passdb_backend=${OCF_RESKEY_smb_passdb_backend_default}} : ${OCF_RESKEY_smb_idmap_backend=${OCF_RESKEY_smb_idmap_backend_default}} : ${OCF_RESKEY_smb_fileid_algorithm=${OCF_RESKEY_smb_fileid_algorithm_default}} ####################################################################### ctdb_version() { $OCF_RESKEY_ctdb_binary version | awk '{print $NF}' | sed "s/[-\.]\?[[:alpha:]].*//" } meta_data() { cat < 1.0 This resource agent manages CTDB, allowing one to use Clustered Samba in a Linux-HA/Pacemaker cluster. You need a shared filesystem (e.g. OCFS2 or GFS2) on which the CTDB lock will be stored. Create /etc/ctdb/nodes containing a list of private IP addresses of each node in the cluster, then configure this RA as a clone. This agent expects the samba and windbind resources to be managed outside of CTDB's control as a separate set of resources controlled by the cluster manager. The optional support for enabling CTDB management of these daemons will be depreciated. For more information see http://linux-ha.org/wiki/CTDB_(resource_agent) CTDB Resource Agent The location of a shared lock file or helper binary, common across all nodes. See CTDB documentation for details. CTDB shared lock file Should CTDB manage starting/stopping the Samba service for you? This will be deprecated in future, in favor of configuring a separate Samba resource. Should CTDB manage Samba? Should CTDB manage starting/stopping the Winbind service for you? This will be deprecated in future, in favor of configuring a separate Winbind resource. Should CTDB manage Winbind? Name of smb init script. Only necessary if CTDB is managing Samba directly. Will usually be auto-detected. Name of smb init script Name of nmb init script. Only necessary if CTDB is managing Samba directly. Will usually be auto-detected. Name of nmb init script Name of winbind init script. Only necessary if CTDB is managing Winbind directly. Will usually be auto-detected. Name of winbind init script If there are very many shares it may not be feasible to check that all of them are available during each monitoring interval. In that case this check can be disabled. Skip share check during monitor? If the amount of free memory drops below this value the node will become unhealthy and ctdb and all managed services will be shutdown. Once this occurs, the administrator needs to find the reason for the OOM situation, rectify it and restart ctdb with "service ctdb start". With CTDB 4.4.0 and later this parameter is ignored. Minimum amount of free memory (MB) When set to yes, the CTDB node will start in DISABLED mode and not host any public ip addresses. Start CTDB disabled? The directory containing various CTDB configuration files. The "nodes" and "notify.sh" scripts are expected to be in this directory. CTDB config file directory Full path to the CTDB binary. CTDB binary path Full path to the CTDB cluster daemon binary. CTDB Daemon binary path Full path to the domain socket that ctdbd will create, used for local clients to attach and communicate with the ctdb daemon. With CTDB 4.9.0 and later the socket path is hardcoded at build time, so this parameter is ignored. CTDB socket location (ignored with CTDB 4.9+) The directory to put the local CTDB database files in. Persistent database files will be put in ctdb_dbdir/persistent. CTDB database directory Full path to log file. To log to syslog instead, use the value "syslog". CTDB log file location Full path to ctdb runtime directory, used for storage of socket lock state. CTDB runtime directory location Indicates that ctdb should wait up to TIMEOUT seconds for a response to most commands sent to the CTDB daemon. CTDB timeout in seconds What debug level to run at (0-10). Higher means more verbose. CTDB debug level Maximum number of open files (for ulimit -n) Max open files Path to default samba config file. Only necessary if CTDB is managing Samba. Path to smb.conf The directory for smbd to use for storing such files as smbpasswd and secrets.tdb. Old versions of CTBD (prior to 1.0.50) required this to be on shared storage. This parameter should not be set for current versions of CTDB, and only remains in the RA for backwards compatibility. Samba private dir (deprecated) Which backend to use for storing user and possibly group information. Only necessary if CTDB is managing Samba. Samba passdb backend Which backend to use for SID/uid/gid mapping. Only necessary if CTDB is managing Samba. Samba idmap backend Which fileid:algorithm to use with vfs_fileid. The correct value depends on which clustered filesystem is in use, e.g.: for OCFS2, this should be set to "fsid". Only necessary if CTDB is managing Samba. Samba VFS fileid algorithm END } ####################################################################### # Figure out path to /etc/sysconfig/ctdb (same logic as # loadconfig() from /etc/ctdb/functions if [ -f /etc/sysconfig/ctdb ]; then CTDB_SYSCONFIG=/etc/sysconfig/ctdb elif [ -f /etc/default/ctdb ]; then CTDB_SYSCONFIG=/etc/default/ctdb elif [ -f "$OCF_RESKEY_ctdb_config_dir/ctdb" ]; then CTDB_SYSCONFIG=$OCF_RESKEY_ctdb_config_dir/ctdb elif [ -f "$OCF_RESKEY_ctdb_config_dir/ctdbd.conf" ]; then CTDB_SYSCONFIG=$OCF_RESKEY_ctdb_config_dir/ctdbd.conf fi # Backup paths CTDB_SYSCONFIG_BACKUP=${CTDB_SYSCONFIG}.ctdb-ra-orig invoke_ctdb() { # CTDB's defaults are: local timelimit timelimit=120 # ...but we override with the timeout for the current