ovn/utilities/ovndb-servers.ocf

   1 #!/bin/bash
   2
   3 : ${OCF_FUNCTIONS_DIR=${OCF_ROOT}/lib/heartbeat}
   4 . ${OCF_FUNCTIONS_DIR}/ocf-shellfuncs
   5 : ${OVN_CTL_DEFAULT="/usr/share/openvswitch/scripts/ovn-ctl"}
   6 : ${NB_MASTER_PORT_DEFAULT="6641"}
   7 : ${NB_MASTER_PROTO_DEFAULT="tcp"}
   8 : ${SB_MASTER_PORT_DEFAULT="6642"}
   9 : ${SB_MASTER_PROTO_DEFAULT="tcp"}
  10 : ${MANAGE_NORTHD_DEFAULT="no"}
  11 CRM_MASTER="${HA_SBIN_DIR}/crm_master -l reboot"
  12 CRM_ATTR_REPL_INFO="${HA_SBIN_DIR}/crm_attribute --type crm_config --name OVN_REPL_INFO -s ovn_ovsdb_master_server"
  13 OVN_CTL=${OCF_RESKEY_ovn_ctl:-${OVN_CTL_DEFAULT}}
  14 MASTER_IP=${OCF_RESKEY_master_ip}
  15 NB_MASTER_PORT=${OCF_RESKEY_nb_master_port:-${NB_MASTER_PORT_DEFAULT}}
  16 NB_MASTER_PROTO=${OCF_RESKEY_nb_master_protocol:-${NB_MASTER_PROTO_DEFAULT}}
  17 SB_MASTER_PORT=${OCF_RESKEY_sb_master_port:-${SB_MASTER_PORT_DEFAULT}}
  18 SB_MASTER_PROTO=${OCF_RESKEY_sb_master_protocol:-${SB_MASTER_PROTO_DEFAULT}}
  19 MANAGE_NORTHD=${OCF_RESKEY_manage_northd:-${MANAGE_NORTHD_DEFAULT}}
  20
  21 # Invalid IP address is an address that can never exist in the network, as
  22 # mentioned in rfc-5737. The ovsdb servers connects to this IP address till
  23 # a master is promoted and the IPAddr2 resource is started.
  24 INVALID_IP_ADDRESS=192.0.2.254
  25
  26 host_name=$(ocf_local_nodename)
