update_identifier=${update_identifier//[^a-zA-Z0-9-_]/}
# seconds to wait for this node to rejoin the cluster after update
-cluster_start_timeout=360
+cluster_start_timeout=600
+galera_sync_timeout=1800
+cluster_settle_timeout=1800
timestamp_file="$timestamp_dir/$update_identifier"
if [[ -a "$timestamp_file" ]]; then
pacemaker_status=$(systemctl is-active pacemaker)
if [[ "$pacemaker_status" == "active" ]] ; then
- echo "Checking for and adding missing constraints"
-
- if ! pcs constraint order show | grep "start openstack-nova-novncproxy-clone then start openstack-nova-api-clone"; then
- pcs constraint order start openstack-nova-novncproxy-clone then openstack-nova-api-clone
- fi
-
- if ! pcs constraint order show | grep "start rabbitmq-clone then start openstack-keystone-clone"; then
- pcs constraint order start rabbitmq-clone then openstack-keystone-clone
- fi
-
- if ! pcs constraint order show | grep "promote galera-master then start openstack-keystone-clone"; then
- pcs constraint order promote galera-master then openstack-keystone-clone
- fi
-
- if ! pcs constraint order show | grep "start haproxy-clone then start openstack-keystone-clone"; then
- pcs constraint order start haproxy-clone then openstack-keystone-clone
- fi
-
- if ! pcs constraint order show | grep "start memcached-clone then start openstack-keystone-clone"; then
- pcs constraint order start memcached-clone then openstack-keystone-clone
- fi
-
- if ! pcs constraint order show | grep "promote redis-master then start openstack-ceilometer-central-clone"; then
- pcs constraint order promote redis-master then start openstack-ceilometer-central-clone require-all=false
- fi
-
- if ! pcs resource defaults | grep "resource-stickiness: INFINITY"; then
- pcs resource defaults resource-stickiness=INFINITY
- fi
-
- echo "Setting resource start/stop timeouts"
-
- # timeouts for non-openstack services and special cases
- pcs resource update haproxy op start timeout=100s
- pcs resource update haproxy op stop timeout=100s
- # mongod start timeout is also higher, setting only stop timeout
- pcs resource update mongod op stop timeout=100s
- # rabbit start timeout is already 100s
- pcs resource update rabbitmq op stop timeout=100s
- pcs resource update memcached op start timeout=100s
- pcs resource update memcached op stop timeout=100s
- pcs resource update httpd op start timeout=100s
- pcs resource update httpd op stop timeout=100s
- # neutron-netns-cleanup stop timeout is 300s, setting only start timeout
- pcs resource update neutron-netns-cleanup op start timeout=100s
- # neutron-ovs-cleanup stop timeout is 300s, setting only start timeout
- pcs resource update neutron-ovs-cleanup op start timeout=100s
-
- # timeouts for openstack services
- pcs resource update neutron-dhcp-agent op start timeout=100s
- pcs resource update neutron-dhcp-agent op stop timeout=100s
- pcs resource update neutron-l3-agent op start timeout=100s
- pcs resource update neutron-l3-agent op stop timeout=100s
- pcs resource update neutron-metadata-agent op start timeout=100s
- pcs resource update neutron-metadata-agent op stop timeout=100s
- pcs resource update neutron-openvswitch-agent op start timeout=100s
- pcs resource update neutron-openvswitch-agent op stop timeout=100s
- pcs resource update neutron-server op start timeout=100s
- pcs resource update neutron-server op stop timeout=100s
- pcs resource update openstack-ceilometer-alarm-evaluator op start timeout=100s
- pcs resource update openstack-ceilometer-alarm-evaluator op stop timeout=100s
- pcs resource update openstack-ceilometer-alarm-notifier op start timeout=100s
- pcs resource update openstack-ceilometer-alarm-notifier op stop timeout=100s
- pcs resource update openstack-ceilometer-api op start timeout=100s
- pcs resource update openstack-ceilometer-api op stop timeout=100s
- pcs resource update openstack-ceilometer-central op start timeout=100s
- pcs resource update openstack-ceilometer-central op stop timeout=100s
- pcs resource update openstack-ceilometer-collector op start timeout=100s
- pcs resource update openstack-ceilometer-collector op stop timeout=100s
- pcs resource update openstack-ceilometer-notification op start timeout=100s