op: if [ -n "$OCF_RESKEY_CRM_meta_timeout" ]; then timelimit=$((OCF_RESKEY_CRM_meta_timeout/1000)) fi local vers=$(ctdb_version) ocf_version_cmp "$vers" "4.9.0" # if version < 4.9.0 specify '--socket' otherwise it's # a compiled option if [ "$?" -eq "0" ]; then $OCF_RESKEY_ctdb_binary --socket="$OCF_RESKEY_ctdb_socket" \ -t ${OCF_RESKEY_ctdb_timeout} -T $timelimit \ "$@" else $OCF_RESKEY_ctdb_binary \ -t ${OCF_RESKEY_ctdb_timeout} -T $timelimit \ "$@" fi } # Enable any event scripts that are explicitly required. # Any others will ultimately be invoked or not based on how they ship # with CTDB, but will generally have no effect, beacuase the relevant # CTDB_MANAGES_* options won't be set in /etc/sysconfig/ctdb. enable_event_scripts_chmod() { local event_dir event_dir=$OCF_RESKEY_ctdb_config_dir/events.d chmod u+x "$event_dir/00.ctdb" # core database health check if [ -f "${OCF_RESKEY_ctdb_config_dir}/public_addresses" ]; then chmod u+x "$event_dir/10.interface" else chmod a-x "$event_dir/10.interface" fi if [ -f "${OCF_RESKEY_ctdb_config_dir}/static-routes" ]; then chmod u+x "$event_dir/11.routing" else chmod a-x "$event_dir/11.routing" fi if ocf_is_true "$OCF_RESKEY_ctdb_manages_samba" || \ ocf_is_true "$OCF_RESKEY_ctdb_manages_winbind"; then chmod u+x "$event_dir/50.samba" else chmod a-x "$event_dir/50.samba" fi } enable_event_scripts_symlink() { # event scripts are symlinked once enabled, with the link source in... mkdir -p "$OCF_RESKEY_ctdb_config_dir/events/legacy" 2>/dev/null invoke_ctdb event script enable legacy 00.ctdb if [ -f "${OCF_RESKEY_ctdb_config_dir}/public_addresses" ]; then invoke_ctdb event script enable legacy 10.interface else invoke_ctdb event script disable legacy 10.interface fi if [ -f "${OCF_RESKEY_ctdb_config_dir}/static-routes" ]; then invoke_ctdb event script enable legacy 11.routing else invoke_ctdb event script disable legacy 11.routing fi if ocf_is_true "$OCF_RESKEY_ctdb_manages_winbind"; then invoke_ctdb event script enable legacy 49.winbind else invoke_ctdb event script disable legacy 49.winbind fi if ocf_is_true "$OCF_RESKEY_ctdb_manages_samba"; then invoke_ctdb event script enable legacy 50.samba else invoke_ctdb event script disable legacy 50.samba fi } # This function has no effect (currently no way to set CTDB_SET_*) # but remains here in case we need it in future. set_ctdb_variables() { rv=$OCF_SUCCESS set | grep ^CTDB_SET_ | cut -d_ -f3- | while read v; do varname=$(echo "$v" | cut -d= -f1) value=$(echo "$v" | cut -d= -f2) invoke_ctdb setvar "$varname" "$value" || rv=$OCF_ERR_GENERIC done || rv=$OCF_ERR_GENERIC return $rv } # Add necessary settings to /etc/samba/smb.conf. In a perfect world, # we'd be able to generate a new, temporary, smb.conf file somewhere, # something like: # include = /etc/samba/smb.conf # [global] # clustering = yes # # ...etc... # Unfortunately, we can't do this, because there's no way to tell the # smb init script where the temporary config is, so we just edit # the default config file. init_smb_conf() { # Don't screw around with the config if CTDB isn't managing Samba! ocf_is_true "$OCF_RESKEY_ctdb_manages_samba" || return 0 # replace these things in smb.conf local repl repl='# CTDB-RA:|passdb backend|clustering|idmap backend|idmap config[[:space:]]*\*[[:space:]]*:[[:space:]]*backend|private dir|ctdbd socket' local private_dir [ -n "$OCF_RESKEY_smb_private_dir" ] && private_dir="\tprivate dir = $OCF_RESKEY_smb_private_dir\n" local vfs_fileid local do_vfs do_vfs=0 if [ -n "$OCF_RESKEY_smb_fileid_algorithm" ]; then repl="${repl}|fileid:algorithm|fileid:mapping" vfs_fileid="\tfileid:algorithm = $OCF_RESKEY_smb_fileid_algorithm\n" if sed -n '/^[[:space:]]*\[global\]/,/^[[:space:]]*\[/p' $OCF_RESKEY_smb_conf | \ grep -Eq '^[[:space:]]*vfs objects'; then # vfs objects already specified, will append fileid to existing line do_vfs=1 else vfs_fileid="$vfs_fileid\tvfs objects = fileid\n" fi fi # Preserve permissions of smb.conf local idmap_config if grep -Eqs '^[[:space:]]*idmap backend[[:space:]]*=' $OCF_RESKEY_smb_conf; then idmap_config=old else idmap_config=new fi cp -a "$OCF_RESKEY_smb_conf" "$OCF_RESKEY_smb_conf.$$" awk ' /^[[:space:]]*\[/ { global = 0 } /^[[:space:]]*\[global\]/ { global = 1 } { if(global) { if ('$do_vfs' && $0 ~ /^[[:space:]]vfs objects/ && $0 !~ /fileid/) { print $0" fileid" } else if ($0 !~ /^[[:space:]]*('"$repl"')/) { print } } else { print } }' "$OCF_RESKEY_smb_conf" | sed "/^[[:space:]]*\[global\]/ a\\ \t# CTDB-RA: Begin auto-generated section (do not change below)\n\ \tpassdb backend = $OCF_RESKEY_smb_passdb_backend\n\ \tclustering = yes\n\ \tctdbd socket = $OCF_RESKEY_ctdb_socket\n$private_dir$vfs_fileid\ \t# CTDB-RA: End auto-generated section (do not change above)" > "$OCF_RESKEY_smb_conf.$$" if [ "$idmap_config" = "old" ]; then sed -i "/^[[:space:]]*clustering = yes/ a\\ \tidmap backend = $OCF_RESKEY_smb_idmap_backend" $OCF_RESKEY_smb_conf.$$ else sed -i "/^[[:space:]]*clustering = yes/ a\\ \tidmap config * : backend = $OCF_RESKEY_smb_idmap_backend" $OCF_RESKEY_smb_conf.$$ fi dd conv=notrunc,fsync of="$OCF_RESKEY_smb_conf.$$" if=/dev/null >/dev/null 2>&1 mv "$OCF_RESKEY_smb_conf.$$" "$OCF_RESKEY_smb_conf" } # Get rid of that section we added cleanup_smb_conf() { ocf_is_true "$OCF_RESKEY_ctdb_manages_samba" || return 0 # preserve permissions of smb.conf cp -a "$OCF_RESKEY_smb_conf" "$OCF_RESKEY_smb_conf.