  27 : ${slave_score=5}
  28 : ${master_score=10}
  29
  30 ovsdb_server_metadata() {
  31     cat <<END
  32 <?xml version="1.0"?>
  33 <!DOCTYPE resource-agent SYSTEM "ra-api-1.dtd">
  34 <resource-agent name="ovsdb-server">
  35   <version>1.0</version>
  36
  37   <longdesc lang="en">
  38     This resource manages ovsdb-server.
  39   </longdesc>
  40
  41   <shortdesc lang="en">
  42     Manages ovsdb-server.
  43   </shortdesc>
  44
  45   <parameters>
  46
  47   <parameter name="ovn_ctl" unique="1">
  48   <longdesc lang="en">
  49   Location to the ovn-ctl script file
  50   </longdesc>
  51   <shortdesc lang="en">ovn-ctl script</shortdesc>
  52   <content type="string" default="${OVN_CTL_DEFAULT}" />
  53   </parameter>
  54
  55   <parameter name="master_ip" unique="1">
  56   <longdesc lang="en">
  57   The IP address resource which will be available on the master ovsdb server
  58   </longdesc>
  59   <shortdesc lang="en">master ip address</shortdesc>
  60   <content type="string" />
  61   </parameter>
  62
  63   <parameter name="nb_master_port" unique="1">
  64   <longdesc lang="en">
  65   The port which the master Northbound database server is listening
  66   </longdesc>
  67   <shortdesc lang="en">master Northbound database port</shortdesc>
  68   <content type="string" />
  69   </parameter>
  70
  71   <parameter name="nb_master_protocol" unique="1">
  72   <longdesc lang="en">
  73   The protocol which the master Northbound database server used, 'tcp' or 'ssl'.
  74   </longdesc>
  75   <shortdesc lang="en">master Northbound database protocol</shortdesc>
  76   <content type="string" />
  77   </parameter>
  78
  79   <parameter name="sb_master_port" unique="1">
  80   <longdesc lang="en">
  81   The port which the master Southbound database server is listening
  82   </longdesc>
  83   <shortdesc lang="en">master Southbound database port</shortdesc>
  84   <content type="string" />
  85   </parameter>
  86
  87   <parameter name="sb_master_protocol" unique="1">
  88   <longdesc lang="en">
  89   The protocol which the master Southbound database server used, 'tcp' or 'ssl'.
  90   </longdesc>
  91   <shortdesc lang="en">master Southbound database protocol</shortdesc>
  92   <content type="string" />
  93   </parameter>
  94
  95   <parameter name="manage_northd" unique="1">
  96   <longdesc lang="en">
  97   If set to yes, manages ovn-northd service. ovn-northd will be started in
  98   the master node.
  99   </longdesc>
 100   <shortdesc lang="en">manage ovn-northd service</shortdesc>
 101   <content type="string" />
 102   </parameter>
 103
 104   </parameters>
 105
 106   <actions>
 107     <action name="notify"       timeout="20s" />
 108     <action name="start"        timeout="30s" />
 109     <action name="stop"         timeout="20s" />
 110     <action name="promote"      timeout="50s" />
 111     <action name="demote"       timeout="50s" />
 112     <action name="monitor"      timeout="20s"  depth="0" interval="10s" />
 113     <action name="meta-data"    timeout="5s" />
 114     <action name="validate-all" timeout="20s" />
 115   </actions>
 116 </resource-agent>
 117 END
 118     exit $OCF_SUCCESS
 119 }
 120
 121 ovsdb_server_notify() {
 122     # requires the notify=true meta resource attribute
 123     local type_op="${OCF_RESKEY_CRM_meta_notify_type}-${OCF_RESKEY_CRM_meta_notify_operation}"
 124
 125     if [ "$type_op" != "post-promote" ]; then
 126         # We are only interested in specific events
 127         return $OCF_SUCCESS
 128     fi
 129
 130     ocf_log debug "ovndb_server: notified of event $type_op"
 131     if [ "x${OCF_RESKEY_CRM_meta_notify_promote_uname}" = "x${host_name}" ]; then
 132         # Record ourselves so that the agent has a better chance of doing
 133         # the right thing at startup
 134         ocf_log debug "ovndb_server: $host_name is the master"
 135         ${CRM_ATTR_REPL_INFO} -v "$host_name"
 136         if [ "$MANAGE_NORTHD" = "yes" ]; then
 137             # Startup ovn-northd service
 138             ${OVN_CTL} --ovn-manage-ovsdb=no start_northd
 139         fi
 140
 141     else
 142         if [ "$MANAGE_NORTHD" = "yes" ]; then
 143             # Stop ovn-northd service. Set --ovn-manage-ovsdb=no so that
 144             # ovn-ctl doesn't stop ovsdb-servers.
 145             ${OVN_CTL} --ovn-manage-ovsdb=no stop_northd
 146         fi
 147         # Synchronize with the new master
 148         ocf_log debug "ovndb_server: Connecting to the new master ${OCF_RESKEY_CRM_meta_notify_promote_uname}"
