diff options
| -rw-r--r-- | ci/environments/multinode_major_upgrade.yaml | 46 | ||||
| -rw-r--r-- | puppet/major_upgrade_steps.j2.yaml | 80 | ||||
| -rw-r--r-- | puppet/services/README.rst | 22 | ||||
| -rw-r--r-- | puppet/services/barbican-api.yaml | 4 | ||||
| -rw-r--r-- | puppet/services/ceph-mon.yaml | 21 | ||||
| -rw-r--r-- | puppet/services/ceph-osd.yaml | 41 | ||||
| -rw-r--r-- | puppet/services/gnocchi-metricd.yaml | 2 | ||||
| -rw-r--r-- | puppet/services/ironic-api.yaml | 4 | ||||
| -rw-r--r-- | puppet/services/ironic-conductor.yaml | 7 | ||||
| -rw-r--r-- | puppet/services/neutron-dhcp.yaml | 5 | ||||
| -rw-r--r-- | puppet/services/services.yaml | 5 | ||||
| -rw-r--r-- | roles_data.yaml | 5 |
12 files changed, 234 insertions, 8 deletions
diff --git a/ci/environments/multinode_major_upgrade.yaml b/ci/environments/multinode_major_upgrade.yaml new file mode 100644 index 00000000..7de1594f --- /dev/null +++ b/ci/environments/multinode_major_upgrade.yaml @@ -0,0 +1,46 @@ +resource_registry: + OS::TripleO::Controller::Net::SoftwareConfig: ../heat-templates/net-config-multinode.yaml + OS::TripleO::Compute::Net::SoftwareConfig: ../heat-templates/net-config-multinode.yaml + OS::TripleO::Services::Core: multinode-core.yaml + +parameter_defaults: + ControllerServices: + - OS::TripleO::Services::CACerts + - OS::TripleO::Services::Core + - OS::TripleO::Services::Kernel + - OS::TripleO::Services::Keystone + - OS::TripleO::Services::GlanceApi + - OS::TripleO::Services::GlanceRegistry + - OS::TripleO::Services::NeutronDhcpAgent + - OS::TripleO::Services::NeutronL3Agent + - OS::TripleO::Services::NeutronMetadataAgent + - OS::TripleO::Services::NeutronServer + - OS::TripleO::Services::NeutronCorePlugin + - OS::TripleO::Services::NeutronOvsAgent + - OS::TripleO::Services::CinderApi + - OS::TripleO::Services::CinderScheduler + - OS::TripleO::Services::CinderVolume + - OS::TripleO::Services::HeatApi + - OS::TripleO::Services::HeatApiCfn + - OS::TripleO::Services::HeatApiCloudwatch + - OS::TripleO::Services::HeatEngine + - OS::TripleO::Services::SwiftProxy + - OS::TripleO::Services::SwiftStorage + - OS::TripleO::Services::SwiftRingBuilder + - OS::TripleO::Services::SaharaApi + - OS::TripleO::Services::SaharaEngine + - OS::TripleO::Services::MySQL + - OS::TripleO::Services::RabbitMQ + - OS::TripleO::Services::HAproxy + - OS::TripleO::Services::Keepalived + - OS::TripleO::Services::Memcached + - OS::TripleO::Services::Ntp + - OS::TripleO::Services::Timezone + - OS::TripleO::Services::TripleoPackages + - OS::TripleO::Services::TripleoFirewall + ControllerExtraConfig: + nova::compute::libvirt::services::libvirt_virt_type: qemu + nova::compute::libvirt::libvirt_virt_type: qemu + # Required for Centos 7.3 and Qemu 2.6.0 + nova::compute::libvirt::libvirt_cpu_mode: 'none' + SwiftCeilometerPipelineEnabled: False diff --git a/puppet/major_upgrade_steps.j2.yaml b/puppet/major_upgrade_steps.j2.yaml index 433b03a0..eae85991 100644 --- a/puppet/major_upgrade_steps.j2.yaml +++ b/puppet/major_upgrade_steps.j2.yaml @@ -1,3 +1,4 @@ +{% set upgrade_steps_max = 8 -%} heat_template_version: ocata description: 'Upgrade steps for all roles' @@ -15,26 +16,94 @@ parameters: Setting to a previously unused value during stack-update will trigger the Upgrade resources to re-run on all roles. +conditions: + # Conditions to disable any steps where the task list is empty +{% for step in range(0, upgrade_steps_max) %} + {% for role in roles %} + UpgradeBatchConfig_Step{{step}}Enabled: + not: + equals: + - {get_param: [role_data, {{role.name}}, upgrade_batch_tasks]} + - [] + UpgradeConfig_Step{{step}}Enabled: + not: + equals: + - {get_param: [role_data, {{role.name}}, upgrade_tasks]} + - [] + {% endfor %} +{% endfor %} + resources: +# Upgrade Steps for all roles, batched updates +# FIXME(shardy): would be nice to make the number of steps configurable +{% for step in range(0, upgrade_steps_max) %} + {% for role