$$" sed '/# CTDB-RA: Begin/,/# CTDB-RA: End/d' "$OCF_RESKEY_smb_conf" > "$OCF_RESKEY_smb_conf.$$" mv "$OCF_RESKEY_smb_conf.$$" "$OCF_RESKEY_smb_conf" } append_conf() { local file_path="$1" shift [ -n "$2" ] && echo "$1=$2" >> "$file_path" } generate_ctdb_config() { local ctdb_config="$OCF_RESKEY_ctdb_config_dir/ctdb.conf" # Backup existing config if we're not already using an auto-generated one grep -qa '# CTDB-RA: Auto-generated' $ctdb_config || cp -p $ctdb_config ${ctdb_config}.ctdb-ra-orig if [ $? -ne 0 ]; then ocf_log warn "Unable to backup $ctdb_config to ${ctdb_config}.ctdb-ra-orig" fi local log_option="file:$OCF_RESKEY_ctdb_logfile" if [ "$OCF_RESKEY_ctdb_logfile" = "syslog" ]; then log_option="syslog" fi local start_as_disabled="false" ocf_is_true "$OCF_RESKEY_ctdb_start_as_disabled" && start_as_disabled="true" local dbdir_volatile="$OCF_RESKEY_ctdb_dbdir/volatile" [ -d "$dbdir_volatile" ] || mkdir -p "$dbdir_volatile" 2>/dev/null local dbdir_persistent="$OCF_RESKEY_ctdb_dbdir/persistent" [ -d "$dbdir_persistent" ] || mkdir -p "$dbdir_persistent" 2>/dev/null local dbdir_state="$OCF_RESKEY_ctdb_dbdir/state" [ -d "$dbdir_state" ] || mkdir -p "$dbdir_state" 2>/dev/null cat >$ctdb_config <$script_options <$CTDB_SYSCONFIG </dev/null # set nofile ulimit for ctdbd process if [ -n "$OCF_RESKEY_ctdb_max_open_files" ]; then ulimit -n "$OCF_RESKEY_ctdb_max_open_files" fi # Start her up invoke_ctdbd "$version" if [ $? -ne 0 ]; then # cleanup smb.conf cleanup_smb_conf ocf_exit_reason "Failed to execute $OCF_RESKEY_ctdbd_binary." return $OCF_ERR_GENERIC else # Wait a bit for CTDB to stabilize # (until start times out if necessary) while true; do # Initial sleep is intentional (ctdb init script # has sleep after ctdbd start, but before invoking # ctdb to talk to it) sleep 1 status=$(invoke_ctdb status 2>/dev/null) if [ $? -ne 0 ]; then # CTDB will be running, kill it before returning ctdb_stop ocf_exit_reason "Can't invoke $OCF_RESKEY_ctdb_binary status" return $OCF_ERR_GENERIC fi if ! echo "$status" | grep -qs 'UNHEALTHY (THIS'; then # Status does not say this node is unhealthy, # so we're good to go. Do a bit of final # setup and (hopefully) return success. set_ctdb_variables return $? fi done fi # ctdbd will (or can) actually still be running at this point, so kill it ctdb_stop ocf_exit_reason "Timeout waiting for CTDB to stabilize" return $OCF_ERR_GENERIC } ctdb_stop() { # Do nothing if already stopped pkill -0 -f "$OCF_RESKEY_ctdbd_binary" || return $OCF_SUCCESS # Tell it to die nicely invoke_ctdb shutdown >/dev/null 2>&1 rv=$? # No more Mr. Nice Guy count=0 while pkill -0 -f "$OCF_RESKEY_ctdbd_binary" ; do sleep 1 count=$((count + 1)) [ $count -gt 10 ] && { ocf_log info "killing ctdbd " pkill -9 -f "$OCF_RESKEY_ctdbd_binary" pkill -9 -f "${OCF_RESKEY_ctdb_config_dir}/events" } done # Cleanup smb.conf cleanup_smb_conf # It was a clean shutdown, return success [ $rv -eq $OCF_SUCCESS ] && return $OCF_SUCCESS # Unclean shutdown, return success if there's no ctdbds left (we # killed them forcibly, but at least they're good and dead). pkill -0 -f "$OCF_RESKEY_ctdbd_binary" || return $OCF_SUCCESS # Problem: ctdb shutdown didn't work and neither did some vigorous # kill -9ing. Only thing to do is report failure. return $OCF_ERR_GENERIC } ctdb_monitor() { local status # "ctdb status" exits non-zero if CTDB isn't running. # It can also exit non-zero if there's a timeout (ctdbd blocked, # stalled, massive load, or otherwise wedged). If it's actually # not running, STDERR will say "Errno:Connection refused(111)", # whereas if it's wedged, it'll say various other unpleasant things. status=$(invoke_ctdb status 2>&1) if [ $? -ne 0 ]; then if echo "$status" | grep -qs 'Connection refused'; then return $OCF_NOT_RUNNING elif echo "$status" | grep -qs 'No such file or directory'; then return $OCF_NOT_RUNNING elif echo $status | grep -qs 'connect() failed'; then return $OCF_NOT_RUNNING else ocf_exit_reason "CTDB status call failed: $status" return $OCF_ERR_GENERIC fi fi if echo "$status" | grep -Eqs '(OK|DISABLED) \(THIS'; then return $OCF_SUCCESS fi ocf_exit_reason "CTDB status is bad: $status" return $OCF_ERR_GENERIC } ctdb_validate() { # Required binaries for binary in pkill; do check_binary $binary done if ocf_is_true "$OCF_RESKEY_ctdb_manages_samba" && [ ! -f "$OCF_RESKEY_smb_conf" ]; then ocf_exit_reason "Samba config file '$OCF_RESKEY_smb_conf' does not exist." return $OCF_ERR_INSTALLED fi if [ -f "${OCF_RESKEY_ctdb_config_dir}/public_addresses" ]; then - ocf_log warn "CTDB file '${OCF_RESKEY_ctdb_config_dir}/public_addresses' exists - CTDB will try to manage IP failover!" + ocf_log info "CTDB file '${OCF_RESKEY_ctdb_config_dir}/public_addresses' exists - CTDB will try to manage IP failover!" fi if [ ! -f "$OCF_RESKEY_ctdb_config_dir/nodes" ]; then ocf_exit_reason "$OCF_RESKEY_ctdb_config_dir/nodes does not exist." return $OCF_ERR_ARGS fi if [ -z "$OCF_RESKEY_ctdb_recovery_lock" ]; then ocf_exit_reason "ctdb_recovery_lock not specified." return $OCF_ERR_CONFIGURED fi if [ "${OCF_RESKEY_ctdb_recovery_lock:0:1}" == '!' ]; then # '!' prefix means recovery lock is handled via a helper binary binary="${OCF_RESKEY_ctdb_recovery_lock:1}" binary="${binary%% *}" # trim any parameters if [ -z "$binary" ]; then ocf_exit_reason "ctdb_recovery_lock invalid helper" return $OCF_ERR_CONFIGURED fi check_binary "${binary}" else lock_dir=$(dirname "$OCF_RESKEY_ctdb_recovery_lock") touch "$lock_dir/$$" 2>/dev/null if [ $? != 0 ]; then ocf_exit_reason "Directory for lock file '$OCF_RESKEY_ctdb_recovery_lock' does not exist, or is not writable." return $OCF_ERR_ARGS fi rm "$lock_dir/$$" fi return $OCF_SUCCESS } case $__OCF_ACTION in meta-data) meta_data exit $OCF_SUCCESS ;; start) ctdb_start;; stop) ctdb_stop;; monitor) ctdb_monitor;; validate-all) ctdb_validate;; usage|help) ctdb_usage exit $OCF_SUCCESS ;; *) ctdb_usage exit $OCF_ERR_UNIMPLEMENTED ;; esac rc=$? ocf_log debug "${OCF_RESOURCE_INSTANCE} $__OCF_ACTION : $rc" exit $rc diff --git a/heartbeat/Filesystem b/heartbeat/Filesystem index 44270ad98..65088029e 100755 --- a/heartbeat/Filesystem +++ b/heartbeat/Filesystem @@ -1,1070 +1,1066 @@ #!