From 4fcbe72aeaf359d56ac636a8722f817ec10c629f Mon Sep 17 00:00:00 2001 From: Jan Chaloupka Date: Fri, 12 May 2017 16:07:34 +0200 Subject: move etcd upgrade related code into etcd_upgrade role --- .../openshift-cluster/redeploy-certificates/ca.yml | 9 ++- .../redeploy-certificates/etcd.yml | 8 +- .../openshift-cluster/upgrades/etcd/backup.yml | 86 ++------------------- .../upgrades/etcd/containerized_tasks.yml | 46 ------------ .../upgrades/etcd/fedora_tasks.yml | 23 ------ .../openshift-cluster/upgrades/etcd/main.yml | 2 +- .../openshift-cluster/upgrades/etcd/rhel_tasks.yml | 20 ----- .../openshift-cluster/upgrades/etcd/upgrade.yml | 87 ++++++++-------------- .../upgrades/etcd/upgrade_image_members.yml | 17 +++++ .../upgrades/etcd/upgrade_rpm_members.yml | 18 +++++ playbooks/common/openshift-etcd/config.yml | 1 + roles/etcd/defaults/main.yaml | 6 -- roles/etcd_common/defaults/main.yml | 11 ++- roles/etcd_common/vars/main.yml | 4 + roles/etcd_upgrade/defaults/main.yml | 9 +++ roles/etcd_upgrade/meta/main.yml | 16 ++++ roles/etcd_upgrade/tasks/backup.yml | 71 ++++++++++++++++++ roles/etcd_upgrade/tasks/main.yml | 14 ++++ roles/etcd_upgrade/tasks/upgrade.yml | 11 +++ roles/etcd_upgrade/tasks/upgrade_image.yml | 48 ++++++++++++ roles/etcd_upgrade/tasks/upgrade_rpm.yml | 32 ++++++++ roles/etcd_upgrade/vars/main.yml | 3 + roles/openshift_facts/tasks/main.yml | 4 + 23 files changed, 309 insertions(+), 237 deletions(-) delete mode 100644 playbooks/common/openshift-cluster/upgrades/etcd/containerized_tasks.yml delete mode 100644 playbooks/common/openshift-cluster/upgrades/etcd/fedora_tasks.yml delete mode 100644 playbooks/common/openshift-cluster/upgrades/etcd/rhel_tasks.yml create mode 100644 playbooks/common/openshift-cluster/upgrades/etcd/upgrade_image_members.yml create mode 100644 playbooks/common/openshift-cluster/upgrades/etcd/upgrade_rpm_members.yml create mode 100644 roles/etcd_common/vars/main.yml create mode 100644 roles/etcd_upgrade/defaults/main.yml create mode 100644 roles/etcd_upgrade/meta/main.yml create mode 100644 roles/etcd_upgrade/tasks/backup.yml create mode 100644 roles/etcd_upgrade/tasks/main.yml create mode 100644 roles/etcd_upgrade/tasks/upgrade.yml create mode 100644 roles/etcd_upgrade/tasks/upgrade_image.yml create mode 100644 roles/etcd_upgrade/tasks/upgrade_rpm.yml create mode 100644 roles/etcd_upgrade/vars/main.yml diff --git a/playbooks/common/openshift-cluster/redeploy-certificates/ca.yml b/playbooks/common/openshift-cluster/redeploy-certificates/ca.yml index 4fa7f9cdf..0d0ff798c 100644 --- a/playbooks/common/openshift-cluster/redeploy-certificates/ca.yml +++ b/playbooks/common/openshift-cluster/redeploy-certificates/ca.yml @@ -9,7 +9,8 @@ - name: Backup existing etcd CA certificate directories hosts: oo_etcd_to_config roles: - - etcd_common + - role: etcd_common + r_etcd_common_etcd_runtime: "{{ openshift.common.etcd_runtime }}" tasks: - name: Determine if CA certificate directory exists stat: @@ -52,7 +53,8 @@ vars: etcd_ca_host: "{{ groups.oo_etcd_to_config.0 }}" roles: - - etcd_common + - role: etcd_common + r_etcd_common_etcd_runtime: "{{ openshift.common.etcd_runtime }}" tasks: - name: Create a tarball of the etcd ca certs command: > @@ -98,7 +100,8 @@ - name: Retrieve etcd CA certificate hosts: oo_first_etcd roles: - - etcd_common + - role: etcd_common + r_etcd_common_etcd_runtime: "{{ openshift.common.etcd_runtime }}" tasks: - name: Retrieve etcd CA certificate fetch: diff --git a/playbooks/common/openshift-cluster/redeploy-certificates/etcd.yml