 149         ${OVN_CTL} demote_ovnnb --db-nb-sync-from-addr=${MASTER_IP} \
 150                                 --db-nb-sync-from-port=${NB_MASTER_PORT} \
 151                                 --db-nb-sync-from-proto=${NB_MASTER_PROTO}
 152         ${OVN_CTL} demote_ovnsb --db-sb-sync-from-addr=${MASTER_IP} \
 153                                 --db-sb-sync-from-port=${SB_MASTER_PORT} \
 154                                 --db-sb-sync-from-proto=${SB_MASTER_PROTO}
 155     fi
 156 }
 157
 158 ovsdb_server_usage() {
 159     cat <<END
 160 usage: $0 {start|stop|status|monitor|notify|validate-all|meta-data}
 161
 162 Expects to have a fully populated OCF RA-compliant environment set.
 163 END
 164     exit $1
 165 }
 166
 167 ovsdb_server_find_active_master() {
 168     # Operation sequence is Demote -> Stop -> Start -> Promote
 169     # At the point this is run, the only active masters will be
 170     # previous masters minus any that were scheduled to be demoted
 171
 172     for master in ${OCF_RESKEY_CRM_meta_notify_master_uname}; do
 173         found=0
 174         for old in ${OCF_RESKEY_CRM_meta_notify_demote_uname}; do
 175             if [ $master = $old ]; then
 176                 found=1
 177             fi
 178         done
 179         if [ $found = 0 ]; then
 180             # Rely on master-max=1
 181             # Pacemaker will demote any additional ones it finds before starting new copies
 182             echo "$master"
 183             return
 184         fi
 185     done
 186
 187     local expected_master=$($CRM_ATTR_REPL_INFO --query  -q 2>/dev/null)
 188     case "x${OCF_RESKEY_CRM_meta_notify_start_uname}x" in
 189         *${expected_master}*) echo "${expected_master}";; # The previous master is expected to start
 190     esac
 191 }
 192
 193 ovsdb_server_find_active_peers() {
 194     # Do we have any peers that are not stopping
 195     for peer in ${OCF_RESKEY_CRM_meta_notify_slave_uname}; do
 196         found=0
 197         for old in ${OCF_RESKEY_CRM_meta_notify_stop_uname}; do
 198             if [ $peer = $old ]; then
 199                 found=1
 200             fi
 201         done
 202         if [ $found = 0 ]; then
 203             # Rely on master-max=1
 204             # Pacemaker will demote any additional ones it finds before starting new copies
 205             echo "$peer"
 206             return
 207         fi
 208     done
 209 }
 210
 211 ovsdb_server_master_update() {
 212
 213     case $1 in
 214         $OCF_SUCCESS)
 215         $CRM_MASTER -v ${slave_score};;
 216         $OCF_RUNNING_MASTER)
 217             $CRM_MASTER -v ${master_score};;
 218         #*) $CRM_MASTER -D;;
 219     esac
 220 }
 221
 222 ovsdb_server_monitor() {
 223     ovsdb_server_check_status
 224     rc=$?
 225
 226     ovsdb_server_master_update $rc
 227     return $rc
 228 }
 229
 230 ovsdb_server_check_status() {
 231     local sb_status=`${OVN_CTL} status_ovnsb`
 232     local nb_status=`${OVN_CTL} status_ovnnb`
 233
 234     if [[ $sb_status == "running/backup" && $nb_status == "running/backup" ]]; then
 235         return $OCF_SUCCESS
 236     fi
 237
 238     if [[ $sb_status == "running/active" && $nb_status == "running/active" ]]; then
 239         return $OCF_RUNNING_MASTER
 240     fi
 241
 242     # TODO: What about service running but not in either state above?
 243     # Eg. a transient state where one db is "active" and the other
 244     # "backup"
 245
 246     return $OCF_NOT_RUNNING
 247 }
 248
 249 ovsdb_server_start() {
 250     ovsdb_server_check_status
 251     local status=$?
 252     # If not in stopped state, return
 253     if [ $status -ne $OCF_NOT_RUNNING ]; then
 254         return $status
 255     fi
 256
 257     local present_master=$(ovsdb_server_find_active_master)
 258
 259     set ${OVN_CTL}
 260
 261     set $@ --db-nb-addr=${MASTER_IP} --db-nb-port=${NB_MASTER_PORT}
 262     set $@ --db-sb-addr=${MASTER_IP} --db-sb-port=${SB_MASTER_PORT}
 263
 264     if [ "x${NB_MASTER_PROTO}" = xtcp ]; then
 265         set $@ --db-nb-create-insecure-remote=yes
 266     fi
 267
 268     if [ "x${SB_MASTER_PROTO}" = xtcp ]; then
 269         set $@ --db-sb-create-insecure-remote=yes
 270     fi
 271
 272     if [ "x${present_master}" = x ]; then
 273         # No master detected, or the previous master is not among the
 274         # set starting.
