diff options
Diffstat (limited to 'roles/openshift_node')
16 files changed, 202 insertions, 138 deletions
diff --git a/roles/openshift_node/defaults/main.yml b/roles/openshift_node/defaults/main.yml index 5904ca9bc..47073ee0f 100644 --- a/roles/openshift_node/defaults/main.yml +++ b/roles/openshift_node/defaults/main.yml @@ -8,7 +8,7 @@ os_firewall_allow: port: 443/tcp - service: OpenShift OVS sdn port: 4789/udp - when: openshift.common.use_openshift_sdn | bool + when: openshift.common.use_openshift_sdn | default(true) | bool - service: Calico BGP Port port: 179/tcp when: openshift.common.use_calico | bool diff --git a/roles/openshift_node/handlers/main.yml b/roles/openshift_node/handlers/main.yml index a6bd12d4e..f2c45a4bd 100644 --- a/roles/openshift_node/handlers/main.yml +++ b/roles/openshift_node/handlers/main.yml @@ -3,10 +3,15 @@ systemd: name: openvswitch state: restarted - when: (not skip_node_svc_handlers | default(False) | bool) and not (ovs_service_status_changed | default(false) | bool) and openshift.common.use_openshift_sdn | bool + when: (not skip_node_svc_handlers | default(False) | bool) and not (ovs_service_status_changed | default(false) | bool) and openshift.common.use_openshift_sdn | default(true) | bool + register: l_openshift_node_stop_openvswitch_result + until: not l_openshift_node_stop_openvswitch_result | failed + retries: 3 + delay: 30 notify: - restart openvswitch pause + - name: restart openvswitch pause pause: seconds=15 when: (not skip_node_svc_handlers | default(False) | bool) and openshift.common.is_containerized | bool @@ -15,7 +20,13 @@ systemd: name: "{{ openshift.common.service_type }}-node" state: restarted - when: (not skip_node_svc_handlers | default(False) | bool) and not (node_service_status_changed | default(false) | bool) + register: l_openshift_node_restart_node_result + until: not l_openshift_node_restart_node_result | failed + retries: 3 + delay: 30 + when: + - (not skip_node_svc_handlers | default(False) | bool) + - not (node_service_status_changed | default(false) | bool) - name: reload sysctl.conf command: /sbin/sysctl -p diff --git a/roles/openshift_node/meta/main.yml b/roles/openshift_node/meta/main.yml index 3b7e8126a..4fb841add 100644 --- a/roles/openshift_node/meta/main.yml +++ b/roles/openshift_node/meta/main.yml @@ -12,6 +12,7 @@ galaxy_info: categories: - cloud dependencies: +- role: openshift_node_facts - role: lib_openshift - role: openshift_common - role: openshift_clock @@ -30,7 +31,7 @@ dependencies: os_firewall_allow: - service: OpenShift OVS sdn port: 4789/udp - when: openshift.common.use_openshift_sdn | bool + when: openshift.common.use_openshift_sdn | default(true) | bool - role: os_firewall os_firewall_allow: - service: Calico BGP Port diff --git a/roles/openshift_node/tasks/config/configure-node-settings.yml b/roles/openshift_node/tasks/config/configure-node-settings.yml new file mode 100644 index 000000000..1186062eb --- /dev/null +++ b/roles/openshift_node/tasks/config/configure-node-settings.yml @@ -0,0 +1,16 @@ +--- +- name: Configure Node settings + lineinfile: + dest: /etc/sysconfig/{{ openshift.common.service_type }}-node + regexp: "{{ item.regex }}" + line: "{{ item.line }}" + create: true + with_items: + - regex: '^OPTIONS=' + line: "OPTIONS=--loglevel={{ openshift.node.debug_level | default(2) }}" + - regex: '^CONFIG_FILE=' + line: "CONFIG_FILE={{ openshift.common.config_base }}/node/node-config.yaml" + - regex: '^IMAGE_VERSION=' + line: "IMAGE_VERSION={{ openshift_image_tag }}" + notify: + - restart node diff --git a/roles/openshift_node/tasks/config/configure-proxy-settings.yml b/roles/openshift_node/tasks/config/configure-proxy-settings.yml new file mode 100644 index 000000000..d60794305 --- /dev/null +++ b/roles/openshift_node/tasks/config/configure-proxy-settings.yml @@ -0,0 +1,17 @@ +--- +- name: Configure Proxy Settings + lineinfile: + dest: /etc/sysconfig/{{ openshift.common.service_type }}-node + regexp: "{{ item.regex }}" + line: "{{ item.line }}" + create: true + with_items: + - regex: '^HTTP_PROXY=' + line: "HTTP_PROXY={{ openshift.common.http_proxy | default('') }}" + - regex: '^HTTPS_PROXY=' + line: "HTTPS_PROXY={{ openshift.common.https_proxy | default('') }}" + - regex: '^NO_PROXY=' + line: "NO_PROXY={{ openshift.common.no_proxy | default([]) }},{{ openshift.common.portal_net }},{{ hostvars[groups.oo_first_master.0].openshift.master.sdn_cluster_network_cidr }}" + when: ('http_proxy' in openshift.common and openshift.common.http_proxy != '') + notify: + - restart node diff --git a/roles/openshift_node/tasks/config/install-node-deps-docker-service-file.yml b/roles/openshift_node/tasks/config/install-node-deps-docker-service-file.yml new file mode 100644 index 000000000..ee91a88ab --- /dev/null +++ b/roles/openshift_node/tasks/config/install-node-deps-docker-service-file.yml @@ -0,0 +1,8 @@ +--- +- name: Install Node dependencies docker service file + template: + dest: "/etc/systemd/system/{{ openshift.common.service_type }}-node-dep.service" + src: openshift.docker.node.dep.service + notify: + - reload systemd units + - restart node diff --git a/roles/openshift_node/tasks/config/install-node-docker-service-file.yml b/roles/openshift_node/tasks/config/install-node-docker-service-file.yml new file mode 100644 index 000000000..f92ff79b5 --- /dev/null +++ b/roles/openshift_node/tasks/config/install-node-docker-service-file.yml @@ -0,0 +1,8 @@ +--- +- name: Install Node docker service file + template: + dest: "/etc/systemd/system/{{ openshift.common.service_type }}-node.service" + src: openshift.docker.node.service + notify: + - reload systemd units + - restart node diff --git a/roles/openshift_node/tasks/config/install-ovs-docker-service-file.yml b/roles/openshift_node/tasks/config/install-ovs-docker-service-file.yml new file mode 100644 index 000000000..c2c5ea1d4 --- /dev/null +++ b/roles/openshift_node/tasks/config/install-ovs-docker-service-file.yml @@ -0,0 +1,8 @@ +--- +- name: Install OpenvSwitch docker service file + template: + dest: "/etc/systemd/system/openvswitch.service" + src: openvswitch.docker.service + notify: + - reload systemd units + - restart openvswitch diff --git a/roles/openshift_node/tasks/config/install-ovs-service-env-file.yml b/roles/openshift_node/tasks/config/install-ovs-service-env-file.yml new file mode 100644 index 000000000..1d75a3355 --- /dev/null +++ b/roles/openshift_node/tasks/config/install-ovs-service-env-file.yml @@ -0,0 +1,8 @@ +--- +- name: Create the openvswitch service env file + template: + src: openvswitch.sysconfig.j2 + dest: /etc/sysconfig/openvswitch + notify: + - reload systemd units + - restart openvswitch diff --git a/roles/openshift_node/tasks/config/workaround-bz1331590-ovs-oom-fix.yml b/roles/openshift_node/tasks/config/workaround-bz1331590-ovs-oom-fix.yml new file mode 100644 index 000000000..5df1abc79 --- /dev/null +++ b/roles/openshift_node/tasks/config/workaround-bz1331590-ovs-oom-fix.yml @@ -0,0 +1,13 @@ +--- +# May be a temporary workaround. +# https://bugzilla.redhat.com/show_bug.cgi?id=1331590 +- name: Create OpenvSwitch service.d directory + file: path=/etc/systemd/system/openvswitch.service.d/ state=directory + +- name: Install OpenvSwitch service OOM fix + template: + dest: "/etc/systemd/system/openvswitch.service.d/01-avoid-oom.conf" + src: openvswitch-avoid-oom.conf + notify: + - reload systemd units + - restart openvswitch diff --git a/roles/openshift_node/tasks/main.yml b/roles/openshift_node/tasks/main.yml index 573051504..87b1f6537 100644 --- a/roles/openshift_node/tasks/main.yml +++ b/roles/openshift_node/tasks/main.yml @@ -6,34 +6,6 @@ (not ansible_selinux or ansible_selinux.status != 'enabled') and deployment_type in ['enterprise', 'online', 'atomic-enterprise', 'openshift-enterprise'] -- name: Set node facts - openshift_facts: - role: "{{ item.role }}" - local_facts: "{{ item.local_facts }}" - with_items: - # Reset node labels to an empty dictionary. - - role: node - local_facts: - labels: {} - - role: node - local_facts: - annotations: "{{ openshift_node_annotations | default(none) }}" - debug_level: "{{ openshift_node_debug_level | default(openshift.common.debug_level) }}" - iptables_sync_period: "{{ openshift_node_iptables_sync_period | default(None) }}" - kubelet_args: "{{ openshift_node_kubelet_args | default(None) }}" - labels: "{{ lookup('oo_option', 'openshift_node_labels') | default( openshift_node_labels | default(none), true) }}" - registry_url: "{{ oreg_url_node | default(oreg_url) | default(None) }}" - schedulable: "{{ openshift_schedulable | default(openshift_scheduleable) | default(None) }}" - sdn_mtu: "{{ openshift_node_sdn_mtu | default(None) }}" - storage_plugin_deps: "{{ osn_storage_plugin_deps | default(None) }}" - set_node_ip: "{{ openshift_set_node_ip | default(None) }}" - node_image: "{{ osn_image | default(None) }}" - ovs_image: "{{ osn_ovs_image | default(None) }}" - proxy_mode: "{{ openshift_node_proxy_mode | default('iptables') }}" - local_quota_per_fsgroup: "{{ openshift_node_local_quota_per_fsgroup | default(None) }}" - dns_ip: "{{ openshift_dns_ip | default(none) | get_dns_ip(hostvars[inventory_hostname])}}" - env_vars: "{{ openshift_node_env_vars | default(None) }}" - # https://docs.openshift.com/container-platform/3.4/admin_guide/overcommit.html#disabling-swap-memory - name: Check for swap usage command: grep "^[^#].*swap" /etc/fstab @@ -90,7 +62,9 @@ package: name: "{{ openshift.common.service_type }}-sdn-ovs{{ openshift_pkg_version | oo_image_tag_to_rpm_version(include_dash=True) }}" state: present - when: openshift.common.use_openshift_sdn and not openshift.common.is_containerized | bool + when: + - openshift.common.use_openshift_sdn | default(true) | bool + - not openshift.common.is_containerized | bool - name: Install conntrack-tools package package: @@ -118,8 +92,14 @@ name: openvswitch.service enabled: yes state: started - when: openshift.common.is_containerized | bool and openshift.common.use_openshift_sdn | bool + daemon_reload: yes + when: + - openshift.common.is_containerized | bool + - openshift.common.use_openshift_sdn | default(true) | bool register: ovs_start_result + until: not ovs_start_result | failed + retries: 3 + delay: 30 - set_fact: ovs_service_status_changed: "{{ ovs_start_result | changed }}" @@ -212,15 +192,27 @@ state: started when: openshift.common.is_containerized | bool + - name: Start and enable node systemd: name: "{{ openshift.common.service_type }}-node" enabled: yes state: started + daemon_reload: yes register: node_start_result until: not node_start_result | failed retries: 1 delay: 30 + ignore_errors: true + +- name: Dump logs from node service if it failed + command: journalctl --no-pager -n 100 -u {{ openshift.common.service_type }}-node + when: node_start_result | failed + +- name: Abort if node failed to start + fail: + msg: Node failed to start please inspect the logs and try again + when: node_start_result | failed - set_fact: node_service_status_changed: "{{ node_start_result | changed }}" diff --git a/roles/openshift_node/tasks/openvswitch_system_container.yml b/roles/openshift_node/tasks/openvswitch_system_container.yml index 8cfa5a026..c8d653880 100644 --- a/roles/openshift_node/tasks/openvswitch_system_container.yml +++ b/roles/openshift_node/tasks/openvswitch_system_container.yml @@ -10,3 +10,5 @@ name: openvswitch image: "{{ openshift.common.system_images_registry }}/{{ openshift.node.ovs_system_image }}:{{ openshift_image_tag }}" state: latest + values: + - "DOCKER_SERVICE={{ openshift.docker.service_name }}.service" diff --git a/roles/openshift_node/tasks/storage_plugins/glusterfs.yml b/roles/openshift_node/tasks/storage_plugins/glusterfs.yml index 7d8c42ee2..1b8a7ad50 100644 --- a/roles/openshift_node/tasks/storage_plugins/glusterfs.yml +++ b/roles/openshift_node/tasks/storage_plugins/glusterfs.yml @@ -3,30 +3,52 @@ package: name=glusterfs-fuse state=present when: not openshift.common.is_atomic | bool -- name: Check for existence of virt_use_fusefs seboolean - command: getsebool virt_use_fusefs - register: virt_use_fusefs_output - when: ansible_selinux and ansible_selinux.status == "enabled" +- name: Check for existence of fusefs sebooleans + command: getsebool {{ item }} + register: fusefs_getsebool_status + when: + - ansible_selinux + - ansible_selinux.status == "enabled" failed_when: false changed_when: false + with_items: + - virt_use_fusefs + - virt_sandbox_use_fusefs - name: Set seboolean to allow gluster storage plugin access from containers seboolean: - name: virt_use_fusefs + name: "{{ item.item }}" state: yes persistent: yes - when: ansible_selinux and ansible_selinux.status == "enabled" and virt_use_fusefs_output.rc == 0 + when: + - ansible_selinux + - ansible_selinux.status == "enabled" + - item.rc == 0 + # We need to detect whether or not the boolean is an alias, since `seboolean` + # will error if it is an alias. We do this by inspecting stdout for the boolean name, + # since getsebool prints the resolved name. (At some point Ansible's seboolean module + # should learn to deal with aliases) + - item.item in item.stdout # Boolean does not have an alias. + - ansible_python_version | version_compare('3', '<') + with_items: "{{ fusefs_getsebool_status.results }}" -- name: Check for existence of virt_sandbox_use_fusefs seboolean - command: getsebool virt_sandbox_use_fusefs - register: virt_sandbox_use_fusefs_output - when: ansible_selinux and ansible_selinux.status == "enabled" - failed_when: false - changed_when: false - -- name: Set seboolean to allow gluster storage plugin access from containers(sandbox) - seboolean: - name: virt_sandbox_use_fusefs - state: yes - persistent: yes - when: ansible_selinux and ansible_selinux.status == "enabled" and virt_sandbox_use_fusefs_output.rc == 0 +# Workaround for https://github.com/openshift/openshift-ansible/issues/4438 +# Use command module rather than seboolean module to set GlusterFS booleans. +# TODO: Remove this task and the ansible_python_version comparison in +# the previous task when the problem has been addressed in current +# ansible release. +- name: Set seboolean to allow gluster storage plugin access from containers (python 3) + command: > + setsebool -P {{ item.item }} on + when: + - ansible_selinux + - ansible_selinux.status == "enabled" + - item.rc == 0 + # We need to detect whether or not the boolean is an alias, since `seboolean` + # will error if it is an alias. We do this by inspecting stdout for the boolean name, + # since getsebool prints the resolved name. (At some point Ansible's seboolean module + # should learn to deal with aliases) + - item.item in item.stdout # Boolean does not have an alias. + - ('--> off' in item.stdout) # Boolean is currently off. + - ansible_python_version | version_compare('3', '>=') + with_items: "{{ fusefs_getsebool_status.results }}" diff --git a/roles/openshift_node/tasks/storage_plugins/nfs.yml b/roles/openshift_node/tasks/storage_plugins/nfs.yml index d40ae66cb..7e1035893 100644 --- a/roles/openshift_node/tasks/storage_plugins/nfs.yml +++ b/roles/openshift_node/tasks/storage_plugins/nfs.yml @@ -3,24 +3,52 @@ package: name=nfs-utils state=present when: not openshift.common.is_atomic | bool -- name: Check for existence of seboolean +- name: Check for existence of nfs sebooleans command: getsebool {{ item }} - register: getsebool_status - when: ansible_selinux and ansible_selinux.status == "enabled" + register: nfs_getsebool_status + when: + - ansible_selinux + - ansible_selinux.status == "enabled" failed_when: false changed_when: false with_items: - - virt_use_nfs - - virt_sandbox_use_nfs + - virt_use_nfs + - virt_sandbox_use_nfs - name: Set seboolean to allow nfs storage plugin access from containers seboolean: name: "{{ item.item }}" state: yes persistent: yes + when: + - ansible_selinux + - ansible_selinux.status == "enabled" + - item.rc == 0 # We need to detect whether or not the boolean is an alias, since `seboolean` # will error if it is an alias. We do this by inspecting stdout for the boolean name, # since getsebool prints the resolved name. (At some point Ansible's seboolean module # should learn to deal with aliases) - when: ansible_selinux and ansible_selinux.status == "enabled" and item.rc == 0 and item.stdout.find(item.item) != -1 - with_items: "{{ getsebool_status.results }}" + - item.item in item.stdout # Boolean does not have an alias. + - ansible_python_version | version_compare('3', '<') + with_items: "{{ nfs_getsebool_status.results }}" + +# Workaround for https://github.com/openshift/openshift-ansible/issues/4438 +# Use command module rather than seboolean module to set NFS booleans. +# TODO: Remove this task and the ansible_python_version comparison in +# the previous task when the problem has been addressed in current +# ansible release. +- name: Set seboolean to allow nfs storage plugin access from containers (python 3) + command: > + setsebool -P {{ item.item }} on + when: + - ansible_selinux + - ansible_selinux.status == "enabled" + - item.rc == 0 + # We need to detect whether or not the boolean is an alias, since `seboolean` + # will error if it is an alias. We do this by inspecting stdout for the boolean name, + # since getsebool prints the resolved name. (At some point Ansible's seboolean module + # should learn to deal with aliases) + - item.item in item.stdout # Boolean does not have an alias. + - ('--> off' in item.stdout) # Boolean is currently off. + - ansible_python_version | version_compare('3', '>=') + with_items: "{{ nfs_getsebool_status.results }}" diff --git a/roles/openshift_node/tasks/systemd_units.yml b/roles/openshift_node/tasks/systemd_units.yml index 2ccc28461..b86bb1549 100644 --- a/roles/openshift_node/tasks/systemd_units.yml +++ b/roles/openshift_node/tasks/systemd_units.yml @@ -2,15 +2,8 @@ # This file is included both in the openshift_master role and in the upgrade # playbooks. -- name: Install Node dependencies docker service file - template: - dest: "/etc/systemd/system/{{ openshift.common.service_type }}-node-dep.service" - src: openshift.docker.node.dep.service - register: install_node_dep_result +- include: config/install-node-deps-docker-service-file.yml when: openshift.common.is_containerized | bool - notify: - - reload systemd units - - restart node - block: - name: Pre-pull node image @@ -19,14 +12,7 @@ register: pull_result changed_when: "'Downloaded newer image' in pull_result.stdout" - - name: Install Node docker service file - template: - dest: "/etc/systemd/system/{{ openshift.common.service_type }}-node.service" - src: openshift.docker.node.service - register: install_node_result - notify: - - reload systemd units - - restart node + - include: config/install-node-docker-service-file.yml when: - openshift.common.is_containerized | bool - not openshift.common.is_node_system_container | bool @@ -35,21 +21,13 @@ template: dest: "/etc/systemd/system/{{ openshift.common.service_type }}-node.service" src: "node.service.j2" - register: install_node_result when: not openshift.common.is_containerized | bool notify: - reload systemd units - restart node -- name: Create the openvswitch service env file - template: - src: openvswitch.sysconfig.j2 - dest: /etc/sysconfig/openvswitch +- include: config/install-ovs-service-env-file.yml when: openshift.common.is_containerized | bool - register: install_ovs_sysconfig - notify: - - reload systemd units - - restart openvswitch - name: Install Node system container include: node_system_container.yml @@ -64,22 +42,9 @@ - openshift.common.is_containerized | bool - openshift.common.is_openvswitch_system_container | bool -# May be a temporary workaround. -# https://bugzilla.redhat.com/show_bug.cgi?id=1331590 -- name: Create OpenvSwitch service.d directory - file: path=/etc/systemd/system/openvswitch.service.d/ state=directory +- include: config/workaround-bz1331590-ovs-oom-fix.yml when: openshift.common.use_openshift_sdn | default(true) | bool -- name: Install OpenvSwitch service OOM fix - template: - dest: "/etc/systemd/system/openvswitch.service.d/01-avoid-oom.conf" - src: openvswitch-avoid-oom.conf - when: openshift.common.use_openshift_sdn | default(true) | bool - register: install_oom_fix_result - notify: - - reload systemd units - - restart openvswitch - - block: - name: Pre-pull openvswitch image command: > @@ -87,47 +52,11 @@ register: pull_result changed_when: "'Downloaded newer image' in pull_result.stdout" - - name: Install OpenvSwitch docker service file - template: - dest: "/etc/systemd/system/openvswitch.service" - src: openvswitch.docker.service - notify: - - reload systemd units - - restart openvswitch + - include: config/install-ovs-docker-service-file.yml when: - openshift.common.is_containerized | bool - openshift.common.use_openshift_sdn | default(true) | bool - not openshift.common.is_openvswitch_system_container | bool -- name: Configure Node settings - lineinfile: - dest: /etc/sysconfig/{{ openshift.common.service_type }}-node - regexp: "{{ item.regex }}" - line: "{{ item.line }}" - create: true - with_items: - - regex: '^OPTIONS=' - line: "OPTIONS=--loglevel={{ openshift.node.debug_level | default(2) }}" - - regex: '^CONFIG_FILE=' - line: "CONFIG_FILE={{ openshift.common.config_base }}/node/node-config.yaml" - - regex: '^IMAGE_VERSION=' - line: "IMAGE_VERSION={{ openshift_image_tag }}" - notify: - - restart node - -- name: Configure Proxy Settings - lineinfile: - dest: /etc/sysconfig/{{ openshift.common.service_type }}-node - regexp: "{{ item.regex }}" - line: "{{ item.line }}" - create: true - with_items: - - regex: '^HTTP_PROXY=' - line: "HTTP_PROXY={{ openshift.common.http_proxy | default('') }}" - - regex: '^HTTPS_PROXY=' - line: "HTTPS_PROXY={{ openshift.common.https_proxy | default('') }}" - - regex: '^NO_PROXY=' - line: "NO_PROXY={{ openshift.common.no_proxy | default([]) }},{{ openshift.common.portal_net }},{{ hostvars[groups.oo_first_master.0].openshift.master.sdn_cluster_network_cidr }}" - when: ('http_proxy' in openshift.common and openshift.common.http_proxy != '') - notify: - - restart node +- include: config/configure-node-settings.yml +- include: config/configure-proxy-settings.yml diff --git a/roles/openshift_node/templates/node.service.j2 b/roles/openshift_node/templates/node.service.j2 index d4f0b7762..e12a52c15 100644 --- a/roles/openshift_node/templates/node.service.j2 +++ b/roles/openshift_node/templates/node.service.j2 @@ -24,6 +24,7 @@ WorkingDirectory=/var/lib/origin/ SyslogIdentifier={{ openshift.common.service_type }}-node Restart=always RestartSec=5s +TimeoutStartSec=300 OOMScoreAdjust=-999 [Install] |