b/playbooks/common/openshift-cluster/redeploy-certificates/etcd.yml index 2963a5940..6b5c805e6 100644 --- a/playbooks/common/openshift-cluster/redeploy-certificates/etcd.yml +++ b/playbooks/common/openshift-cluster/redeploy-certificates/etcd.yml @@ -3,7 +3,8 @@ hosts: oo_first_etcd any_errors_fatal: true roles: - - etcd_common + - role: etcd_common + r_etcd_common_etcd_runtime: "{{ openshift.common.etcd_runtime }}" post_tasks: - name: Determine if generated etcd certificates exist stat: @@ -27,7 +28,8 @@ hosts: oo_etcd_to_config any_errors_fatal: true roles: - - etcd_common + - role: etcd_common + r_etcd_common_etcd_runtime: "{{ openshift.common.etcd_runtime }}" post_tasks: - name: Backup etcd certificates command: > @@ -50,6 +52,7 @@ etcd_peers: "{{ groups.oo_etcd_to_config | default([], true) }}" etcd_certificates_etcd_hosts: "{{ groups.oo_etcd_to_config | default([], true) }}" openshift_ca_host: "{{ groups.oo_first_master.0 }}" + r_etcd_common_etcd_runtime: "{{ openshift.common.etcd_runtime }}" - name: Redeploy etcd client certificates for masters hosts: oo_masters_to_config @@ -63,4 +66,5 @@ etcd_cert_prefix: "master.etcd-" openshift_ca_host: "{{ groups.oo_first_master.0 }}" openshift_master_count: "{{ openshift.master.master_count | default(groups.oo_masters | length) }}" + r_etcd_common_etcd_runtime: "{{ openshift.common.etcd_runtime }}" when: groups.oo_etcd_to_config is defined and groups.oo_etcd_to_config diff --git a/playbooks/common/openshift-cluster/upgrades/etcd/backup.yml b/playbooks/common/openshift-cluster/upgrades/etcd/backup.yml index 9d0333ca8..b7fd2c0c5 100644 --- a/playbooks/common/openshift-cluster/upgrades/etcd/backup.yml +++ b/playbooks/common/openshift-cluster/upgrades/etcd/backup.yml @@ -1,84 +1,14 @@ --- - name: Backup etcd hosts: oo_etcd_hosts_to_backup - vars: - embedded_etcd: "{{ groups.oo_etcd_to_config | default([]) | length == 0 }}" - etcdctl_command: "{{ 'etcdctl' if not openshift.common.is_containerized or embedded_etcd else 'docker exec etcd_container etcdctl' if not openshift.common.is_etcd_system_container else 'runc exec etcd etcdctl' }}" - timestamp: "{{ lookup('pipe', 'date +%Y%m%d%H%M%S') }}" roles: - - openshift_facts - tasks: - # Ensure we persist the etcd role for this host in openshift_facts - - openshift_facts: - role: etcd - local_facts: {} - when: "'etcd' not in openshift" - - set_fact: - etcd_backup_dir: "{{ openshift.etcd.etcd_data_dir }}/openshift-backup-{{ backup_tag | default('') }}{{ timestamp }}" - - # TODO: replace shell module with command and update later checks - - name: Check available disk space for etcd backup - shell: df --output=avail -k {{ openshift.etcd.etcd_data_dir }} | tail -n 1 - register: avail_disk - # AUDIT:changed_when: `false` because we are only inspecting - # state, not manipulating anything - changed_when: false - - # TODO: replace shell module with command and update later checks - - name: Check current etcd disk usage - shell: du --exclude='*openshift-backup*' -k {{ openshift.etcd.etcd_data_dir }} | tail -n 1 | cut -f1 - register: etcd_disk_usage - when: embedded_etcd | bool - # AUDIT:changed_when: `false` because we are only inspecting - # state, not manipulating anything - changed_when: false - - - name: Abort if insufficient disk space for etcd backup - fail: - msg: > - {{ etcd_disk_usage.stdout }} Kb disk space required for etcd backup, - {{ avail_disk.stdout }} Kb available. - when: (embedded_etcd | bool) and (etcd_disk_usage.stdout|int > avail_disk.stdout|int) - - # For non containerized and non embedded we should have the correct