- pcs resource update openstack-ceilometer-notification op stop timeout=100s
- pcs resource update openstack-cinder-api op start timeout=100s
- pcs resource update openstack-cinder-api op stop timeout=100s
- pcs resource update openstack-cinder-scheduler op start timeout=100s
- pcs resource update openstack-cinder-scheduler op stop timeout=100s
- pcs resource update openstack-cinder-volume op start timeout=100s
- pcs resource update openstack-cinder-volume op stop timeout=100s
- pcs resource update openstack-glance-api op start timeout=100s
- pcs resource update openstack-glance-api op stop timeout=100s
- pcs resource update openstack-glance-registry op start timeout=100s
- pcs resource update openstack-glance-registry op stop timeout=100s
- pcs resource update openstack-heat-api op start timeout=100s
- pcs resource update openstack-heat-api op stop timeout=100s
- pcs resource update openstack-heat-api-cfn op start timeout=100s
- pcs resource update openstack-heat-api-cfn op stop timeout=100s
- pcs resource update openstack-heat-api-cloudwatch op start timeout=100s
- pcs resource update openstack-heat-api-cloudwatch op stop timeout=100s
- pcs resource update openstack-heat-engine op start timeout=100s
- pcs resource update openstack-heat-engine op stop timeout=100s
- pcs resource update openstack-keystone op start timeout=100s
- pcs resource update openstack-keystone op stop timeout=100s
- pcs resource update openstack-nova-api op start timeout=100s
- pcs resource update openstack-nova-api op stop timeout=100s
- pcs resource update openstack-nova-conductor op start timeout=100s
- pcs resource update openstack-nova-conductor op stop timeout=100s
- pcs resource update openstack-nova-consoleauth op start timeout=100s
- pcs resource update openstack-nova-consoleauth op stop timeout=100s
- pcs resource update openstack-nova-novncproxy op start timeout=100s
- pcs resource update openstack-nova-novncproxy op stop timeout=100s
- pcs resource update openstack-nova-scheduler op start timeout=100s
- pcs resource update openstack-nova-scheduler op stop timeout=100s
-
echo "Pacemaker running, stopping cluster node and doing full package update"
node_count=$(pcs status xml | grep -o "<nodes_configured.*/>" | grep -o 'number="[0-9]*"' | grep -o "[0-9]*")
if [[ "$node_count" == "1" ]] ; then
pcs cluster stop
fi
else
- echo "Excluding upgrading packages that are handled by config management tooling"
- command_arguments="$command_arguments --skip-broken"
- for exclude in $(cat /var/lib/tripleo/installed-packages/* | sort -u); do
- command_arguments="$command_arguments --exclude $exclude"
- done
+ echo "Upgrading openstack-puppet-modules"
+ yum -q -y update openstack-puppet-modules
+ echo "Upgrading other packages is handled by config management tooling"
+ echo -n "true" > $heat_outputs_path.update_managed_packages
+ exit 0
+fi
+
+
+# Special-case OVS for https://bugs.launchpad.net/tripleo/+bug/1635205
+if [[ -n $(rpm -q --scripts openvswitch | awk '/postuninstall/,/*/' | grep "systemctl.*try-restart") ]]; then
+ echo "Manual upgrade of openvswitch - restart in postun detected"
+ mkdir OVS_UPGRADE || true
+ pushd OVS_UPGRADE
+ echo "Attempting to downloading latest openvswitch with yumdownloader"
+ yumdownloader --resolve openvswitch
+ echo "Updating openvswitch with nopostun"
+ rpm -U --nopostun ./*.rpm
+ popd
+else
+ echo "Skipping manual upgrade of openvswitch - no restart in postun detected"
fi
command=${command:-update}
-full_command="yum -y $command $command_arguments"
+full_command="yum -q -y $command $command_arguments"
echo "Running: $full_command"
result=$($full_command)
exit 1
fi
done
- pcs status
-else
- echo -n "true" > $heat_outputs_path.update_managed_packages
+ tstart=$(date +%s)
+ while ! clustercheck; do
+ sleep 5
+ tnow=$(date +%s)
+ if (( tnow-tstart > galera_sync_timeout )) ; then
+ echo "ERROR galera sync timed out"
+ exit 1
+ fi
+ done
+
+ echo "Waiting for pacemaker cluster to settle"
+ if ! timeout -k 10 $cluster_settle_timeout crm_resource --wait; then
+ echo "ERROR timed out while waiting for the cluster to settle"
+ exit 1
+ fi
+
+ pcs status
fi
echo "Finished yum_update.sh on server $deploy_server_id at `date`"
Code Review / apex-tripleo-heat-templates.git / blobdiff