/bin/sh # # Support: users@clusterlabs.org # License: GNU General Public License (GPL) # # Filesystem # Description: Manages a Filesystem on a shared storage medium. # Original Author: Eric Z. Ayers (eric.ayers@compgen.com) # Original Release: 25 Oct 2000 # # usage: ./Filesystem {start|stop|status|monitor|validate-all|meta-data} # # OCF parameters are as below: # OCF_RESKEY_device # OCF_RESKEY_directory # OCF_RESKEY_fstype # OCF_RESKEY_options # OCF_RESKEY_statusfile_prefix # OCF_RESKEY_run_fsck # OCF_RESKEY_fast_stop # OCF_RESKEY_force_clones # #OCF_RESKEY_device : name of block device for the filesystem. e.g. /dev/sda1, /dev/md0 # Or a -U or -L option for mount, or an NFS mount specification #OCF_RESKEY_directory : the mount point for the filesystem #OCF_RESKEY_fstype : optional name of the filesystem type. e.g. ext2 #OCF_RESKEY_options : options to be given to the mount command via -o #OCF_RESKEY_statusfile_prefix : the prefix used for a status file for monitoring #OCF_RESKEY_run_fsck : fsck execution mode: auto(default)/force/no #OCF_RESKEY_fast_stop : fast stop: yes(default)/no #OCF_RESKEY_force_clones : allow running the resource as clone. e.g. local xfs mounts # for each brick in a glusterfs setup # # # This assumes you want to manage a filesystem on a shared (SCSI) bus, # on a replicated device (such as DRBD), or a network filesystem (such # as NFS or Samba). # # Do not put this filesystem in /etc/fstab. This script manages all of # that for you. # # NOTE: If 2 or more nodes mount the same file system read-write, and # that file system is not designed for that specific purpose # (such as GFS or OCFS2), and is not a network file system like # NFS or Samba, then the filesystem is going to become # corrupted. # # As a result, you should use this together with the stonith # option and redundant, independent communications paths. # # If you don't do this, don't blame us when you scramble your # disk. ####################################################################### # Initialization: : ${OCF_FUNCTIONS_DIR=${OCF_ROOT}/lib/heartbeat} . ${OCF_FUNCTIONS_DIR}/ocf-shellfuncs # Defaults DFLT_STATUSDIR=".Filesystem_status/" # Parameter defaults OCF_RESKEY_device_default="" OCF_RESKEY_directory_default="" OCF_RESKEY_fstype_default="" OCF_RESKEY_options_default="" OCF_RESKEY_statusfile_prefix_default="${DFLT_STATUSDIR}" OCF_RESKEY_run_fsck_default="auto" OCF_RESKEY_fast_stop_default="no" OCF_RESKEY_force_clones_default="false" OCF_RESKEY_force_unmount_default="true" # RHEL specific defaults if is_redhat_based; then get_os_ver ocf_version_cmp "$VER" "9.0" 2>/dev/null case "$?" in # RHEL >= 9 1|2) OCF_RESKEY_force_unmount_default="safe";; # RHEL < 9 and fallback if ocf_version_cmp() fails *) OCF_RESKEY_fast_stop_default="yes";; esac fi : ${OCF_RESKEY_device=${OCF_RESKEY_device_default}} : ${OCF_RESKEY_directory=${OCF_RESKEY_directory_default}} : ${OCF_RESKEY_fstype=${OCF_RESKEY_fstype_default}} : ${OCF_RESKEY_options=${OCF_RESKEY_options_default}} : ${OCF_RESKEY_statusfile_prefix=${OCF_RESKEY_statusfile_prefix_default}} : ${OCF_RESKEY_run_fsck=${OCF_RESKEY_run_fsck_default}} if [ -z "${OCF_RESKEY_fast_stop}" ]; then case "$OCF_RESKEY_fstype" in gfs2) OCF_RESKEY_fast_stop="no";; *) OCF_RESKEY_fast_stop=${OCF_RESKEY_fast_stop_default};; esac fi : ${OCF_RESKEY_force_clones=${OCF_RESKEY_force_clones_default}} : ${OCF_RESKEY_force_unmount=${OCF_RESKEY_force_unmount_default}} # Variables used by multiple methods HOSTOS=$(uname) TAB=' ' # The status file is going to an extra directory, by default # prefix=${OCF_RESKEY_statusfile_prefix} : ${prefix:=$DFLT_STATUSDIR} suffix="${OCF_RESOURCE_INSTANCE}" [ "$OCF_RESKEY_CRM_meta_clone" ] && suffix="${suffix}_$OCF_RESKEY_CRM_meta_clone" suffix="${suffix}_$(uname -n)" STATUSFILE="${OCF_RESKEY_directory}/$prefix$suffix" ####################################################################### usage() { cat <<-EOT usage: $0 {start|stop|status|monitor|validate-all|meta-data} EOT } meta_data() { cat < 1.0 Resource script for Filesystem. It manages a Filesystem on a shared storage medium. The standard monitor operation of depth 0 (also known as probe) checks if the filesystem is mounted. If you want deeper tests, set OCF_CHECK_LEVEL to one of the following values: 10: read first 16 blocks of the device (raw read) This doesn't exercise the filesystem at all, but the device on which the filesystem lives. This is noop for non-block devices such as NFS, SMBFS, or bind mounts. 