version of - # etcd installed already. So don't do anything. - # - # For containerized installs we now exec into etcd_container - # - # For embedded non containerized we need to ensure we have the latest version - # etcd on the host. - - name: Install latest etcd for embedded - package: - name: etcd - state: latest - when: - - embedded_etcd | bool - - not openshift.common.is_atomic | bool - - - name: Generate etcd backup - command: > - {{ etcdctl_command }} backup --data-dir={{ openshift.etcd.etcd_data_dir }} - --backup-dir={{ etcd_backup_dir }} - - # According to the docs change you can simply copy snap/db - # https://github.com/openshift/openshift-docs/commit/b38042de02d9780842dce95cfa0ef45d53b58bc6 - - name: Check for v3 data store - stat: - path: "{{ openshift.etcd.etcd_data_dir }}/member/snap/db" - register: v3_db - - - name: Copy etcd v3 data store - command: > - cp -a {{ openshift.etcd.etcd_data_dir }}/member/snap/db - {{ etcd_backup_dir }}/member/snap/ - when: v3_db.stat.exists - - - set_fact: - etcd_backup_complete: True - - - name: Display location of etcd backup - debug: - msg: "Etcd backup created in {{ etcd_backup_dir }}" + - role: openshift_facts + - role: etcd_upgrade + r_etcd_upgrade_action: backup + r_etcd_backup_tag: etcd_backup_tag + r_etcd_common_etcd_runtime: "{{ openshift.common.etcd_runtime }}" + r_etcd_upgrade_embedded_etcd: "{{ groups.oo_etcd_to_config | default([]) | length == 0 }}" + r_etcd_backup_sufix_name: "{{ lookup('pipe', 'date +%Y%m%d%H%M%S') }}" - name: Gate on etcd backup hosts: localhost @@ -88,7 +18,7 @@ - set_fact: etcd_backup_completed: "{{ hostvars | oo_select_keys(groups.oo_etcd_hosts_to_backup) - | oo_collect('inventory_hostname', {'etcd_backup_complete': true}) }}" + | oo_collect('inventory_hostname', {'r_etcd_upgrade_backup_complete': true}) }}" - set_fact: etcd_backup_failed: "{{ groups.oo_etcd_hosts_to_backup | difference(etcd_backup_completed) }}" - fail: diff --git a/playbooks/common/openshift-cluster/upgrades/etcd/containerized_tasks.yml b/playbooks/common/openshift-cluster/upgrades/etcd/containerized_tasks.yml deleted file mode 100644 index 5f8b59e17..000000000 --- a/playbooks/common/openshift-cluster/upgrades/etcd/containerized_tasks.yml +++ /dev/null @@ -1,46 +0,0 @@ ---- -- name: Verify cluster is healthy pre-upgrade - command: "etcdctl --cert-file /etc/etcd/peer.crt --key-file /etc/etcd/peer.key --ca-file /etc/etcd/ca.crt -C https://{{ openshift.common.hostname }}:2379 cluster-health" - -- name: Get current image - shell: grep 'ExecStart=' /etc/systemd/system/etcd_container.service | awk '{print $NF}' - register: current_image - -- name: Set new_etcd_image - set_fact: - new_etcd_image: "{{ current_image.stdout | regex_replace('/etcd.*$','/etcd:' ~ upgrade_version ) }}" - -- name: Pull new etcd image - command: "docker pull {{ new_etcd_image }}" - -- name: Update to latest etcd image - replace: - dest: /etc/systemd/system/etcd_container.service - regexp: "{{ current_image.stdout }}$" - replace: "{{ new_etcd_image }}" - -- name: Restart etcd_container - systemd: - name: etcd_container - daemon_reload: yes - state: restarted - -## TODO: probably should just move this into the backup playbooks, also this -## will fail on atomic host. We need to revisit how to do etcd backups there as -## the container may be newer than etcdctl on the host. Assumes etcd3 obsoletes etcd (7.3.1) -- name: Upgrade etcd for etcdctl when not atomic - package: name=etcd state=latest - when: not openshift.common.is_atomic | bool - -- name: Verify cluster is healthy - command: "etcdctl --cert-file /etc/etcd/peer.crt --key-file /etc/etcd/peer.key --ca-file /etc/etcd/ca.crt -C https://{{ openshift.common.hostname }}:2379 cluster-health" - register: etcdctl - until: etcdctl.rc == 0 - retries: 3 - delay: 10 - -- name: Store new etcd_image - openshift_facts: - role: etcd - local_facts: - etcd_image: "{{ new_etcd_image }}" diff --git a/playbooks/common/openshift-cluster/upgrades/etcd/fedora_tasks.yml b/playbooks/common/openshift-cluster/upgrades/etcd/fedora_tasks.yml deleted file mode 100644 index 30232110e..000000000 --- a/playbooks/common/openshift-cluster/upgrades/etcd/fedora_tasks.yml +++ /dev/null @@ -1,23 +0,0 @@ ---- -# F23 GA'd with etcd 2.0, currently has 2.2 in updates -# F24 GA'd with etcd-2.2, currently has 2.2 in updates -# F25 Beta currently has etcd 3.0 -- name: Verify cluster is healthy pre-upgrade - command: "etcdctl --cert-file /etc/etcd/peer.crt --key-file /etc/etcd/peer.key --ca-file /etc/etcd/ca.crt -C https://{{ openshift.common.hostname }}:2379 cluster-health" - -- name: Update etcd - package: - name: "etcd" - state: "latest" - -- name: Restart etcd - service: - name: etcd - state: restarted - -- name: Verify cluster is healthy - command: "etcdctl --cert-file /etc/etcd/peer.crt --key-file /etc/etcd/peer.key --ca-file /etc/etcd/ca.crt -C https://{{ openshift.common.hostname }}:2379 cluster-health" - register: etcdctl - until: etcdctl.rc == 0 - retries: 3 - delay: 10 diff --git a/playbooks/common/openshift-cluster/upgrades/etcd/main.yml b/playbooks/common/openshift-cluster/upgrades/etcd/main.yml index d9b59edcb..3e01883ae 100644 --- a/playbooks/common/openshift-cluster/upgrades/etcd/main.yml +++ b/playbooks/common/openshift-cluster/upgrades/etcd/main.yml @@ -8,7 +8,7 @@ - name: Backup etcd before upgrading anything include: backup.yml vars: - backup_tag: "pre-upgrade-" + etcd_backup_tag: "pre-upgrade-" when: openshift_etcd_backup | default(true) | bool - name: Drop etcdctl profiles diff --git a/playbooks/common/openshift-cluster/upgrades/etcd/rhel_tasks.yml b/playbooks/common/openshift-cluster/upgrades/etcd/rhel_tasks.yml deleted file mode 100644 index 3a972e8ab..000000000 --- a/playbooks/common/openshift-cluster/upgrades/etcd/rhel_tasks.yml +++ /dev/null @@ -1,20 +0,0 @@ ---- -- name: Verify cluster is healthy pre-upgrade - command: "etcdctl --cert-file /etc/etcd/peer.crt --key-file /etc/etcd/peer.key --ca-file /etc/etcd/ca.crt -C https://{{ openshift.common.hostname }}:2379 cluster-health" - -- name: Update etcd RPM - package: - name: etcd-{{ upgrade_version }}* - state: latest - -- name: Restart etcd - service: - name: etcd - state: restarted - -- name: Verify cluster is healthy - command: "etcdctl --cert-file /etc/etcd/peer.crt --key-file /etc/etcd/peer.key --ca-file /etc/etcd/ca.crt -C https://{{ openshift.common.hostname }}:2379 cluster-health" - register: etcdctl - until: etcdctl.rc == 0 - retries: 3 - delay: 10 diff --git a/playbooks/common/openshift-cluster/upgrades/etcd/upgrade.yml b/playbooks/common/openshift-cluster/upgrades/etcd/upgrade.yml index 54f9e21a1..0431c1ce0 100644 --- a/playbooks/common/openshift-cluster/upgrades/etcd/upgrade.yml +++ b/playbooks/common/openshift-cluster/upgrades/etcd/upgrade.yml @@ -12,10 +12,10 @@ # AUDIT:changed_when: `false` because we are only inspecting # state, not manipulating anything changed_when: false - - debug: msg: "Etcd rpm version {{ etcd_rpm_version.stdout }} detected" - when: not openshift.common.is_containerized | bool + when: + - not openshift.common.is_containerized | bool - block: - name: Record