Fix race during major-upgrade-pacemaker step

Currently when we call the major-upgrade step we do the following:
"""
...
if [[ -n $(is_bootstrap_node) ]]; then
    check_clean_cluster
fi
...
if [[ -n $(is_bootstrap_node) ]]; then
    migrate_full_to_ng_ha
fi
...
for service in $(services_to_migrate); do
    manage_systemd_service stop "${service%%-clone}"
    ...
done
"""

The problem with the above code is that it is open to the following race
condition:
1. Code gets run first on a non-bootstrap controller node so we start
stopping a bunch of services
2. Pacemaker notices will notice that services are down and will mark
the service as stopped
3. Code gets run on the bootstrap node (controller-0) and the
check_clean_cluster function will fail and exit
4. Eventually also the script on the non-bootstrap controller node will
timeout and exit because the cluster never shut down (it never actually
started the shutdown because we failed at 3)

Let's make sure we first only call the HA NG migration step as a
separate heat step. Only afterwards we start shutting down the systemd
services on all nodes.

We also need to move the STONITH_STATE variable into a file because it
is being used across two different scripts (1 and 2) and we need to
store that state.

Co-Authored-By: Athlan-Guyot Sofer <sathlang@redhat.com>

Closes-Bug: #1640407
Change-Id: Ifb9b9e633fcc77604cca2590071656f4b2275c60

1 files changed, 5 insertions, 177 deletions

diff --git a/extraconfig/tasks/major_upgrade_controller_pacemaker_1.sh b/extraconfig/tasks/major_upgrade_controller_pacemaker_1.sh
index fbdbc30b..739c01dd 100755
--- a/extraconfig/tasks/major_upgrade_controller_pacemaker_1.sh
+++ b/extraconfig/tasks/major_upgrade_controller_pacemaker_1.sh
@@ -2,8 +2,6 @@
 
 set -eu
 
-cluster_sync_timeout=1800
-
 check_cluster
 check_pcsd
 if [[ -n $(is_bootstrap_node) ]]; then
@@ -19,6 +17,11 @@ check_disk_for_mysql_dump
 # at the end of this script
 if [[ -n $(is_bootstrap_node) ]]; then
     STONITH_STATE=$(pcs property show stonith-enabled | grep "stonith-enabled" | awk '{ print $2 }')
+    # We create this empty file if stonith was set to true so we can reenable stonith in step2
+    rm -f /var/tmp/stonith-true
+    if [ $STONITH_STATE == "true" ]; then
+        touch /var/tmp/stonith-true
+    fi
     pcs property set stonith-enabled=false
 fi
 
@@ -31,178 +34,3 @@ if [[ -n $(is_bootstrap_node) ]]; then
     rabbitmq_mitaka_newton_upgrade
 fi
 
