3 # A heat-config-script which runs yum update during a stack-update.
5 # deploy_action - yum will only be run if this is UPDATE
6 # update_identifier - yum will only run for previously unused values of update_identifier
7 # command - yum sub-command to run, defaults to "update"
8 # command_arguments - yum command arguments, defaults to ""
10 echo "Started yum_update.sh on server $deploy_server_id at `date`"
11 echo -n "false" > $heat_outputs_path.update_managed_packages
13 if [[ -z "$update_identifier" ]]; then
14 echo "Not running due to unset update_identifier"
18 timestamp_dir=/var/lib/overcloud-yum-update
19 mkdir -p $timestamp_dir
21 # sanitise to remove unusual characters
22 update_identifier=${update_identifier//[^a-zA-Z0-9-_]/}
24 # seconds to wait for this node to rejoin the cluster after update
25 cluster_start_timeout=600
26 galera_sync_timeout=360
28 timestamp_file="$timestamp_dir/$update_identifier"
29 if [[ -a "$timestamp_file" ]]; then
30 echo "Not running for already-run timestamp \"$update_identifier\""
33 touch "$timestamp_file"
35 command_arguments=${command_arguments:-}
37 list_updates=$(yum list updates)
39 if [[ "$list_updates" == "" ]]; then
40 echo "No packages require updating"
44 pacemaker_status=$(systemctl is-active pacemaker)
45 pacemaker_dumpfile=$(mktemp)
47 if [[ "$pacemaker_status" == "active" ]] ; then
48 echo "Dumping Pacemaker config"
49 pcs cluster cib $pacemaker_dumpfile
51 echo "Checking for missing constraints"
53 if ! pcs constraint order show | grep "start openstack-nova-novncproxy-clone then start openstack-nova-api-clone"; then
54 pcs -f $pacemaker_dumpfile constraint order start openstack-nova-novncproxy-clone then openstack-nova-api-clone
57 if ! pcs constraint order show | grep "start rabbitmq-clone then start openstack-keystone-clone"; then
58 pcs -f $pacemaker_dumpfile constraint order start rabbitmq-clone then openstack-keystone-clone
61 if ! pcs constraint order show | grep "promote galera-master then start openstack-keystone-clone"; then
62 pcs -f $pacemaker_dumpfile constraint order promote galera-master then openstack-keystone-clone
65 if ! pcs constraint order show | grep "start haproxy-clone then start openstack-keystone-clone"; then
66 pcs -f $pacemaker_dumpfile constraint order start haproxy-clone then openstack-keystone-clone
69 if ! pcs constraint order show | grep "start memcached-clone then start openstack-keystone-clone"; then
70 pcs -f $pacemaker_dumpfile constraint order start memcached-clone then openstack-keystone-clone
73 if ! pcs constraint order show | grep "promote redis-master then start openstack-ceilometer-central-clone"; then
74 pcs -f $pacemaker_dumpfile constraint order promote redis-master then start openstack-ceilometer-central-clone require-all=false
77 if ! pcs resource defaults | grep "resource-stickiness: INFINITY"; then
78 pcs -f $pacemaker_dumpfile resource defaults resource-stickiness=INFINITY
81 echo "Setting resource start/stop timeouts"
88 neutron-metadata-agent
89 neutron-openvswitch-agent
91 openstack-ceilometer-alarm-evaluator
92 openstack-ceilometer-alarm-notifier
93 openstack-ceilometer-api
94 openstack-ceilometer-central
95 openstack-ceilometer-collector
96 openstack-ceilometer-notification
98 openstack-cinder-scheduler
99 openstack-cinder-volume
101 openstack-glance-registry
103 openstack-heat-api-cfn
104 openstack-heat-api-cloudwatch
105 openstack-heat-engine
108 openstack-nova-conductor
109 openstack-nova-consoleauth
110 openstack-nova-novncproxy
111 openstack-nova-scheduler"
112 for service in $SERVICES; do
113 pcs -f $pacemaker_dumpfile resource update $service op start timeout=100s op stop timeout=100s
115 # mongod start timeout is higher, setting only stop timeout
116 pcs resource update mongod op stop timeout=100s
118 echo "Applying new Pacemaker config"
119 pcs cluster cib-push $pacemaker_dumpfile
121 echo "Pacemaker running, stopping cluster node and doing full package update"
122 node_count=$(pcs status xml | grep -o "<nodes_configured.*/>" | grep -o 'number="[0-9]*"' | grep -o "[0-9]*")
123 if [[ "$node_count" == "1" ]] ; then
124 echo "Active node count is 1, stopping node with --force"
125 pcs cluster stop --force
130 echo "Excluding upgrading packages that are handled by config management tooling"
131 command_arguments="$command_arguments --skip-broken"
132 for exclude in $(cat /var/lib/tripleo/installed-packages/* | sort -u); do
133 command_arguments="$command_arguments --exclude $exclude"
137 command=${command:-update}
138 full_command="yum -y $command $command_arguments"
139 echo "Running: $full_command"
141 result=$($full_command)
144 echo "yum return code: $return_code"
146 if [[ "$pacemaker_status" == "active" ]] ; then
147 echo "Starting cluster node"
150 hostname=$(hostname -s)
152 while [[ "$(pcs status | grep "^Online" | grep -F -o $hostname)" == "" ]]; do
155 if (( tnow-tstart > cluster_start_timeout )) ; then
156 echo "ERROR $hostname failed to join cluster in $cluster_start_timeout seconds"
163 while ! clustercheck; do
166 if (( tnow-tstart > galera_sync_timeout )) ; then
167 echo "ERROR galera sync timed out"
175 echo -n "true" > $heat_outputs_path.update_managed_packages
178 echo "Finished yum_update.sh on server $deploy_server_id at `date`"