containerized etcd version (docker) @@ -54,84 +54,57 @@ - debug: msg: "Etcd containerized version {{ etcd_container_version }} detected" - when: - openshift.common.is_containerized | bool -# I really dislike this copy/pasta but I wasn't able to find a way to get it to loop -# through hosts, then loop through tasks only when appropriate -- name: Upgrade to 2.1 - hosts: oo_etcd_hosts_to_upgrade - serial: 1 +- include: upgrade_rpm_members.yml vars: - upgrade_version: '2.1' - tasks: - - include: rhel_tasks.yml - when: etcd_rpm_version.stdout | default('99') | version_compare('2.1','<') and ansible_distribution == 'RedHat' and not openshift.common.is_containerized | bool + etcd_upgrade_version: '2.1' -- name: Upgrade RPM hosts to 2.2 - hosts: oo_etcd_hosts_to_upgrade - serial: 1 +- include: upgrade_rpm_members.yml vars: - upgrade_version: '2.2' - tasks: - - include: rhel_tasks.yml - when: etcd_rpm_version.stdout | default('99') | version_compare('2.2','<') and ansible_distribution == 'RedHat' and not openshift.common.is_containerized | bool + etcd_upgrade_version: '2.2' -- name: Upgrade containerized hosts to 2.2.5 - hosts: oo_etcd_hosts_to_upgrade - serial: 1 +- include: upgrade_image_members.yml vars: - upgrade_version: 2.2.5 - tasks: - - include: containerized_tasks.yml - when: etcd_container_version | default('99') | version_compare('2.2','<') and openshift.common.is_containerized | bool + etcd_upgrade_version: '2.2.5' -- name: Upgrade RPM hosts to 2.3 - hosts: oo_etcd_hosts_to_upgrade - serial: 1 +- include: upgrade_rpm_members.yml vars: - upgrade_version: '2.3' - tasks: - - include: rhel_tasks.yml - when: etcd_rpm_version.stdout | default('99') | version_compare('2.3','<') and ansible_distribution == 'RedHat' and not openshift.common.is_containerized | bool + etcd_upgrade_version: '2.3' -- name: Upgrade containerized hosts to 2.3.7 - hosts: oo_etcd_hosts_to_upgrade - serial: 1 +- include: upgrade_image_members.yml vars: - upgrade_version: 2.3.7 - tasks: - - include: containerized_tasks.yml - when: etcd_container_version | default('99') | version_compare('2.3','<') and openshift.common.is_containerized | bool + etcd_upgrade_version: '2.3.7' -- name: Upgrade RPM hosts to 3.0 - hosts: oo_etcd_hosts_to_upgrade - serial: 1 +- include: upgrade_rpm_members.yml vars: - upgrade_version: '3.0' - tasks: - - include: rhel_tasks.yml - when: etcd_rpm_version.stdout | default('99') | version_compare('3.0','<') and ansible_distribution == 'RedHat' and not openshift.common.is_containerized | bool + etcd_upgrade_version: '3.0' -- name: Upgrade containerized hosts to etcd3 image - hosts: oo_etcd_hosts_to_upgrade - serial: 1 +- include: upgrade_image_members.yml vars: - upgrade_version: 3.0.15 - tasks: - - include: containerized_tasks.yml - when: etcd_container_version | default('99') | version_compare('3.0','<') and openshift.common.is_containerized | bool + etcd_upgrade_version: '3.0.15' + +- include: upgrade_rpm_members.yml + vars: + etcd_upgrade_version: '3.1' + +- include: upgrade_image_members.yml + vars: + etcd_upgrade_version: '3.1.3' - name: Upgrade fedora to latest hosts: oo_etcd_hosts_to_upgrade serial: 1 tasks: - - include: fedora_tasks.yml - when: ansible_distribution == 'Fedora' and not openshift.common.is_containerized | bool + - include_role: + name: etcd_upgrade + when: + - ansible_distribution == 'Fedora' + - not openshift.common.is_containerized | bool - name: Backup etcd include: backup.yml vars: - backup_tag: "post-3.0-" + etcd_backup_tag: "post-3.0-" when: openshift_etcd_backup | default(true) | bool