diff options
Diffstat (limited to 'playbooks/common')
18 files changed, 641 insertions, 19 deletions
diff --git a/playbooks/common/openshift-cluster/update_repos_and_packages.yml b/playbooks/common/openshift-cluster/update_repos_and_packages.yml index 9a303c62d..88736ee03 100644 --- a/playbooks/common/openshift-cluster/update_repos_and_packages.yml +++ b/playbooks/common/openshift-cluster/update_repos_and_packages.yml @@ -4,7 +4,7 @@ openshift_deployment_type: "{{ deployment_type }}" roles: - role: rhel_subscribe - when: deployment_type == "enterprise" and + when: deployment_type in ["enterprise", "atomic-enterprise", "openshift-enterprise"] and ansible_distribution == "RedHat" and lookup('oo_option', 'rhel_skip_subscription') | default(rhsub_skip, True) | default('no', True) | lower in ['no', 'false'] diff --git a/playbooks/common/openshift-cluster/upgrades/files/versions.sh b/playbooks/common/openshift-cluster/upgrades/files/versions.sh index b46407ed7..3a1a8ebb1 100644 --- a/playbooks/common/openshift-cluster/upgrades/files/versions.sh +++ b/playbooks/common/openshift-cluster/upgrades/files/versions.sh @@ -1,8 +1,8 @@ #!/bin/bash -yum_installed=$(yum list installed -e 0 -q "$@" 2>&1 | tail -n +2 | awk '{ print $2 }' | tr '\n' ' ') +yum_installed=$(yum list installed -e 0 -q "$@" 2>&1 | tail -n +2 | awk '{ print $2 }' | sort -r | tr '\n' ' ') -yum_available=$(yum list available -e 0 -q "$@" 2>&1 | tail -n +2 | grep -v 'el7ose' | awk '{ print $2 }' | tr '\n' ' ') +yum_available=$(yum list available -e 0 -q "$@" 2>&1 | tail -n +2 | grep -v 'el7ose' | awk '{ print $2 }' | sort -r | tr '\n' ' ') echo "---" echo "curr_version: ${yum_installed}" diff --git a/playbooks/common/openshift-cluster/upgrades/v3_0_to_v3_1/upgrade.yml b/playbooks/common/openshift-cluster/upgrades/v3_0_to_v3_1/upgrade.yml index 68df2153d..8ec379109 100644 --- a/playbooks/common/openshift-cluster/upgrades/v3_0_to_v3_1/upgrade.yml +++ b/playbooks/common/openshift-cluster/upgrades/v3_0_to_v3_1/upgrade.yml @@ -248,7 +248,31 @@ config_base: "{{ hostvars[inventory_hostname].openshift.common.config_base }}" - set_fact: - master_certs_missing: True + openshift_master_certs_no_etcd: + - admin.crt + - master.kubelet-client.crt + - "{{ 'master.proxy-client.crt' if openshift.common.version_greater_than_3_1_or_1_1 else omit }}" + - master.server.crt + - openshift-master.crt + - openshift-registry.crt + - openshift-router.crt + - etcd.server.crt + openshift_master_certs_etcd: + - master.etcd-client.crt + + - set_fact: + openshift_master_certs: "{{ (openshift_master_certs_no_etcd | union(openshift_master_certs_etcd)) if (groups.oo_etcd_to_config is defined and groups.oo_etcd_to_config) else openshift_master_certs_no_etcd }}" + + - name: Check status of master certificates + stat: + path: "{{ openshift.common.config_base }}/master/{{ item }}" + with_items: openshift_master_certs + register: g_master_cert_stat_result + + - set_fact: + master_certs_missing: "{{ False in (g_master_cert_stat_result.results + | oo_collect(attribute='stat.exists') + | list ) }}" master_cert_subdir: master-{{ openshift.common.hostname }} master_cert_config_dir: "{{ openshift.common.config_base }}/master" @@ -262,8 +286,8 @@ | oo_flatten | unique }}" master_generated_certs_dir: "{{ openshift.common.config_base }}/generated-configs" masters_needing_certs: "{{ hostvars - | oo_select_keys(groups.oo_masters_to_config) - | difference([groups.oo_first_master.0]) }}" + | oo_select_keys(groups['oo_masters_to_config'] | difference(groups['oo_first_master'])) + | oo_filter_list(filter_attr='master_certs_missing') }}" sync_tmpdir: "{{ hostvars.localhost.g_master_mktemp.stdout }}" openshift_deployment_type: "{{ deployment_type }}" roles: @@ -398,6 +422,24 @@ - name: Ensure node service enabled service: name="{{ openshift.common.service_type }}-node" state=started enabled=yes + - name: Install Ceph storage plugin dependencies + action: "{{ ansible_pkg_mgr }} name=ceph-common state=present" + + - name: Install GlusterFS storage plugin dependencies + action: "{{ ansible_pkg_mgr }} name=glusterfs-fuse state=present" + + - name: Set sebooleans to allow gluster storage plugin access from containers + seboolean: + name: "{{ item }}" + state: yes + persistent: yes + when: ansible_selinux and ansible_selinux.status == "enabled" + with_items: + - virt_use_fusefs + - virt_sandbox_use_fusefs + register: sebool_result + failed_when: "'state' not in sebool_result and 'msg' in sebool_result and 'SELinux boolean {{ item }} does not exist' not in sebool_result.msg" + - set_fact: node_update_complete: True diff --git a/playbooks/common/openshift-cluster/upgrades/v3_1_minor/filter_plugins b/playbooks/common/openshift-cluster/upgrades/v3_1_minor/filter_plugins new file mode 120000 index 000000000..27ddaa18b --- /dev/null +++ b/playbooks/common/openshift-cluster/upgrades/v3_1_minor/filter_plugins @@ -0,0 +1 @@ +../../../../../filter_plugins