in roles %} + # Step {{step}} resources + {{role.name}}UpgradeBatchConfig_Step{{step}}: + type: OS::TripleO::UpgradeConfig + condition: UpgradeBatchConfig_Step{{step}}Enabled + # The UpgradeConfig resources could actually be created without + # serialization, but the event output is easier to follow if we + # do, and there should be minimal performance hit (creating the + # config is cheap compared to the time to apply the deployment). + {% if step > 0 %} + depends_on: + {% for dep in roles %} + - {{dep.name}}UpgradeBatch_Step{{step -1}} + {% endfor %} + {% endif %} + properties: + UpgradeStepConfig: {get_param: [role_data, {{role.name}}, upgrade_batch_tasks]} + step: {{step}} + + {{role.name}}UpgradeBatch_Step{{step}}: + type: OS::Heat::StructuredDeploymentGroup + condition: UpgradeBatchConfig_Step{{step}}Enabled + {% if step > 0 %} + depends_on: + {% for dep in roles %} + - {{dep.name}}UpgradeBatch_Step{{step -1}} + {% endfor %} + {% endif %} + update_policy: + batch_create: + max_batch_size: {{role.upgrade_batch_size|default(1)}} + rolling_update: + max_batch_size: {{role.upgrade_batch_size|default(1)}} + properties: + name: {{role.name}}UpgradeBatch_Step{{step}} + servers: {get_param: [servers, {{role.name}}]} + config: {get_resource: {{role.name}}UpgradeBatchConfig_Step{{step}}} + input_values: + role: {{role.name}} + update_identifier: {get_param: UpdateIdentifier} + {% endfor %} +{% endfor %} + # Upgrade Steps for all roles # FIXME(shardy): would be nice to make the number of steps configurable -{% for step in range(0, 8) %} +{% for step in range(0, upgrade_steps_max) %} {% for role in roles %} # Step {{step}} resources {{role.name}}UpgradeConfig_Step{{step}}: type: OS::TripleO::UpgradeConfig + condition: UpgradeConfig_Step{{step}}Enabled # The UpgradeConfig resources could actually be created without # serialization, but the event output is easier to follow if we # do, and there should be minimal performance hit (creating the # config is cheap compared to the time to apply the deployment). - {% if step > 0 %} depends_on: + {% if step > 0 %} {% for dep in roles %} {% if not dep.disable_upgrade_deployment|default(false) %} - {{dep.name}}Upgrade_Step{{step -1}} {% endif %} {% endfor %} + {% else %} + {% for dep in roles %} + - {{dep.name}}UpgradeBatch_Step{{upgrade_steps_max -1}} + {% endfor %} {% endif %} properties: UpgradeStepConfig: {get_param: [role_data, {{role.name}}, upgrade_tasks]} @@ -42,13 +111,18 @@ resources: {% if not role.disable_upgrade_deployment|default(false) %} {{role.name}}Upgrade_Step{{step}}: type: OS::Heat::StructuredDeploymentGroup - {% if step > 0 %} + condition: UpgradeConfig_Step{{step}}Enabled depends_on: + {% if step > 0 %} {% for dep in roles %} {% if not dep.disable_upgrade_deployment|default(false) %} - {{dep.name}}Upgrade_Step{{step -1}} {% endif %} {% endfor %} + {% else %} + {% for dep in roles %} + - {{dep.name}}UpgradeBatch_Step{{upgrade_steps_max -1}} + {% endfor %} {% endif %} properties: name: {{role.name}}Upgrade_Step{{step}} diff --git a/puppet/services/README.rst b/puppet/services/README.rst index 6e4e9c1d..34cb350b 100644 --- a/puppet/services/README.rst +++ b/puppet/services/README.rst @@ -49,6 +49,28 @@ are re-asserted when applying latter ones. 5) Service activation (Pacemaker) +Batch Upgrade Steps +------------------- + +Each service template may optionally define a `upgrade_batch_tasks` key, which +is a list of ansible tasks to be performed during the upgrade process. + +Similar to the step_config, we allow a series of steps for the per-service +upgrade sequence, defined as ansible tasks with a tag e.g "step1" for the first +step, "step2" for the second, etc. Note that each step is performed in batches, +then we move on to the next step which is also performed in batches (we don't +perform all steps on one node, then move on to the next one which means you +can sequence rolling upgrades of dependent services via