 275         #
 276         # Force all copies to come up as slaves by pointing them into
 277         # space and let pacemaker pick one to promote:
 278         #
 279         set $@ --db-nb-sync-from-addr=${INVALID_IP_ADDRESS} --db-sb-sync-from-addr=${INVALID_IP_ADDRESS}
 280
 281     elif [ ${present_master} != ${host_name} ]; then
 282         # An existing master is active, connect to it
 283         set $@ --db-nb-sync-from-addr=${MASTER_IP} --db-sb-sync-from-addr=${MASTER_IP}
 284         set $@ --db-nb-sync-from-port=${NB_MASTER_PORT}
 285         set $@ --db-nb-sync-from-proto=${NB_MASTER_PROTO}
 286         set $@ --db-sb-sync-from-port=${SB_MASTER_PORT}
 287         set $@ --db-sb-sync-from-proto=${SB_MASTER_PROTO}
 288     fi
 289
 290     $@ start_ovsdb
 291
 292     while [ 1 = 1 ]; do
 293         # It is important that we don't return until we're in a functional state
 294         ovsdb_server_monitor
 295         rc=$?
 296         case $rc in
 297             $OCF_SUCCESS)        return $rc;;
 298             $OCF_RUNNING_MASTER)
 299                 # When a slave node is promoted as master, the action would be
 300                 # STOP -> START -> PROMOTE.
 301                 # When the start action is called, it is possible for the
 302                 # ovsdb-server's to be started as active. This could happen
 303                 # if the node owns the $MASTER_IP. At this point, pacemaker
 304                 # has not promoted this node yet. So return OCF_SUCCESS.
 305                 # Let pacemaker promote it in subsequent actions.
 306                 # As per the OCF guidelines, only monitor action should return
 307                 # OCF_RUNNING_MASTER.
 308                 # http://www.linux-ha.org/doc/dev-guides/_literal_ocf_running_master_literal_8.html
 309                 return $OCF_SUCCESS;;
 310             $OCF_ERR_GENERIC)    return $rc;;
 311             # Otherwise loop, waiting for the service to start, until
 312             # the cluster times the operation out
 313         esac
 314         ocf_log warn "ovndb_servers: After starting ovsdb, status is $rc. Checking the status again"
 315     done
 316 }
 317
 318 ovsdb_server_stop() {
 319     if [ "$MANAGE_NORTHD" = "yes" ]; then
 320         # Stop ovn-northd service in case it was running. This is required
 321         # when the master is demoted. For other cases, it would be a no-op.
 322         # Set --ovn-manage-ovsdb=no so that ovn-ctl doesn't stop ovsdb-servers.
 323         ${OVN_CTL} --ovn-manage-ovsdb=no stop_northd
 324     fi
 325
 326     ovsdb_server_check_status
 327     case $? in
 328         $OCF_NOT_RUNNING)    return ${OCF_SUCCESS};;
 329     esac
 330
 331     ${OVN_CTL} stop_ovsdb
 332     ovsdb_server_master_update ${OCF_NOT_RUNNING}
 333
 334     while [ 1 = 1 ]; do
 335         # It is important that we don't return until we're stopped
 336         ovsdb_server_check_status
 337         rc=$?
 338         case $rc in
 339         $OCF_SUCCESS)
 340             # Loop, waiting for the service to stop, until the
 341             # cluster times the operation out
 342             ocf_log warn "ovndb_servers: Even after stopping, the servers seems to be running"
 343             ;;
 344         $OCF_NOT_RUNNING)
 345             return $OCF_SUCCESS
 346             ;;
 347         *)
 348             return $rc
 349             ;;
 350         esac
 351     done
 352
 353     return $OCF_ERR_GENERIC
 354 }
 355
 356 ovsdb_server_promote() {
 357     ovsdb_server_check_status
 358     rc=$?