\ No newline at end of file diff --git a/playbooks/common/openshift-cluster/upgrades/v3_1_minor/library b/playbooks/common/openshift-cluster/upgrades/v3_1_minor/library new file mode 120000 index 000000000..53bed9684 --- /dev/null +++ b/playbooks/common/openshift-cluster/upgrades/v3_1_minor/library @@ -0,0 +1 @@ +../library
\ No newline at end of file diff --git a/playbooks/common/openshift-cluster/upgrades/v3_1_minor/lookup_plugins b/playbooks/common/openshift-cluster/upgrades/v3_1_minor/lookup_plugins new file mode 120000 index 000000000..cf407f69b --- /dev/null +++ b/playbooks/common/openshift-cluster/upgrades/v3_1_minor/lookup_plugins @@ -0,0 +1 @@ +../../../../../lookup_plugins
\ No newline at end of file diff --git a/playbooks/common/openshift-cluster/upgrades/v3_1_minor/post.yml b/playbooks/common/openshift-cluster/upgrades/v3_1_minor/post.yml new file mode 100644 index 000000000..d8336fcae --- /dev/null +++ b/playbooks/common/openshift-cluster/upgrades/v3_1_minor/post.yml @@ -0,0 +1,50 @@ +--- +############################################################################### +# Post upgrade - Upgrade default router, default registry and examples +############################################################################### +- name: Upgrade default router and default registry + hosts: oo_first_master + vars: + openshift_deployment_type: "{{ deployment_type }}" + registry_image: "{{ openshift.master.registry_url | replace( '${component}', 'docker-registry' ) | replace ( '${version}', 'v' + g_new_version ) }}" + router_image: "{{ openshift.master.registry_url | replace( '${component}', 'haproxy-router' ) | replace ( '${version}', 'v' + g_new_version ) }}" + oc_cmd: "{{ openshift.common.client_binary }} --config={{ openshift.common.config_base }}/master/admin.kubeconfig" + roles: + # Create the new templates shipped in 3.1.z, existing templates are left + # unmodified. This prevents the subsequent role definition for + # openshift_examples from failing when trying to replace templates that do + # not already exist. We could have potentially done a replace --force to + # create and update in one step. + - openshift_examples + # Update the existing templates + - role: openshift_examples + openshift_examples_import_command: replace + pre_tasks: + - name: Check for default router + command: > + {{ oc_cmd }} get -n default dc/router + register: _default_router + failed_when: false + changed_when: false + + - name: Check for default registry + command: > + {{ oc_cmd }} get -n default dc/docker-registry + register: _default_registry + failed_when: false + changed_when: false + + - name: Update router image to current version + when: _default_router.rc == 0 + command: > + {{ oc_cmd }} patch dc/router -p + '{"spec":{"template":{"spec":{"containers":[{"name":"router","image":"{{ router_image }}"}]}}}}' + --api-version=v1 + + - name: Update registry image to current version + when: _default_registry.rc == 0 + command: > + {{ oc_cmd }} patch dc/docker-registry -p + '{"spec":{"template":{"spec":{"containers":[{"name":"registry","image":"{{ registry_image }}"}]}}}}' + --api-version=v1 + diff --git a/playbooks/common/openshift-cluster/upgrades/v3_1_minor/pre.yml b/playbooks/common/openshift-cluster/upgrades/v3_1_minor/pre.yml new file mode 100644 index 000000000..91780de09 --- /dev/null +++ b/playbooks/common/openshift-cluster/upgrades/v3_1_minor/pre.yml @@ -0,0 +1,87 @@ +--- +############################################################################### +# Evaluate host groups and gather facts +############################################################################### +- name: Load openshift_facts + hosts: oo_masters_to_config:oo_nodes_to_config:oo_etcd_to_config:oo_lb_to_config + roles: + - openshift_facts + +############################################################################### +# Pre-upgrade checks +############################################################################### +- name: Verify upgrade can proceed + hosts: oo_first_master + vars: + openshift_master_ha: "{{ groups.oo_masters_to_config | length > 1 }}" + target_version: "{{ '1.1.1' if deployment_type == 'origin' else '3.1.1' }}" + gather_facts: no + tasks: + - fail: + msg: > + This upgrade is only supported for origin, openshift-enterprise, and online + deployment types + when: deployment_type not in ['origin','openshift-enterprise', 'online'] + + - fail: + msg: > + openshift_pkg_version is {{ openshift_pkg_version }} which is not a + valid version for a {{ target_version }} upgrade + when: openshift_pkg_version is defined and openshift_pkg_version.split('-',1).1 | version_compare(target_version ,'<') + +- name: Verify upgrade can proceed + hosts: oo_masters_to_config:oo_nodes_to_config + vars: + target_version: "{{ '1.1.1' if deployment_type == 'origin' else '3.1.1' }}" + tasks: + - name: Clean package cache + command: "{{ ansible_pkg_mgr }} clean all" + + - set_fact: + g_new_service_name: "{{ 'origin' if deployment_type =='origin' else 'atomic-openshift' }}" + + - name: Determine available versions + script: ../files/versions.sh {{ g_new_service_name }} openshift + register: g_versions_result + + - set_fact: + g_aos_versions: "{{ g_versions_result.stdout | from_yaml }}" + + - set_fact: + g_new_version: "{{ g_aos_versions.curr_version.split('-', 1).0 if g_aos_versions.avail_version is none else g_aos_versions.avail_version.split('-', 1).0 }}" + + - fail: + msg: This playbook requires Origin 1.1 or later + when: deployment_type == 'origin' and g_aos_versions.curr_version | version_compare('1.1','<') + + - fail: + msg: This playbook requires Atomic Enterprise Platform/OpenShift Enterprise 3.1 or later + when: deployment_type == 'atomic-openshift' and g_aos_versions.curr_version | version_compare('3.1','<') + + - fail: + msg: Upgrade packages not found + when: (g_aos_versions.avail_version | default(g_aos_versions.curr_version, true) | version_compare(target_version, '<')) + + - set_fact: + pre_upgrade_complete: True + + +############################################################################## +# Gate on pre-upgrade checks +############################################################################## +- name: Gate on pre-upgrade checks + hosts: localhost + connection: local + become: no + vars: + pre_upgrade_hosts: "{{ groups.oo_masters_to_config | union(groups.oo_nodes_to_config) }}" + tasks: + - set_fact: + pre_upgrade_completed: "{{ hostvars + | oo_select_keys(pre_upgrade_hosts) + | oo_collect('inventory_hostname', {'pre_upgrade_complete': true}) }}" + - set_fact: + pre_upgrade_failed: "{{ pre_upgrade_hosts | difference(pre_upgrade_completed) }}" + - fail: + msg: "Upgrade cannot continue. The following hosts did not complete pre-upgrade checks: {{ pre_upgrade_failed | join(',') }}" + when: pre_upgrade_failed | length > 0 diff --git a/playbooks/common/openshift-cluster/upgrades/v3_1_minor/roles b/playbooks/common/openshift-cluster/upgrades/v3_1_minor/roles new file mode 120000 index 000000000..6bc1a7aef --- /dev/null +++ b/playbooks/common/openshift-cluster/upgrades/v3_1_minor/roles @@ -0,0 +1 @@ +../../../../../roles
\ No newline at end of file diff --git a/playbooks/common/openshift-cluster/upgrades/v3_1_minor/upgrade.yml b/playbooks/common/openshift-cluster/upgrades/v3_1_minor/upgrade.yml new file mode 100644 index 000000000..81dbba1e3 --- /dev/null +++ b/playbooks/common/openshift-cluster/upgrades/v3_1_minor/upgrade.yml @@ -0,0 +1,137 @@ +--- +############################################################################### +# The restart playbook should be run after this playbook completes. +############################################################################### + +############################################################################### +# Upgrade Masters +############################################################################### +- name: Upgrade master packages and configuration + hosts: oo_masters_to_config + vars: + openshift_version: "{{ openshift_pkg_version | default('') }}" + tasks: + - name: Upgrade master