20: test if a status file can be written and read The status file must be writable by root. This is not always the case with an NFS mount, as NFS exports usually have the "root_squash" option set. In such a setup, you must either use read-only monitoring (depth=10), export with "no_root_squash" on your NFS server, or grant world write permissions on the directory where the status file is to be placed. Manages filesystem mounts The name of block device for the filesystem, or -U, -L options for mount, or NFS mount specification. NOTE: On Linux /dev/disk/by-{uuid,label}/ are preferred to -U/-L. block device The mount point for the filesystem. mount point The type of filesystem to be mounted. filesystem type Any extra options to be given as -o options to mount. For bind mounts, add "bind" here and set fstype to "none". We will do the right thing for options such as "bind,ro". options The prefix to be used for a status file for resource monitoring with depth 20. If you don't specify this parameter, all status files will be created in a separate directory. status file prefix Specify how to decide whether to run fsck or not. "auto" : decide to run fsck depending on the fstype(default) "force" : always run fsck regardless of the fstype "no" : do not run fsck ever. run_fsck Normally, we expect no users of the filesystem and the stop operation to finish quickly. If you cannot control the filesystem users easily and want to prevent the stop action from failing, then set this parameter to "no" and add an appropriate timeout for the stop operation. This defaults to "no" for GFS2 filesystems. fast stop The use of a clone setup for local filesystems is forbidden by default. For special setups like glusterfs, cloning a mount of a local device with a filesystem like ext4 or xfs independently on several nodes is a valid use case. Only set this to "true" if you know what you are doing! allow running as a clone, regardless of filesystem type This option allows specifying how to handle processes that are currently accessing the mount directory. "true" : Kill processes accessing mount point "safe" : Kill processes accessing mount point using methods that avoid functions that could potentially block during process detection "false" : Do not kill any processes. The 'safe' option uses shell logic to walk the /procs/ directory for pids using the mount point while the default option uses the fuser cli tool. fuser is known to perform operations that can potentially block if unresponsive nfs mounts are in use on the system. Kill processes before unmount END } # # Make sure the kernel does the right thing with the FS buffers # This function should be called after unmounting and before mounting # It may not be necessary in 2.4 and later kernels, but it shouldn't hurt # anything either... # # It's really a bug that you have to do this at all... # flushbufs() { if have_binary $BLOCKDEV ; then if [ "$blockdevice" = "yes" ] ; then $BLOCKDEV --flushbufs $1 return $? fi fi return 0 } # Take advantage of /etc/mtab if present, use portable mount command # otherwise. Normalize format to "dev mountpoint fstype". is_bind_mount() { echo "$options" | grep -w bind >/dev/null 2>&1 } list_mounts() { local inpf="" local mount_list="" local check_list="x" if [ -e "/proc/mounts" ] && ! is_bind_mount; then inpf=/proc/mounts elif [ -f "/etc/mtab" -a -r "/etc/mtab" ]; then inpf=/etc/mtab fi # Make sure that the mount list has not been changed while reading. while [ "$mount_list" != "$check_list" ]; do check_list="$mount_list" if [ "$inpf" ]; then # ... # Spaces in device or mountpoint are octal \040 in $inpf # Convert literal spaces (field separators) to tabs mount_list=$(cut -d' ' -f1,2,3 < $inpf | tr ' ' "$TAB") else # on type ... # Use tabs as field separators match_string='\(.*\) on \(.*\) type \([^[:space:]]\)' replace_string="\\1${TAB}\\3${TAB}\\5" mount_list=$($MOUNT | sed "s/$match_string/$replace_string/g") fi done # Convert octal \040 to space characters printf "$mount_list" } determine_blockdevice() { if [ $blockdevice = "yes" ]; then return fi # Get the current real device name, if possible. # (specified devname could be -L or -U...) case "$FSTYPE" in nfs4|nfs|efs|smbfs|cifs|glusterfs|ceph|tmpfs|overlay|overlayfs|rozofs|zfs|cvfs|none|lustre) : ;; *) match_string="${TAB}${CANONICALIZED_MOUNTPOINT}${TAB}" DEVICE=$(list_mounts | grep "$match_string" | cut -d"$TAB" -f1) if [ -b "$DEVICE" ]; then blockdevice=yes fi ;; esac } # Lists all filesystems potentially mounted under a given path, # excluding the path itself. list_submounts() { list_mounts | grep "${TAB}${1}/" | cut -d"$TAB" -f2 | sort -r } # Lists all bind mounts of a given file system, # excluding the path itself. list_bindmounts() { if is_bind_mount; then # skip bind mount # we should not umount the original file system via a bind mount return fi match_string="${TAB}${1}${TAB}" if list_mounts | grep "$match_string" >/dev/null 2>&1; then mount_disk=$(list_mounts | grep "$match_string" | cut -d"$TAB" -f1) else return fi if [ -b "$mount_disk" ]; then list_mounts | grep "$mount_disk" | grep -v "$match_string" | cut -d"$TAB" -f2 | sort -r fi } # kernels < 2.6.26 can't handle bind remounts bind_kernel_check() { echo "$options" | grep -w ro >/dev/null 2>&1 || return uname -r | awk -F. ' $1==2 && $2==6 { sub("[^0-9].