the step value). + +The tasks performed at each step is service specific, but note that all batch +upgrade steps are performed before the `upgrade_tasks` described below. This +means that all services that support rolling upgrades can be upgraded without +downtime during `upgrade_batch_tasks`, then any remaining services are stopped +and upgraded during `upgrade_tasks` + +The default batch size is 1, but this can be overridden for each role via the +`upgrade_batch_size` option in roles_data.yaml + Upgrade Steps ------------- diff --git a/puppet/services/barbican-api.yaml b/puppet/services/barbican-api.yaml index 1f220e6b..4e420750 100644 --- a/puppet/services/barbican-api.yaml +++ b/puppet/services/barbican-api.yaml @@ -136,13 +136,13 @@ outputs: nova::compute::barbican_endpoint: get_param: [EndpointMap, BarbicanInternal, uri] nova::compute::barbican_auth_endpoint: - get_param: [EndpointMap, KeystoneV3Internal, uri] + get_param: [EndpointMap, KeystoneV3Internal, uri_no_suffix] cinder_api: cinder::api::keymgr_api_class: > castellan.key_manager.barbican_key_manager.BarbicanKeyManager cinder::api::keymgr_encryption_api_url: get_param: [EndpointMap, BarbicanInternal, uri] cinder::api::keymgr_encryption_auth_url: - get_param: [EndpointMap, KeystoneV3Internal, uri] + get_param: [EndpointMap, KeystoneV3Internal, uri_no_suffix] metadata_settings: get_attr: [ApacheServiceBase, role_data, metadata_settings] diff --git a/puppet/services/ceph-mon.yaml b/puppet/services/ceph-mon.yaml index 68ad69b7..0c61305d 100644 --- a/puppet/services/ceph-mon.yaml +++ b/puppet/services/ceph-mon.yaml @@ -113,3 +113,24 @@ outputs: get_attr: [CephBase, role_data, service_config_settings] step_config: | include ::tripleo::profile::base::ceph::mon + upgrade_batch_tasks: + # Note we perform these tasks in list order, but they are all step0 so + # we can perform a rolling upgrade of all mon nodes in step0, then a + # rolling upgrade of all osd nodes in step1 + # FIXME(shardy) I suspect we can use heat or ansible facts here instead? + - name: Get hostname + tags: step0 + shell: hostname -s + register: mon_id + - name: Stop Ceph Mon + tags: step0 + service: name=ceph-mon@{{mon_id.stdout}} pattern=ceph-mon state=stopped + - name: Update ceph packages + tags: step0 + yum: name=ceph-mon,ceph state=latest + - name: Start ceph-mon service + tags: step0 + service: name=ceph-mon@{{mon_id.stdout}} state=started + - name: ceph osd crush tunables default + tags: step0 + shell: ceph osd crush tunables default diff --git a/puppet/services/ceph-osd.yaml b/puppet/services/ceph-osd.yaml index df0ee6c3..e9ed6c29 100644 --- a/puppet/services/ceph-osd.yaml +++ b/puppet/services/ceph-osd.yaml @@ -45,3 +45,44 @@ outputs: - '6800-7300' step_config: | include ::tripleo::profile::base::ceph::osd + upgrade_batch_tasks: + - name: Get OSD IDs + tags: step1 + shell: ls /var/lib/ceph/osd | awk 'BEGIN { FS = "-" } ; { print $2 }' + register: osd_ids + # "so that mirrors aren't rebalanced as if the OSD died" - gfidente / leseb + - name: ceph osd set noout + tags: step1 + command: ceph osd set noout + - name: ceph osd set norebalance + tags: step1 + command: ceph osd set norebalance + - name: ceph osd set nodeep-scrub + tags: step1 + command: ceph osd set nodeep-scrub + - name: ceph osd set noscrub + tags: step1 + command: ceph osd set noscrub + - name: Stop Ceph OSD + tags: step1 + service: name=ceph-osd@$item state=stopped + with_items: "{{osd_ids.stdout.strip().split()}}" + - name: Update ceph OSD packages + tags: step1 + yum: name=ceph-osd state=latest + - name: Start ceph-osd service + tags: step1 + service: name=ceph-osd@$item state=started + with_items: "{{osd_ids.stdout.strip().split()}}" + - name: ceph osd unset noout + tags: step1 + command: ceph osd unset noout + - name: ceph osd unset norebalance + tags: step1 + command: ceph osd unset norebalance + - name: ceph osd unset nodeep-scrub + tags: step1 + command: ceph osd unset