packages + command: "{{ ansible_pkg_mgr}} update -y {{ openshift.common.service_type }}-master{{ openshift_version }}" + + - name: Ensure python-yaml present for config upgrade + action: "{{ ansible_pkg_mgr }} name=PyYAML state=present" + when: not openshift.common.is_atomic | bool + +# Currently 3.1.1 does not have any new configuration settings +# +# - name: Upgrade master configuration +# openshift_upgrade_config: +# from_version: '3.0' +# to_version: '3.1' +# role: master +# config_base: "{{ hostvars[inventory_hostname].openshift.common.config_base }}" + +- name: Set master update status to complete + hosts: oo_masters_to_config + tasks: + - set_fact: + master_update_complete: True + +############################################################################## +# Gate on master update complete +############################################################################## +- name: Gate on master update + hosts: localhost + connection: local + become: no + tasks: + - set_fact: + master_update_completed: "{{ hostvars + | oo_select_keys(groups.oo_masters_to_config) + | oo_collect('inventory_hostname', {'master_update_complete': true}) }}" + - set_fact: + master_update_failed: "{{ groups.oo_masters_to_config | difference(master_update_completed) }}" + - fail: + msg: "Upgrade cannot continue. The following masters did not finish updating: {{ master_update_failed | join(',') }}" + when: master_update_failed | length > 0 + +############################################################################### +# Upgrade Nodes +############################################################################### +- name: Upgrade nodes + hosts: oo_nodes_to_config + vars: + openshift_version: "{{ openshift_pkg_version | default('') }}" + roles: + - openshift_facts + tasks: + - name: Upgrade node packages + command: "{{ ansible_pkg_mgr }} update -y {{ openshift.common.service_type }}-node{{ openshift_version }}" + + - name: Restart node service + service: name="{{ openshift.common.service_type }}-node" state=restarted + + - set_fact: + node_update_complete: True + +############################################################################## +# Gate on nodes update +############################################################################## +- name: Gate on nodes update + hosts: localhost + connection: local + become: no + tasks: + - set_fact: + node_update_completed: "{{ hostvars + | oo_select_keys(groups.oo_nodes_to_config) + | oo_collect('inventory_hostname', {'node_update_complete': true}) }}" + - set_fact: + node_update_failed: "{{ groups.oo_nodes_to_config | difference(node_update_completed) }}" + - fail: + msg: "Upgrade cannot continue. The following nodes did not finish updating: {{ node_update_failed | join(',') }}" + when: node_update_failed | length > 0 + +############################################################################### +# Reconcile Cluster Roles and Cluster Role Bindings +############################################################################### +- name: Reconcile Cluster Roles and Cluster Role Bindings + hosts: oo_masters_to_config + vars: + origin_reconcile_bindings: "{{ deployment_type == 'origin' and g_new_version | version_compare('1.0.6', '>') }}" + ent_reconcile_bindings: true + openshift_master_ha: "{{ groups.oo_masters_to_config | length > 1 }}" + tasks: + - name: Reconcile Cluster Roles + command: > + {{ openshift.common.admin_binary}} --config={{ openshift.common.config_base }}/master/admin.kubeconfig + policy reconcile-cluster-roles --confirm + run_once: true + + - name: Reconcile Cluster Role Bindings + command: > + {{ openshift.common.admin_binary}} --config={{ openshift.common.config_base }}/master/admin.kubeconfig + policy reconcile-cluster-role-bindings + --exclude-groups=system:authenticated + --exclude-groups=system:unauthenticated + --exclude-users=system:anonymous + --additive-only=true --confirm + when: origin_reconcile_bindings | bool or ent_reconcile_bindings | bool + run_once: true + + - set_fact: + reconcile_complete: True + +############################################################################## +# Gate on reconcile +############################################################################## +- name: Gate on reconcile + hosts: localhost + connection: local + become: no + tasks: + - set_fact: + reconcile_completed: "{{ hostvars + | oo_select_keys(groups.oo_masters_to_config) + | oo_collect('inventory_hostname', {'reconcile_complete': true}) }}" + - set_fact: + reconcile_failed: "{{ groups.oo_masters_to_config | difference(reconcile_completed) }}" + - fail: + msg: "Upgrade cannot continue. The following masters did not finish reconciling: {{ reconcile_failed | join(',') }}" + when: reconcile_failed | length > 0 diff --git a/playbooks/common/openshift-etcd/config.yml b/playbooks/common/openshift-etcd/config.yml index 9a5ae0e6b..d23a54511 100644 --- a/playbooks/common/openshift-etcd/config.yml +++ b/playbooks/common/openshift-etcd/config.yml @@ -89,7 +89,6 @@ roles: - etcd - role: nickhammond.logrotate - when: not openshift.common.is_containerized | bool - name: Delete temporary directory on localhost hosts: localhost diff --git a/playbooks/common/openshift-master/config.yml b/playbooks/common/openshift-master/config.yml index 677c274c4..12497bf94 100644 --- a/playbooks/common/openshift-master/config.yml +++ b/playbooks/common/openshift-master/config.yml @@ -43,6 +43,7 @@ api_port: "{{ openshift_master_api_port | default(None) }}" api_url: "{{ openshift_master_api_url | default(None) }}" api_use_ssl: "{{ openshift_master_api_use_ssl | default(None) }}" + controllers_port: "{{ openshift_master_controllers_port | default(None) }}" public_api_url: "{{ openshift_master_public_api_url | default(None) }}" cluster_hostname: "{{ openshift_master_cluster_hostname | default(None) }}" cluster_public_hostname: "{{ openshift_master_cluster_public_hostname | default(None) }}" @@ -51,6 +52,7 @@ console_url: "{{ openshift_master_console_url | default(None) }}" console_use_ssl: "{{ openshift_master_console_use_ssl | default(None) }}" public_console_url: "{{ openshift_master_public_console_url | default(None) }}" + portal_net: "{{ openshift_master_portal_net | default(None) }}" - name: Check status of external etcd certificatees stat: path: "{{ openshift.common.config_base }}/master/{{ item }}" @@ -84,6 +86,7 @@ etcd_generated_certs_dir: /etc/etcd/generated_certs etcd_needing_client_certs: "{{ hostvars | oo_select_keys(groups['oo_masters_to_config']) + | default([]) | oo_filter_list(filter_attr='etcd_client_certs_missing') }}" sync_tmpdir: "{{ hostvars.localhost.g_master_mktemp.stdout }}" roles: @@ -217,6 +220,7 @@ hosts: oo_lb_to_config vars: sync_tmpdir: "{{ hostvars.localhost.g_master_mktemp.stdout }}" + haproxy_frontend_port: "{{ hostvars[groups.oo_first_master.0].openshift.master.api_port }}" haproxy_frontends: - name: atomic-openshift-api mode: tcp @@ -312,6 +316,7 @@ - name: Configure master instances hosts: oo_masters_to_config + any_errors_fatal: true serial: 1 vars: sync_tmpdir: "{{ hostvars.localhost.g_master_mktemp.stdout }}" @@ -333,9 +338,10 @@ roles: - openshift_master - role: nickhammond.logrotate - when: not openshift.common.is_containerized | bool - role: fluentd_master when: openshift.common.use_fluentd | bool + - role: nuage_master + when: openshift.common.use_nuage | bool post_tasks: - name: Create group for deployment type group_by: key=oo_masters_deployment_type_{{ openshift.common.deployment_type }} @@ -362,7 +368,7 @@ cockpit_plugins: "{{ osm_cockpit_plugins | default(['cockpit-kubernetes']) }}" roles: - role: cockpit - when: not openshift.common.is_containerized and ( deployment_type in ['atomic-enterprise','openshift-enterprise'] ) and + when: not openshift.common.is_atomic and ( deployment_type in ['atomic-enterprise','openshift-enterprise'] ) and (osm_use_cockpit | bool or osm_use_cockpit is undefined ) - name: Configure flannel diff --git a/playbooks/common/openshift-master/restart.yml b/playbooks/common/openshift-master/restart.yml new file mode 100644 index 000000000..02449e40d --- /dev/null +++ b/playbooks/common/openshift-master/restart.yml @@ -0,0 +1,151 @@ +--- +- include: ../openshift-cluster/evaluate_groups.yml + +- name: Validate configuration for rolling restart + hosts: oo_masters_to_config + roles: + - openshift_facts + tasks: + - fail: + msg: "openshift_rolling_restart_mode must be set to either 'services' or 'system'" + when: openshift_rolling_restart_mode is defined and openshift_rolling_restart_mode not in ["services", "system"] + - openshift_facts: + role: "{{ item.role }}" + local_facts: "{{ item.local_facts }}" + with_items: + - role: common + local_facts: + rolling_restart_mode: "{{ openshift_rolling_restart_mode | default('services') }}" + - role: master + local_facts: + cluster_method: "{{ openshift_master_cluster_method | default(None) }}" + +# Creating a temp file on localhost, we then check each system that will +# be rebooted to see if that file exists, if so we know we're running +# ansible on a machine that needs a reboot, and we need to error out. +- name: Create temp file on localhost + hosts: localhost + connection: local + become: no + gather_facts: no + tasks: + - local_action: command mktemp + register: mktemp + changed_when: false + +- name: Check if temp file exists on any masters + hosts: oo_masters_to_config + tasks: + - stat: path="{{ hostvars.localhost.mktemp.stdout }}" + register: exists + changed_when: false + +- name: Cleanup temp file on localhost + hosts: localhost + connection: local + become: no + gather_facts: no + tasks: + - file: path="{{ hostvars.localhost.mktemp.stdout }}" state=absent + changed_when: false + +- name: Warn if restarting the system where ansible is running + hosts: oo_masters_to_config + tasks: + - pause: + prompt: > + Warning: Running playbook from a host that will be restarted! + Press CTRL+C and A to abort playbook execution. You may + continue by pressing ENTER but the playbook will stop + executing after this system has been restarted and services + must be verified manually. To only restart services, set + openshift_master_rolling_restart_mode=services in host + inventory and relaunch the playbook. + when: exists.stat.exists and openshift.common.rolling_restart_mode == 'system' + - set_fact: + current_host: "{{ exists.stat.exists }}" + when: openshift.common.rolling_restart_mode == 'system' + +- name: Determine which masters are currently active + hosts: oo_masters_to_config + any_errors_fatal: true + tasks: + - name: Check master service status + command: > + systemctl is-active {{ openshift.common.service_type }}-master + register: active_check_output + when: openshift.master.cluster_method | default(None) == 'pacemaker' + failed_when: false + changed_when: false + - set_fact: + is_active: "{{ active_check_output.stdout == 'active' }}" + when: openshift.master.cluster_method | default(None) == 'pacemaker' + +- name: Evaluate master groups + hosts: localhost + become: no + tasks: + - fail: + msg: > + Did not receive active status from any masters. Please verify pacemaker cluster. + when: "{{ hostvars[groups.oo_first_master.0].openshift.master.cluster_method | default(None) == 'pacemaker' and 'True' not in (hostvars + | oo_select_keys(groups['oo_masters_to_config']) + | oo_collect('is_active') + | list) }}" + - name: Evaluate oo_active_masters + add_host: + name: "{{ item }}" + groups: oo_active_masters + ansible_ssh_user: "{{ g_ssh_user | default(omit) }}" + ansible_sudo: "{{ g_sudo | default(omit) }}" + with_items: "{{ groups.oo_masters_to_config | default([]) }}" + when: (hostvars[item]['is_active'] | default(false)) | bool + - name: Evaluate oo_current_masters + add_host: + name: "{{ item }}" + groups: oo_current_masters + ansible_ssh_user: "{{ g_ssh_user | default(omit) }}" + ansible_sudo: "{{ g_sudo | default(omit) }}" + with_items: "{{ groups.oo_masters_to_config | default([]) }}" + when: (hostvars[item]['current_host'] | default(false)) | bool + +- name: Validate pacemaker cluster + hosts: oo_active_masters + tasks: + - name: Retrieve pcs status + command: pcs status + register: pcs_status_output + changed_when: false + - fail: + msg: > + Pacemaker cluster validation failed. One or more nodes are not online. + when: not (pcs_status_output.stdout | validate_pcs_cluster(groups.oo_masters_to_config)) | bool + +- name: Restart masters + hosts: oo_masters_to_config:!oo_active_masters:!oo_current_masters + vars: + openshift_master_ha: "{{ groups.oo_masters_to_config | length > 1 }}" + serial: 1 + tasks: + - include: restart_hosts.yml + when: openshift.common.rolling_restart_mode == 'system' + - include: restart_services.yml + when: openshift.common.rolling_restart_mode == 'services' + +- name: Restart active masters + hosts: oo_active_masters + serial: 1 + tasks: + - include: restart_hosts_pacemaker.yml + when: openshift.common.rolling_restart_mode == 'system' + - include: restart_services_pacemaker.yml + when: openshift.common.rolling_restart_mode == 'services' + +- name: Restart current masters + hosts: oo_current_masters + serial: 1 + tasks: + - include: restart_hosts.yml + when: openshift.common.rolling_restart_mode == 'system' + - include: restart_services.yml + when: openshift.common.rolling_restart_mode == 'services' diff --git a/playbooks/common/openshift-master/restart_hosts.yml b/playbooks/common/openshift-master/restart_hosts.yml new file mode 100644 index 000000000..ff206f5a2 --- /dev/null +++ b/playbooks/common/openshift-master/restart_hosts.yml @@ -0,0 +1,39 @@ +- name: Restart master system + # https://github.com/ansible/ansible/issues/10616 + shell: sleep 2 && shutdown -r now "OpenShift Ansible master rolling restart" + async: 1 + poll: 0 + ignore_errors: true + become: yes +# When cluster_method != pacemaker we can ensure the api_port is +# available. +- name: Wait for master API to come back online + become: no + local_action: + module: wait_for + host="{{ inventory_hostname }}" + state=started + delay=10 + port="{{ openshift.master.api_port }}" + when: openshift.master.cluster_method != 'pacemaker' +- name: Wait for master to start + become: no + local_action: + module: wait_for + host="{{ inventory_hostname }}" + state=started + delay=10 + port=22 + when: openshift.master.cluster_method == 'pacemaker' +- name: Wait for master to become available + command: pcs status + register: pcs_status_output + until: pcs_status_output.stdout | validate_pcs_cluster([inventory_hostname]) | bool + retries: 15 + delay: 2 + changed_when: false + when: openshift.master.cluster_method == 'pacemaker' +- fail: + msg: > + Pacemaker cluster validation failed {{ inventory hostname }} is not online. + when: openshift.master.cluster_method == 'pacemaker' and not (pcs_status_output.stdout | validate_pcs_cluster([inventory_hostname])) | bool diff --git a/playbooks/common/openshift-master/restart_hosts_pacemaker.yml b/playbooks/common/openshift-master/restart_hosts_pacemaker.yml new file mode 100644 index 000000000..c9219e8de --- /dev/null +++ b/playbooks/common/openshift-master/restart_hosts_pacemaker.yml @@ -0,0 +1,25 @@ +- name: Fail over master resource + command: > + pcs resource move master {{ hostvars | oo_select_keys(groups['oo_masters_to_config']) | oo_collect('openshift.common.hostname', {'is_active': 'False'}) | list | first }} +- name: Wait for master API to come back online + become: no + local_action: + module: wait_for + host="{{ openshift.master.cluster_hostname }}" + state=started + delay=10 + port="{{ openshift.master.api_port }}" +- name: Restart master system + # https://github.com/ansible/ansible/issues/10616 + shell: sleep 2 && shutdown -r now "OpenShift Ansible master rolling restart" + async: 1 + poll: 0 + ignore_errors: true + become: yes +- name: Wait for master to start + become: no + local_action: + module: wait_for + host="{{ inventory_hostname }}" + state=started + delay=10 diff --git a/playbooks/common/openshift-master/restart_services.yml b/playbooks/common/openshift-master/restart_services.yml new file mode 100644 index 000000000..5e539cd65 --- /dev/null +++ b/playbooks/common/openshift-master/restart_services.yml @@ -0,0 +1,27 @@ +- name: Restart master + service: + name: "{{ openshift.common.service_type }}-master" + state: restarted + when: not openshift_master_ha | bool +- name: Restart master API + service: + name: "{{ openshift.common.service_type }}-master-api" + state: restarted + when: openshift_master_ha | bool and openshift.master.cluster_method != 'pacemaker' +- name: Wait for master API to come back online + become: no + local_action: + module: wait_for + host="{{ inventory_hostname }}" + state=started + delay=10 + port="{{ openshift.master.api_port }}" + when: openshift_master_ha | bool and openshift.master.cluster_method != 'pacemaker' +- name: Restart master controllers + service: + name: "{{ openshift.common.service_type }}-master-controllers" + state: restarted + # Ignore errrors since it is possible that type != simple for + # pre-3.1.1 installations. + ignore_errors: true + when: openshift_master_ha | bool and openshift.master.cluster_method != 'pacemaker' diff --git a/playbooks/common/openshift-master/restart_services_pacemaker.yml b/playbooks/common/openshift-master/restart_services_pacemaker.yml new file mode 100644 