*","",$3); if ($3<26) exit(1); }' [ $? -ne 0 ] && ocf_log warn "kernel $(uname -r) cannot handle read only bind mounts" } bind_root_mount_check() { if [ "$(df -P "$1" | awk 'END{print $6}')" = "/" ]; then return 1 else return 0 fi } bind_mount() { if is_bind_mount && [ "$options" != "-o bind" ] then bind_kernel_check bind_opts=$(echo "$options" | sed 's/bind/remount/') $MOUNT $bind_opts "$MOUNTPOINT" else true # make sure to return OK fi } is_option() { echo "$OCF_RESKEY_options" | grep -w "$1" >/dev/null 2>&1 } is_fsck_needed() { case $OCF_RESKEY_run_fsck in force) true;; no) false;; ""|auto) case "$FSTYPE" in ext4|ext4dev|ext3|reiserfs|reiser4|nss|xfs|jfs|vfat|fat|nfs4|nfs|efs|cifs|smbfs|ocfs2|gfs2|none|lustre|glusterfs|ceph|tmpfs|overlay|overlayfs|rozofs|zfs|cvfs) false;; *) true;; esac;; *) ocf_log warn "Invalid parameter value for fsck: '$OCF_RESKEY_run_fsck'; setting to 'auto'" OCF_RESKEY_run_fsck="auto" is_fsck_needed;; esac } fstype_supported() { local support="$FSTYPE" local rc if [ "X${HOSTOS}" = "XOpenBSD" ];then # skip checking /proc/filesystems for obsd return $OCF_SUCCESS fi if [ -z "$FSTYPE" -o "$FSTYPE" = none ]; then : No FSTYPE specified, rely on the system has the right file-system support already return $OCF_SUCCESS fi # support fuse-filesystems (e.g. GlusterFS) and Amazon Elastic File # System (EFS) case "$FSTYPE" in fuse.*|glusterfs|rozofs) support="fuse";; efs) support="nfs4";; esac if [ "$support" != "$FSTYPE" ]; then ocf_log info "Checking support for $FSTYPE as \"$support\"" fi grep -w "$support"'$' /proc/filesystems >/dev/null if [ $? -eq 0 ]; then # found the fs type return $OCF_SUCCESS fi # if here, we should attempt to load the module and then # check the if the filesystem support exists again. $MODPROBE $support >/dev/null if [ $? -ne 0 ]; then ocf_exit_reason "Couldn't find filesystem $support in /proc/filesystems and failed to load kernel module" return $OCF_ERR_INSTALLED fi # It is possible for the module to load and not be complete initialized # before we check /proc/filesystems again. Give this a few trys before # giving up entirely. for try in $(seq 5); do grep -w "$support"'$' /proc/filesystems >/dev/null if [ $? -eq 0 ] ; then # yes. found the filesystem after doing the modprobe return $OCF_SUCCESS fi ocf_log debug "Unable to find support for $support in /proc/filesystems after modprobe, trying again" sleep 1 done ocf_exit_reason "Couldn't find filesystem $support in /proc/filesystems" return $OCF_ERR_INSTALLED } # # In the case a fresh filesystem is just created from another # node on the shared storage, and is not visible yet. Then try # partprobe to refresh /dev/disk/by-{label,uuid}/* up to date. # # DEVICE can be /dev/xxx, -U, -L # trigger_udev_rules_if_needed() { local refresh_flag="no" local tmp local timeout if [ $blockdevice = "yes" ]; then tmp="$DEVICE" if [ "$DEVICE" != "/dev/null" -a ! -b "$DEVICE" ] ; then refresh_flag="yes" fi else tmp="$(echo $DEVICE|awk '{$1=""; print substr($0,2)}')" case "$DEVICE" in -U*|--uuid*) tmp="/dev/disk/by-uuid/$tmp" ;; -L*|--label*) tmp="/dev/disk/by-label/$tmp" ;; *) # bind mount? return ;; esac [ ! -b "$tmp" ] && refresh_flag="yes" fi [ "$refresh_flag" = "no" ] && return have_binary partprobe && partprobe >/dev/null 2>&1 timeout=${OCF_RESKEY_CRM_meta_timeout:="60000"} timeout=$((timeout/1000)) have_binary udevadm && udevadm settle -t $timeout --exit-if-exists=$tmp return $? } # # START: Start up the filesystem # Filesystem_start() { # Check if there are any mounts mounted under the mountpoint match_string="${TAB}${CANONICALIZED_MOUNTPOINT}" if list_mounts | grep -E "$match_string/\w+" >/dev/null 2>&1; then ocf_log err "There is one or more mounts mounted under $MOUNTPOINT." return $OCF_ERR_CONFIGURED fi # See if the device is already mounted. if Filesystem_status >/dev/null 2>&1 ; then ocf_log info "Filesystem $MOUNTPOINT is already mounted." return $OCF_SUCCESS fi fstype_supported || exit $OCF_ERR_INSTALLED # Check the filesystem & auto repair. # NOTE: Some filesystem types don't need this step... Please modify # accordingly trigger_udev_rules_if_needed if [ $blockdevice = "yes" ]; then if [ "$DEVICE" != "/dev/null" -a ! -b "$DEVICE" ] ; then ocf_exit_reason "Couldn't find device [$DEVICE]. Expected /dev/??? to exist" exit $OCF_ERR_INSTALLED fi if is_fsck_needed; then ocf_log info "Starting filesystem check on $DEVICE" if [ -z "$FSTYPE" ]; then $FSCK -p "$DEVICE" else $FSCK -t "$FSTYPE" -p "$DEVICE" fi # NOTE: if any errors at all are detected, it returns non-zero # if the error is >= 4 then there is a big problem if [ $? -ge 4 ]; then ocf_exit_reason "Couldn't successfully fsck filesystem for $DEVICE" return $OCF_ERR_GENERIC fi fi fi [ -d "$MOUNTPOINT" ] || ocf_run mkdir -p "$MOUNTPOINT" if [ ! -d "$MOUNTPOINT" ] ; then ocf_exit_reason "Couldn't find directory [$MOUNTPOINT] to use as a mount point" exit $OCF_ERR_INSTALLED fi flushbufs "$DEVICE" # Mount the filesystem. case "$FSTYPE" in none) $MOUNT $options $device_opt "$DEVICE" "$MOUNTPOINT" && bind_mount ;; "") $MOUNT $options $device_opt "$DEVICE" "$MOUNTPOINT" ;; *) $MOUNT -t "$FSTYPE" $options $device_opt "$DEVICE" "$MOUNTPOINT" ;; esac if [ $? -ne 0 ]; then ocf_exit_reason "Couldn't mount device [$DEVICE] as $MOUNTPOINT" return $OCF_ERR_GENERIC fi return $OCF_SUCCESS } # end of Filesystem_start get_pids() { local dir=$1 local procs local mmap_procs if is_bind_mount && ocf_is_true "$FORCE_UNMOUNT" && ! bind_root_mount_check "$DEVICE"; then ocf_log debug "Change force_umount from '$FORCE_UNMOUNT' to 'safe'" FORCE_UNMOUNT=safe fi if ocf_is_true "$FORCE_UNMOUNT"; then if [ "X${HOSTOS}" = "XOpenBSD" ];then fstat | grep $dir | awk '{print $3}' else $FUSER -m $dir 2>/dev/null fi elif [ "$FORCE_UNMOUNT" = "safe" ]; then procs=$(find /proc/[0-9]*/ -type l -lname "${dir}/*" -or -lname "${dir}" 2>/dev/null | awk -F/ '{print $3}') mmap_procs=$(grep " ${dir}/" /proc/[0-9]*/maps | awk -F/ '{print $3}') printf "${procs}\n${mmap_procs}" | sort | uniq fi } signal_processes() { local dir=$1 local sig=$2 local pids pid # fuser returns a non-zero return code if none of the # specified files is accessed or in case of a fatal # error. pids=$(get_pids "$dir") if [ -z "$pids" ]; then ocf_log info "No processes on $dir were signalled. force_unmount is set to '$FORCE_UNMOUNT'" return fi for pid in $pids; do ocf_log info "sending signal $sig to: $(ps -f $pid | tail -1)" kill -s $sig $pid done } try_umount() { local SUB="$1" $UMOUNT $umount_force "$SUB" list_mounts | grep "${TAB}${SUB}${TAB}" >/dev/null 2>&1 || { ocf_log info "unmounted $SUB successfully" return $OCF_SUCCESS } return $OCF_ERR_GENERIC } fs_stop() { local SUB="$1" timeout=$2 sig cnt for sig in TERM KILL; do cnt=$((timeout/2)) # try half time with TERM while [ $cnt -gt 0 ]; do try_umount "$SUB" && return $OCF_SUCCESS ocf_exit_reason "Couldn't unmount $SUB; trying cleanup with $sig" signal_processes "$SUB" $sig cnt=$((cnt-1)) sleep 1 done done return $OCF_ERR_GENERIC } # # STOP: Unmount the filesystem # Filesystem_stop() { # See if the device is currently mounted Filesystem_status >/dev/null 2>&1 if [ $? -eq $OCF_NOT_RUNNING ]; then # Already unmounted, wonderful. rc=$OCF_SUCCESS else # Wipe the status file, but continue with a warning if # removal fails -- the file system might be read only if [ $OCF_CHECK_LEVEL -eq 20 ]; then rm -f "${STATUSFILE}" if [ $? -ne 0 ]; then ocf_log warn "Failed to remove status file ${STATUSFILE}." fi fi # Determine the real blockdevice this is mounted on (if # possible) prior to unmounting. determine_blockdevice # For networked filesystems, there's merit in trying -f: case "$FSTYPE" in nfs4|nfs|efs|cifs|smbfs) umount_force="-f" ;; esac # Umount all sub-filesystems mounted under $MOUNTPOINT/ too. local timeout while read SUB; do ocf_log info "Trying to unmount $SUB" if ocf_is_true "$FAST_STOP"; then timeout=6 else timeout=${OCF_RESKEY_CRM_meta_timeout:="20000"} timeout=$((timeout/1000)) fi fs_stop "$SUB" $timeout rc=$? if [ $rc -ne $OCF_SUCCESS ]; then ocf_exit_reason "Couldn't unmount $SUB, giving up!" fi done <<-EOF $(list_submounts "$CANONICALIZED_MOUNTPOINT"; \ list_bindmounts "$CANONICALIZED_MOUNTPOINT"; \ echo $CANONICALIZED_MOUNTPOINT) EOF fi flushbufs "$DEVICE" return $rc } # end of Filesystem_stop # # STATUS: is the filesystem mounted or not? # Filesystem_status() { match_string="${TAB}${CANONICALIZED_MOUNTPOINT}${TAB}" if list_mounts | grep "$match_string" >/dev/null 2>&1; then rc=$OCF_SUCCESS msg="$MOUNTPOINT is mounted (running)" else rc=$OCF_NOT_RUNNING msg="$MOUNTPOINT is unmounted (stopped)" fi # Special case "monitor" to check whether the UUID cached and # on-disk still match? case "$OP" in status) ocf_log info "$msg";; esac return $rc } # end of Filesystem_status # Note: the read/write tests below will stall in case the # underlying block device (or in the case of a NAS mount, the # NAS server) has gone away. In that case, if I/O does not # return to normal in time, the operation hits its timeout # and it is up to the CRM to initiate appropriate recovery # actions (such as fencing the node). # # MONITOR 10: read the device # Filesystem_monitor_10() { if [ "$blockdevice" = "no" ] ; then ocf_log warn "$DEVICE is not a block device, monitor 10 is noop" return $OCF_SUCCESS fi dd_opts="iflag=direct bs=4k count=1" err_output=$(dd if="$DEVICE" $dd_opts 2>&1 >/dev/null) if [ $? -ne 0 ]; then ocf_exit_reason "Failed to read device $DEVICE" ocf_log err "dd said: $err_output" return $OCF_ERR_GENERIC fi return $OCF_SUCCESS } # # MONITOR 20: write and read a status file # Filesystem_monitor_20() { if [ "$blockdevice" = "no" ] ; then # O_DIRECT not supported on cifs/smbfs dd_opts="oflag=sync bs=4k conv=fsync,sync" else # Writing to the device in O_DIRECT mode is imperative # to bypass caches. dd_opts="oflag=direct,sync bs=4k conv=fsync,sync" fi status_dir=$(dirname "$STATUSFILE") [ -d "$status_dir" ] || mkdir -p "$status_dir" err_output=$(echo "${OCF_RESOURCE_INSTANCE}" | dd of="${STATUSFILE}" $dd_opts 2>&1) if [ $? -ne 0 ]; then ocf_exit_reason "Failed to write status file ${STATUSFILE}" ocf_log err "dd said: $err_output" return $OCF_ERR_GENERIC fi test -f "${STATUSFILE}" if [ $? -ne 0 ]; then ocf_exit_reason "Cannot stat the status file ${STATUSFILE}" return $OCF_ERR_GENERIC fi cat "${STATUSFILE}" > /dev/null if [ $? -ne 0 ]; then ocf_exit_reason "Cannot read the status file ${STATUSFILE}" return $OCF_ERR_GENERIC fi return $OCF_SUCCESS } Filesystem_monitor() { Filesystem_status rc=$? if [ $rc -ne $OCF_SUCCESS ]; then return $rc fi if [ $rc -eq $OCF_SUCCESS -a $OCF_CHECK_LEVEL -gt 0 ]; then case "$OCF_CHECK_LEVEL" in 10) Filesystem_monitor_10; rc=$?;; 20) Filesystem_monitor_20; rc=$?;; *) ocf_exit_reason "unsupported monitor level $OCF_CHECK_LEVEL" rc=$OCF_ERR_CONFIGURED ;; esac fi return $rc } # end of Filesystem_monitor # # VALIDATE_ALL: Are the instance parameters valid? # FIXME!! The only part that's useful is the return code. # This code always returns $OCF_SUCCESS (!) # FIXME!! Needs some tuning to match fstype_supported() (e.g., for # fuse). Can we just call fstype_supported() with a flag like # "no_modprobe" instead? # Filesystem_validate_all() { - if [ -n "$MOUNTPOINT" ] && [ ! -d "$MOUNTPOINT" ]; then - ocf_log warn "Mountpoint $MOUNTPOINT does not exist" - fi - # Check if the $FSTYPE is workable # NOTE: Without inserting the $FSTYPE module, this step may be imprecise # TODO: This is Linux specific crap. if [ ! -z "$FSTYPE" -a "$FSTYPE" != none ]; then cut -f2 /proc/filesystems | grep "^${FSTYPE}$" >/dev/null 2>&1 if [ $? -ne 0 ]; then modpath=/lib/modules/$(uname -r) moddep=$modpath/modules.dep # Do we have $FSTYPE in modules.dep? cut -d' ' -f1 $moddep \ | grep "^${modpath}.