nodeep-scrub + - name: ceph osd unset noscrub + tags: step1 + command: ceph osd unset noscrub diff --git a/puppet/services/gnocchi-metricd.yaml b/puppet/services/gnocchi-metricd.yaml index e5f9a8e7..27700606 100644 --- a/puppet/services/gnocchi-metricd.yaml +++ b/puppet/services/gnocchi-metricd.yaml @@ -22,7 +22,7 @@ parameters: default: 'overcloud-gnocchi-metricd' type: string GnocchiMetricdWorkers: - default: '' + default: '%{::os_workers}' description: Number of workers for Gnocchi MetricD type: string diff --git a/puppet/services/ironic-api.yaml b/puppet/services/ironic-api.yaml index aebb37b2..ff91eb63 100644 --- a/puppet/services/ironic-api.yaml +++ b/puppet/services/ironic-api.yaml @@ -81,3 +81,7 @@ outputs: ironic::db::mysql::allowed_hosts: - '%' - "%{hiera('mysql_bind_host')}" + upgrade_tasks: + - name: Stop ironic_api service + tags: step2 + service: name=openstack-ironic-api state=stopped diff --git a/puppet/services/ironic-conductor.yaml b/puppet/services/ironic-conductor.yaml index 194afec7..a10c03a5 100644 --- a/puppet/services/ironic-conductor.yaml +++ b/puppet/services/ironic-conductor.yaml @@ -98,3 +98,10 @@ outputs: step_config: | include ::tripleo::profile::base::ironic::conductor + upgrade_tasks: + - name: Stop ironic_conductor service + tags: step2 + service: name=openstack-ironic-conductor state=stopped + - name: Sync ironic_conductor DB + tags: step5 + command: ironic-dbsync diff --git a/puppet/services/neutron-dhcp.yaml b/puppet/services/neutron-dhcp.yaml index 5e7de18e..bb4742c9 100644 --- a/puppet/services/neutron-dhcp.yaml +++ b/puppet/services/neutron-dhcp.yaml @@ -39,6 +39,10 @@ parameters: default: tag: openstack.neutron.agent.dhcp path: /var/log/neutron/dhcp-agent.log + NeutronDhcpAgentDnsmasqDnsServers: + default: [] + description: List of servers to use as dnsmasq forwarders + type: comma_delimited_list resources: @@ -64,6 +68,7 @@ outputs: - neutron::agents::dhcp::enable_isolated_metadata: {get_param: NeutronEnableIsolatedMetadata} neutron::agents::dhcp::enable_force_metadata: {get_param: NeutronEnableForceMetadata} neutron::agents::dhcp::enable_metadata_network: {get_param: NeutronEnableMetadataNetwork} + neutron::agents::dhcp::dnsmasq_dns_servers: {get_param: NeutronDhcpAgentDnsmasqDnsServers} tripleo.neutron_dhcp.firewall_rules: '115 neutron dhcp input': proto: 'udp' diff --git a/puppet/services/services.yaml b/puppet/services/services.yaml index 90268c78..80da5352 100644 --- a/puppet/services/services.yaml +++ b/puppet/services/services.yaml @@ -118,4 +118,9 @@ outputs: # Note we use distinct() here to filter any identical tasks, e.g yum update for all services expression: $.data.where($ != null).select($.get('upgrade_tasks')).where($ != null).flatten().distinct() data: {get_attr: [ServiceChain, role_data]} + upgrade_batch_tasks: + yaql: + # Note we use distinct() here to filter any identical tasks, e.g yum update for all services + expression: $.data.where($ != null).select($.get('upgrade_batch_tasks')).where($ != null).flatten().distinct() + data: {get_attr: [ServiceChain, role_data]} service_metadata_settings: {get_attr: [ServiceServerMetadataHook, metadata]} diff --git a/roles_data.yaml b/roles_data.yaml index 05b8b23b..a21ef961 100644 --- a/roles_data.yaml +++ b/roles_data.yaml @@ -17,8 +17,9 @@ # disable_constraints: (boolean) optional, whether to disable Nova and Glance # constraints for each role specified in the templates. # -# disable_upgrade_deployment: (boolean) optional, whether to run the composable upgrade -# steps for all services that are deployed on the particular role. +# upgrade_batch_size: (number): batch size for upgrades where tasks are +# specified by services to run in batches vs all nodes at once. +# This defaults to 1, but larger batches may be specified here. # # ServicesDefault: (list) optional default list of services to be deployed # on the role, defaults to an empty list. Sets the default for the |