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=360
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=$(systemctl is-active pacemaker)
46 if [[ "$pacemaker_status" == "active" ]] ; then
47 echo "Checking for and adding missing constraints"
49 if ! pcs constraint order show | grep "start openstack-nova-novncproxy-clone then start openstack-nova-api-clone"; then
50 pcs constraint order start openstack-nova-novncproxy-clone then openstack-nova-api-clone
53 if ! pcs constraint order show | grep "start rabbitmq-clone then start openstack-keystone-clone"; then
54 pcs constraint order start rabbitmq-clone then openstack-keystone-clone
57 if ! pcs constraint order show | grep "promote galera-master then start openstack-keystone-clone"; then
58 pcs constraint order promote galera-master then openstack-keystone-clone
61 if ! pcs constraint order show | grep "start haproxy-clone then start openstack-keystone-clone"; then
62 pcs constraint order start haproxy-clone then openstack-keystone-clone
65 if ! pcs constraint order show | grep "start memcached-clone then start openstack-keystone-clone"; then
66 pcs constraint order start memcached-clone then openstack-keystone-clone
69 if ! pcs constraint order show | grep "promote redis-master then start openstack-ceilometer-central-clone"; then
70 pcs constraint order promote redis-master then start openstack-ceilometer-central-clone require-all=false
73 if ! pcs resource defaults | grep "resource-stickiness: INFINITY"; then
74 pcs resource defaults resource-stickiness=INFINITY
77 echo "Setting resource start/stop timeouts"
79 # timeouts for non-openstack services and special cases
80 pcs resource update haproxy op start timeout=100s
81 pcs resource update haproxy op stop timeout=100s
82 # mongod start timeout is also higher, setting only stop timeout
83 pcs resource update mongod op stop timeout=100s
84 # rabbit start timeout is already 100s
85 pcs resource update rabbitmq op stop timeout=100s
86 pcs resource update memcached op start timeout=100s
87 pcs resource update memcached op stop timeout=100s
88 pcs resource update httpd op start timeout=100s
89 pcs resource update httpd op stop timeout=100s
90 # neutron-netns-cleanup stop timeout is 300s, setting only start timeout
91 pcs resource update neutron-netns-cleanup op start timeout=100s
92 # neutron-ovs-cleanup stop timeout is 300s, setting only start timeout
93 pcs resource update neutron-ovs-cleanup op start timeout=100s
95 # timeouts for openstack services
96 pcs resource update neutron-dhcp-agent op start timeout=100s
97 pcs resource update neutron-dhcp-agent op stop timeout=100s
98 pcs resource update neutron-l3-agent op start timeout=100s
99 pcs resource update neutron-l3-agent op stop timeout=100s
100 pcs resource update neutron-metadata-agent op start timeout=100s
101 pcs resource update neutron-metadata-agent op stop timeout=100s
102 pcs resource update neutron-openvswitch-agent op start timeout=100s
103 pcs resource update neutron-openvswitch-agent op stop timeout=100s
104 pcs resource update neutron-server op start timeout=100s
105 pcs resource update neutron-server op stop timeout=100s
106 pcs resource update openstack-ceilometer-alarm-evaluator op start timeout=100s
107 pcs resource update openstack-ceilometer-alarm-evaluator op stop timeout=100s
108 pcs resource update openstack-ceilometer-alarm-notifier op start timeout=100s
109 pcs resource update openstack-ceilometer-alarm-notifier op stop timeout=100s
110 pcs resource update openstack-ceilometer-api op start timeout=100s
111 pcs resource update openstack-ceilometer-api op stop timeout=100s
112 pcs resource update openstack-ceilometer-central op start timeout=100s
113 pcs resource update openstack-ceilometer-central op stop timeout=100s
114 pcs resource update openstack-ceilometer-collector op start timeout=100s
115 pcs resource update openstack-ceilometer-collector op stop timeout=100s
116 pcs resource update openstack-ceilometer-notification op start timeout=100s
117 pcs resource update openstack-ceilometer-notification op stop timeout=100s
118 pcs resource update openstack-cinder-api op start timeout=100s
119 pcs resource update openstack-cinder-api op stop timeout=100s
120 pcs resource update openstack-cinder-scheduler op start timeout=100s
121 pcs resource update openstack-cinder-scheduler op stop timeout=100s
122 pcs resource update openstack-cinder-volume op start timeout=100s
123 pcs resource update openstack-cinder-volume op stop timeout=100s
124 pcs resource update openstack-glance-api op start timeout=100s
125 pcs resource update openstack-glance-api op stop timeout=100s
126 pcs resource update openstack-glance-registry op start timeout=100s
127 pcs resource update openstack-glance-registry op stop timeout=100s
128 pcs resource update openstack-heat-api op start timeout=100s
129 pcs resource update openstack-heat-api op stop timeout=100s
130 pcs resource update openstack-heat-api-cfn op start timeout=100s
131 pcs resource update openstack-heat-api-cfn op stop timeout=100s
132 pcs resource update openstack-heat-api-cloudwatch op start timeout=100s
133 pcs resource update openstack-heat-api-cloudwatch op stop timeout=100s
134 pcs resource update openstack-heat-engine op start timeout=100s
135 pcs resource update openstack-heat-engine op stop timeout=100s
136 pcs resource update openstack-keystone op start timeout=100s
137 pcs resource update openstack-keystone op stop timeout=100s
138 pcs resource update openstack-nova-api op start timeout=100s
139 pcs resource update openstack-nova-api op stop timeout=100s
140 pcs resource update openstack-nova-conductor op start timeout=100s
141 pcs resource update openstack-nova-conductor op stop timeout=100s
142 pcs resource update openstack-nova-consoleauth op start timeout=100s
143 pcs resource update openstack-nova-consoleauth op stop timeout=100s
144 pcs resource update openstack-nova-novncproxy op start timeout=100s
145 pcs resource update openstack-nova-novncproxy op stop timeout=100s
146 pcs resource update openstack-nova-scheduler op start timeout=100s
147 pcs resource update openstack-nova-scheduler op stop timeout=100s
149 echo "Pacemaker running, stopping cluster node and doing full package update"
150 node_count=$(pcs status xml | grep -o "<nodes_configured.*/>" | grep -o 'number="[0-9]*"' | grep -o "[0-9]*")
151 if [[ "$node_count" == "1" ]] ; then
152 echo "Active node count is 1, stopping node with --force"
153 pcs cluster stop --force
158 echo "Excluding upgrading packages that are handled by config management tooling"
159 command_arguments="$command_arguments --skip-broken"
160 for exclude in $(cat /var/lib/tripleo/installed-packages/* | sort -u); do
161 command_arguments="$command_arguments --exclude $exclude"
165 command=${command:-update}
166 full_command="yum -y $command $command_arguments"
167 echo "Running: $full_command"
169 result=$($full_command)
172 echo "yum return code: $return_code"
174 if [[ "$pacemaker_status" == "active" ]] ; then
175 echo "Starting cluster node"
178 hostname=$(hostname -s)
180 while [[ "$(pcs status | grep "^Online" | grep -F -o $hostname)" == "" ]]; do
183 if (( tnow-tstart > cluster_start_timeout )) ; then
184 echo "ERROR $hostname failed to join cluster in $cluster_start_timeout seconds"
191 while ! clustercheck; do
194 if (( tnow-tstart > galera_sync_timeout )) ; then
195 echo "ERROR galera sync timed out"
203 echo -n "true" > $heat_outputs_path.update_managed_packages
206 echo "Finished yum_update.sh on server $deploy_server_id at `date`"