*${FSTYPE}\.k\?o:$" >/dev/null 2>&1 if [ $? -ne 0 ]; then ocf_log info "It seems we do not have $FSTYPE support" fi fi fi # If we are supposed to do monitoring with status files, then # we need a utility to write in O_DIRECT mode. if [ $OCF_CHECK_LEVEL -gt 0 ]; then check_binary dd # Note: really old coreutils version do not support # the "oflag" option for dd. We don't check for that # here. In case dd does not support oflag, monitor is # bound to fail, with dd spewing an error message to # the logs. On such systems, we must do without status # file monitoring. fi #TODO: How to check the $options ? return $OCF_SUCCESS } # # set the blockdevice variable to "no" or "yes" # set_blockdevice_var() { blockdevice=no # these are definitely not block devices case "$FSTYPE" in nfs4|nfs|efs|smbfs|cifs|none|glusterfs|ceph|tmpfs|overlay|overlayfs|rozofs|zfs|cvfs|lustre) return;; esac if $(is_option "loop"); then return fi case "$DEVICE" in --uuid=*|--uuid\ *|--label=*|--label\ *) device_opt=$(echo $DEVICE | sed "s/\([[:blank:]]\|=\).*//") DEVICE=$(echo $DEVICE | sed -E "s/$device_opt([[:blank:]]*|=)//") ;; -U*|-L*) # short versions of --uuid/--label device_opt=$(echo $DEVICE | cut -c1-2) DEVICE=$(echo $DEVICE | sed "s/$device_opt[[:blank:]]*//") ;; /dev/null) # Special case for BSC blockdevice=yes ;; *) if [ ! -b "$DEVICE" -a ! -d "$DEVICE" -a "X$OP" != Xstart ] ; then ocf_log warn "Couldn't find device [$DEVICE]. Expected /dev/??? to exist" fi if [ ! -d "$DEVICE" ]; then blockdevice=yes fi ;; esac } # Check the arguments passed to this script if [ $# -ne 1 ]; then usage exit $OCF_ERR_ARGS fi # Check the OCF_RESKEY_ environment variables... FORCE_UNMOUNT="yes" if [ -n "${OCF_RESKEY_force_unmount}" ]; then FORCE_UNMOUNT=$OCF_RESKEY_force_unmount fi DEVICE="$OCF_RESKEY_device" FSTYPE=$OCF_RESKEY_fstype if [ ! -z "$OCF_RESKEY_options" ]; then options="-o $OCF_RESKEY_options" fi FAST_STOP=${OCF_RESKEY_fast_stop:="yes"} OP=$1 # These operations do not require instance parameters case $OP in meta-data) meta_data exit $OCF_SUCCESS ;; usage) usage exit $OCF_SUCCESS ;; esac if [ x = x"$DEVICE" ]; then ocf_exit_reason "Please set OCF_RESKEY_device to the device to be managed" exit $OCF_ERR_CONFIGURED fi set_blockdevice_var # Normalize instance parameters: # It is possible that OCF_RESKEY_directory has one or even multiple trailing "/". # But the output of `mount` and /proc/mounts do not. if [ -z "$OCF_RESKEY_directory" ]; then if [ X$OP = "Xstart" -o $blockdevice = "no" ]; then ocf_exit_reason "Please specify the directory" exit $OCF_ERR_CONFIGURED fi else MOUNTPOINT="$(echo "$OCF_RESKEY_directory" | sed 's/\/*$//')" : ${MOUNTPOINT:=/} if [ -e "$MOUNTPOINT" ] ; then CANONICALIZED_MOUNTPOINT="$(readlink -f "$MOUNTPOINT")" if [ $? -ne 0 ]; then ocf_exit_reason "Could not canonicalize $MOUNTPOINT because readlink failed" exit $OCF_ERR_GENERIC fi else CANONICALIZED_MOUNTPOINT="$MOUNTPOINT" fi # At this stage, $MOUNTPOINT does not contain trailing "/" unless it is "/" # TODO: / mounted via Filesystem sounds dangerous. On stop, we'll # kill the whole system. Is that a good idea? fi # Check to make sure the utilites are found if [ "X${HOSTOS}" != "XOpenBSD" ];then check_binary $MODPROBE check_binary $FUSER fi check_binary $FSCK check_binary $MOUNT check_binary $UMOUNT if [ "$OP" != "monitor" ]; then ocf_log info "Running $OP for $DEVICE on $MOUNTPOINT" fi case $OP in status) Filesystem_status exit $? ;; monitor) Filesystem_monitor exit $? ;; validate-all) Filesystem_validate_all exit $? ;; stop) Filesystem_stop exit $? ;; esac CLUSTERSAFE=0 is_option "ro" && CLUSTERSAFE=2 case "$FSTYPE" in nfs4|nfs|efs|smbfs|cifs|none|gfs2|glusterfs|ceph|ocfs2|overlay|overlayfs|tmpfs|cvfs|lustre) CLUSTERSAFE=1 # this is kind of safe too ;; # add here CLUSTERSAFE=0 for all filesystems which are not # cluster aware and which, even if when mounted read-only, # could still modify parts of it such as journal/metadata ext4|ext4dev|ext3|reiserfs|reiser4|xfs|jfs) if ocf_is_true "$OCF_RESKEY_force_clones"; then CLUSTERSAFE=2 else CLUSTERSAFE=0 # these are not allowed fi ;; esac if ocf_is_clone; then case $CLUSTERSAFE in 0) ocf_exit_reason "DANGER! $FSTYPE on $DEVICE is NOT cluster-aware!" ocf_log err "DO NOT RUN IT AS A CLONE!" ocf_log err "Politely refusing to proceed to avoid data corruption." exit $OCF_ERR_CONFIGURED ;; 2) ocf_log warn "$FSTYPE on $DEVICE is NOT cluster-aware!" if ocf_is_true "$OCF_RESKEY_force_clones"; then ocf_log warn "But we'll let it run because we trust _YOU_ verified it's safe to do so." else ocf_log warn "But we'll let it run because it is mounted read-only." ocf_log warn "Please make sure that it's meta data is read-only too!" fi ;; esac fi case $OP in start) Filesystem_start ;; *) usage exit $OCF_ERR_UNIMPLEMENTED ;; esac exit $?