-# After migrating the cluster to HA-NG the services not under pacemaker's control
-# are still up and running. We need to stop them explicitely otherwise during the yum
-# upgrade the rpm %post sections will try to do a systemctl try-restart <service>, which
-# is going to take a long time because rabbit is down. By having the service stopped
-# systemctl try-restart is a noop
-
-for service in $(services_to_migrate); do
-    manage_systemd_service stop "${service%%-clone}"
-    # So the reason for not reusing check_resource_systemd is that
-    # I have observed systemctl is-active returning unknown with at least
-    # one service that was stopped (See LP 1627254)
-    timeout=600
-    tstart=$(date +%s)
-    tend=$(( $tstart + $timeout ))
-    check_interval=3
-    while (( $(date +%s) < $tend )); do
-      if [[ "$(systemctl is-active ${service%%-clone})" = "active" ]]; then
-        echo "$service still active, sleeping $check_interval seconds."
-        sleep $check_interval
-      else
-        # we do not care if it is inactive, unknown or failed as long as it is
-        # not running
-        break
-      fi
-
-    done
-done
-
-# In case the mysql package is updated, the database on disk must be
-# upgraded as well. This typically needs to happen during major
-# version upgrades (e.g. 5.5 -> 5.6, 5.5 -> 10.1...)
-#
-# Because in-place upgrades are not supported across 2+ major versions
-# (e.g. 5.5 -> 10.1), we rely on logical upgrades via dump/restore cycle
-# https://bugzilla.redhat.com/show_bug.cgi?id=1341968
-#
-# The default is to determine automatically if upgrade is needed based
-# on mysql package versionning, but this can be overriden manually
-# to support specific upgrade scenario
-
-if [[ -n $(is_bootstrap_node) ]]; then
-    if [ $DO_MYSQL_UPGRADE -eq 1 ]; then
-        mysqldump $backup_flags > "$MYSQL_BACKUP_DIR/openstack_database.sql"
-        cp -rdp /etc/my.cnf* "$MYSQL_BACKUP_DIR"
-    fi
-
-    pcs resource disable redis
-    check_resource redis stopped 600
-    pcs resource disable rabbitmq
-    check_resource rabbitmq stopped 600
-    pcs resource disable galera
-    check_resource galera stopped 600
-    pcs resource disable openstack-cinder-volume
-    check_resource openstack-cinder-volume stopped 600
-    # Disable all VIPs before stopping the cluster, so that pcs doesn't use one as a source address:
-    #   https://bugzilla.redhat.com/show_bug.cgi?id=1330688
-    for vip in $(pcs resource show | grep ocf::heartbeat:IPaddr2 | grep Started | awk '{ print $1 }'); do
-      pcs resource disable $vip
-      check_resource $vip stopped 60
-    done
-    pcs cluster stop --all
-fi
-
-
-# Swift isn't controlled by pacemaker
-systemctl_swift stop
-
-tstart=$(date +%s)
-while systemctl is-active pacemaker; do
-    sleep 5
-    tnow=$(date +%s)
-    if (( tnow-tstart > cluster_sync_timeout )) ; then
-        echo_error "ERROR: cluster shutdown timed out"
-        exit 1
-    fi
-done
-
-# The reason we do an sql dump *and* we move the old dir out of
-# the way is because it gives us an extra level of safety in case
-# something goes wrong during the upgrade. Once the restore is
-# successful we go ahead and remove it. If the directory exists
-# we bail out as it means the upgrade process had issues in the last
-# run.
-if [ $DO_MYSQL_UPGRADE -eq 1 ]; then
-    if [ -d $MYSQL_TEMP_UPGRADE_BACKUP_DIR ]; then
-        echo_error "ERROR: mysql backup dir already exist"
-        exit 1
-    fi
-    mv /var/lib/mysql $MYSQL_TEMP_UPGRADE_BACKUP_DIR
-fi
-
-# Special-case OVS for https://bugs.launchpad.net/tripleo/+bug/1635205
-if [[ -n $(rpm -q --scripts openvswitch | awk '/postuninstall/,/*/' | grep "systemctl.*try-restart") ]]; then
-    echo "Manual upgrade of openvswitch - restart in postun detected"
-    mkdir OVS_UPGRADE || true
-    pushd OVS_UPGRADE
-    echo "Attempting to downloading latest openvswitch with yumdownloader"
-    yumdownloader --resolve openvswitch
-    echo "Updating openvswitch with nopostun option"
-    rpm -U --replacepkgs --nopostun ./*.rpm
-    popd
-else
-    echo "Skipping manual upgrade of openvswitch - no restart in postun detected"
-fi
-
-yum -y install python-zaqarclient  # needed for os-collect-config
-yum -y -q update
-
-# We need to ensure at least those two configuration settings, otherwise
-# mariadb 10.1+ won't activate galera replication.
-# wsrep_cluster_address must only be set though, its value does not
-# matter because it's overriden by the galera resource agent.
-cat >> /etc/my.cnf.d/galera.cnf <<EOF
-[mysqld]
-wsrep_on = ON
-wsrep_cluster_address = gcomm://localhost
-EOF
-
-if [ $DO_MYSQL_UPGRADE -eq 1 ]; then
-    # Scripts run via heat have no HOME variable set and this confuses
-    # mysqladmin
-    export HOME=/root
-
-    mkdir /var/lib/mysql || /bin/true
-    chown mysql:mysql /var/lib/mysql
-    chmod 0755 /var/lib/mysql
-    restorecon -R /var/lib/mysql/
-    mysql_install_db --datadir=/var/lib/mysql --user=mysql
-    chown -R mysql:mysql /var/lib/mysql/
-
-    if [ "$(hiera -c /etc/puppet/hiera.yaml bootstrap_nodeid)" = "$(facter hostname)" ]; then
-        mysqld_safe --wsrep-new-cluster &
-        # We have a populated /root/.my.cnf with root/password here so
-        # we need to temporarily rename it because the newly created
-        # db is empty and no root password is set
-        mv /root/.my.cnf /root/.my.cnf.temporary
-        timeout 60 sh -c 'while ! mysql -e "" &> /dev/null; do sleep 1; done'
-        mysql -u root < "$MYSQL_BACKUP_DIR/openstack_database.sql"
-        mv /root/.my.cnf.temporary /root/.my.cnf
-        mysqladmin -u root shutdown
-        # The import was successful so we may remove the folder
-        rm -r "$MYSQL_BACKUP_DIR"
-    fi
-fi
-
-# If we reached here without error we can safely blow away the origin
-# mysql dir from every controller
-
-# TODO: What if the upgrade fails on the bootstrap node, but not on
-# this controller.  Data may be lost.
-if [ $DO_MYSQL_UPGRADE -eq 1 ]; then
-    rm -r $MYSQL_TEMP_UPGRADE_BACKUP_DIR
-fi
-
-# Let's reset the stonith back to true if it was true, before starting the cluster
-if [[ -n $(is_bootstrap_node) ]]; then
-    if [ $STONITH_STATE == "true" ]; then
-        pcs -f /var/lib/pacemaker/cib/cib.xml property set stonith-enabled=true
-    fi
-fi
-
-# Pin messages sent to compute nodes to kilo, these will be upgraded later
-crudini  --set /etc/nova/nova.conf upgrade_levels compute "$upgrade_level_nova_compute"
-# https://bugzilla.redhat.com/show_bug.cgi?id=1284047
-# Change-Id: Ib3f6c12ff5471e1f017f28b16b1e6496a4a4b435
-crudini  --set /etc/ceilometer/ceilometer.conf DEFAULT rpc_backend rabbit
-# https://bugzilla.redhat.com/show_bug.cgi?id=1284058
-# Ifd1861e3df46fad0e44ff9b5cbd58711bbc87c97 Swift Ceilometer middleware no longer exists
-crudini --set /etc/swift/proxy-server.conf pipeline:main pipeline "catch_errors healthcheck cache ratelimit tempurl formpost authtoken keystone staticweb proxy-logging proxy-server"
-# LP: 1615035, required only for M/N upgrade.
-crudini --set /etc/nova/nova.conf DEFAULT scheduler_host_manager host_manager
-# LP: 1627450, required only for M/N upgrade
-crudini --set /etc/nova/nova.conf DEFAULT scheduler_driver filter_scheduler
-
-crudini --set /etc/sahara/sahara.conf DEFAULT plugins ambari,cdh,mapr,vanilla,spark,storm