index 000000000..e738f3fb6 --- /dev/null +++ b/playbooks/common/openshift-master/restart_services_pacemaker.yml @@ -0,0 +1,10 @@ +- name: Restart master services + command: pcs resource restart master +- name: Wait for master API to come back online + become: no + local_action: + module: wait_for + host="{{ openshift.master.cluster_hostname }}" + state=started + delay=10 + port="{{ openshift.master.api_port }}" diff --git a/playbooks/common/openshift-node/config.yml b/playbooks/common/openshift-node/config.yml index 483a7768c..81ec9ab6d 100644 --- a/playbooks/common/openshift-node/config.yml +++ b/playbooks/common/openshift-node/config.yml @@ -154,21 +154,15 @@ validate_checksum: yes with_items: nodes_needing_certs -- name: Configure node instances +- name: Deploy node certificates hosts: oo_nodes_to_config vars: sync_tmpdir: "{{ hostvars.localhost.mktemp.stdout }}" - openshift_node_master_api_url: "{{ hostvars[groups.oo_first_master.0].openshift.master.api_url }}" - # TODO: Prefix flannel role variables. - etcd_urls: "{{ hostvars[groups.oo_first_master.0].openshift.master.etcd_urls }}" - embedded_etcd: "{{ hostvars[groups.oo_first_master.0].openshift.master.embedded_etcd }}" - openshift_node_first_master_ip: "{{ hostvars[groups.oo_first_master.0].openshift.common.ip }}" - pre_tasks: + tasks: - name: Ensure certificate directory exists file: path: "{{ node_cert_dir }}" state: directory - # TODO: notify restart node # possibly test service started time against certificate/config file # timestamps in node to trigger notify @@ -177,12 +171,50 @@ src: "{{ sync_tmpdir }}/{{ node_subdir }}.tgz" dest: "{{ node_cert_dir }}" when: certs_missing + +- name: Evaluate node groups + hosts: localhost + become: no + tasks: + - name: Evaluate oo_containerized_master_nodes + add_host: + name: "{{ item }}" + groups: oo_containerized_master_nodes + ansible_ssh_user: "{{ g_ssh_user | default(omit) }}" + ansible_sudo: "{{ g_sudo | default(omit) }}" + with_items: "{{ groups.oo_nodes_to_config | default([]) }}" + when: hostvars[item].openshift.common.is_containerized | bool and (item in groups.oo_nodes_to_config and item in groups.oo_masters_to_config) + +- name: Configure node instances + hosts: oo_containerized_master_nodes + serial: 1 + vars: + openshift_node_master_api_url: "{{ hostvars[groups.oo_first_master.0].openshift.master.api_url }}" + openshift_node_first_master_ip: "{{ hostvars[groups.oo_first_master.0].openshift.common.ip }}" roles: - openshift_node + +- name: Configure node instances + hosts: oo_nodes_to_config:!oo_containerized_master_nodes + vars: + openshift_node_master_api_url: "{{ hostvars[groups.oo_first_master.0].openshift.master.api_url }}" + openshift_node_first_master_ip: "{{ hostvars[groups.oo_first_master.0].openshift.common.ip }}" + roles: + - openshift_node + +- name: Additional node config + hosts: oo_nodes_to_config + vars: + # TODO: Prefix flannel role variables. + openshift_node_master_api_url: "{{ hostvars[groups.oo_first_master.0].openshift.master.api_url }}" + etcd_urls: "{{ hostvars[groups.oo_first_master.0].openshift.master.etcd_urls }}" + embedded_etcd: "{{ hostvars[groups.oo_first_master.0].openshift.master.embedded_etcd }}" + roles: - role: flannel when: openshift.common.use_flannel | bool + - role: nuage_node + when: openshift.common.use_nuage | bool - role: nickhammond.logrotate - when: not openshift.common.is_containerized | bool - role: fluentd_node when: openshift.common.use_fluentd | bool tasks: @@ -215,6 +247,19 @@ | oo_collect('openshift.common.hostname') }}" openshift_node_vars: "{{ hostvars | oo_select_keys(groups['oo_nodes_to_config']) }}" pre_tasks: - + # Necessary because when you're on a node that's also a master the master will be + # restarted after the node restarts docker and it will take up to 60 seconds for + # systemd to start the master again + - name: Wait for master API to become available before proceeding + # Using curl here since the uri module requires python-httplib2 and + # wait_for port doesn't provide health information. + command: > + curl -k --head --silent {{ openshift.master.api_url }} + register: api_available_output + until: api_available_output.stdout.find("200 OK") != -1 + retries: 120 + delay: 1 + changed_when: false + when: openshift.common.is_containerized | bool roles: - openshift_manage_node |