 359     case $rc in
 360         ${OCF_SUCCESS}) ;;
 361         ${OCF_RUNNING_MASTER}) return ${OCF_SUCCESS};;
 362         *)
 363             ovsdb_server_master_update $OCF_RUNNING_MASTER
 364             return ${rc}
 365             ;;
 366     esac
 367
 368     ${OVN_CTL} promote_ovnnb
 369     ${OVN_CTL} promote_ovnsb
 370
 371     ocf_log debug "ovndb_servers: Promoting $host_name as the master"
 372     # Record ourselves so that the agent has a better chance of doing
 373     # the right thing at startup
 374     ${CRM_ATTR_REPL_INFO} -v "$host_name"
 375     ovsdb_server_master_update $OCF_RUNNING_MASTER
 376     return $OCF_SUCCESS
 377 }
 378
 379 ovsdb_server_demote() {
 380     ovsdb_server_check_status
 381     if [ $? = $OCF_NOT_RUNNING ]; then
 382         return $OCF_NOT_RUNNING
 383     fi
 384
 385     local present_master=$(ovsdb_server_find_active_master)
 386     local recorded_master=$($CRM_ATTR_REPL_INFO --query  -q 2>/dev/null)
 387
 388     ocf_log debug "ovndb_servers: Demoting $host_name, present master ${present_master}, recorded master ${recorded_master}"
 389     if [ "x${recorded_master}" = "x${host_name}" -a "x${present_master}" = x ]; then
 390         # We are the one and only master
 391         # This should be the "normal" case
 392         # The only way to be demoted is to call demote_ovn*
 393         #
 394         # The local database is only reset once we successfully
 395         # connect to the peer.  So specify one that doesn't exist.
 396         #
 397         # Eventually a new master will be promoted and we'll resync
 398         # using the logic in ovsdb_server_notify()
 399         ${OVN_CTL} demote_ovnnb --db-nb-sync-from-addr=${INVALID_IP_ADDRESS}
 400         ${OVN_CTL} demote_ovnsb --db-sb-sync-from-addr=${INVALID_IP_ADDRESS}
 401
 402     elif [ "x${present_master}" = "x${host_name}" ]; then
 403         # Safety check, should never be called
 404         #
 405         # Never allow sync'ing from ourselves, its a great way to
 406         # erase the local DB
 407         ${OVN_CTL} demote_ovnnb --db-nb-sync-from-addr=${INVALID_IP_ADDRESS}
 408         ${OVN_CTL} demote_ovnsb --db-sb-sync-from-addr=${INVALID_IP_ADDRESS}
 409
 410     elif [ "x${present_master}" != x ]; then
 411         # There are too many masters and we're an extra one that is
 412         # being demoted. Sync to the surviving one
 413         ${OVN_CTL} demote_ovnnb --db-nb-sync-from-addr=${MASTER_IP} \
 414                                 --db-nb-sync-from-port=${NB_MASTER_PORT} \
 415                                 --db-nb-sync-from-proto=${NB_MASTER_PROTO}
 416         ${OVN_CTL} demote_ovnsb --db-sb-sync-from-addr=${MASTER_IP} \
 417                                 --db-sb-sync-from-port=${SB_MASTER_PORT} \
 418                                 --db-sb-sync-from-proto=${SB_MASTER_PROTO}
 419
 420     else
 421         # For completeness, should never be called
 422         #
 423         # Something unexpected happened, perhaps CRM_ATTR_REPL_INFO is incorrect
 424         ${OVN_CTL} demote_ovnnb --db-nb-sync-from-addr=${INVALID_IP_ADDRESS}
 425         ${OVN_CTL} demote_ovnsb --db-sb-sync-from-addr=${INVALID_IP_ADDRESS}
 426     fi
 427
 428     ovsdb_server_master_update $OCF_SUCCESS
 429     return $OCF_SUCCESS
 430 }
 431
 432 ovsdb_server_validate() {
 433     if [ ! -e ${OVN_CTL} ]; then
 434         return $OCF_ERR_INSTALLED
 435     fi
 436     return $OCF_SUCCESS
 437 }
 438
 439
 440 case $__OCF_ACTION in
 441 start)          ovsdb_server_start;;
 442 stop)           ovsdb_server_stop;;
 443 promote)        ovsdb_server_promote;;
 444 demote)         ovsdb_server_demote;;
 445 notify)         ovsdb_server_notify;;
 446 meta-data)      ovsdb_server_metadata;;
 447 validate-all)   ovsdb_server_validate;;
 448 status|monitor) ovsdb_server_monitor;;
 449 usage|help)     ovsdb_server_usage $OCF_SUCCESS;;
 450 *)              ovsdb_server_usage $OCF_ERR_UNIMPLEMENTED ;;
 451 esac
 452
 453 rc=$?
 454 exit $rc