diff --git a/playbooks/common/openshift-cluster/upgrades/etcd/upgrade_image_members.yml b/playbooks/common/openshift-cluster/upgrades/etcd/upgrade_image_members.yml new file mode 100644 index 000000000..831ca8f57 --- /dev/null +++ b/playbooks/common/openshift-cluster/upgrades/etcd/upgrade_image_members.yml @@ -0,0 +1,17 @@ +--- +# INPUT etcd_upgrade_version +# INPUT etcd_container_version +# INPUT openshift.common.is_containerized +- name: Upgrade containerized hosts to {{ etcd_upgrade_version }} + hosts: oo_etcd_hosts_to_upgrade + serial: 1 + roles: + - role: etcd_upgrade + r_etcd_upgrade_action: upgrade + r_etcd_upgrade_mechanism: image + r_etcd_upgrade_version: "{{ etcd_upgrade_version }}" + r_etcd_common_etcd_runtime: "{{ openshift.common.etcd_runtime }}" + etcd_peer: "{{ openshift.common.hostname }}" + when: + - etcd_container_version | default('99') | version_compare(etcd_upgrade_version,'<') + - openshift.common.is_containerized | bool diff --git a/playbooks/common/openshift-cluster/upgrades/etcd/upgrade_rpm_members.yml b/playbooks/common/openshift-cluster/upgrades/etcd/upgrade_rpm_members.yml new file mode 100644 index 000000000..2e79451e0 --- /dev/null +++ b/playbooks/common/openshift-cluster/upgrades/etcd/upgrade_rpm_members.yml @@ -0,0 +1,18 @@ +--- +# INPUT etcd_upgrade_version +# INPUT etcd_rpm_version +# INPUT openshift.common.is_containerized +- name: Upgrade to {{ etcd_upgrade_version }} + hosts: oo_etcd_hosts_to_upgrade + serial: 1 + roles: + - role: etcd_upgrade + r_etcd_upgrade_action: upgrade + r_etcd_upgrade_mechanism: rpm + r_etcd_upgrade_version: "{{ etcd_upgrade_version }}" + r_etcd_common_etcd_runtime: "host" + etcd_peer: "{{ openshift.common.hostname }}" + when: + - etcd_rpm_version.stdout | default('99') | version_compare(etcd_upgrade_version, '<') + - ansible_distribution == 'RedHat' + - not openshift.common.is_containerized | bool diff --git a/playbooks/common/openshift-etcd/config.yml b/playbooks/common/openshift-etcd/config.yml index 1b8106e0e..2cb6197d1 100644 --- a/playbooks/common/openshift-etcd/config.yml +++ b/playbooks/common/openshift-etcd/config.yml @@ -7,4 +7,5 @@ etcd_peers: "{{ groups.oo_etcd_to_config | default([], true) }}" etcd_ca_host: "{{ groups.oo_etcd_to_config.0 }}" etcd_certificates_etcd_hosts: "{{ groups.oo_etcd_to_config | default([], true) }}" + r_etcd_common_etcd_runtime: "{{ openshift.common.etcd_runtime }}" - role: nickhammond.logrotate diff --git a/roles/etcd/defaults/main.yaml b/roles/etcd/defaults/main.yaml index e45f53219..c0d1d5946 100644 --- a/roles/etcd/defaults/main.yaml +++ b/roles/etcd/defaults/main.yaml @@ -1,10 +1,4 @@ --- -etcd_service: "{{ 'etcd' if openshift.common.is_etcd_system_container | bool or not etcd_is_containerized | bool else 'etcd_container' }}" -etcd_client_port: 2379 -etcd_peer_port: 2380 -etcd_url_scheme: http -etcd_peer_url_scheme: http - etcd_initial_cluster_state: new etcd_initial_cluster_token: etcd-cluster-1 diff --git a/roles/etcd_common/defaults/main.yml b/roles/etcd_common/defaults/main.yml index d12e6a07f..e1a080b34 100644 --- a/roles/etcd_common/defaults/main.yml +++ b/roles/etcd_common/defaults/main.yml @@ -1,6 +1,9 @@ --- +# runc, docker, host +r_etcd_common_etcd_runtime: "docker" + # etcd server vars -etcd_conf_dir: "{{ '/etc/etcd' if not openshift.common.is_etcd_system_container else '/var/lib/etcd/etcd.etcd/etc' }}" +etcd_conf_dir: "{{ '/etc/etcd' if r_etcd_common_etcd_runtime != 'runc' else '/var/lib/etcd/etcd.etcd/etc' }}" etcd_system_container_conf_dir: /var/lib/etcd/etc etcd_conf_file: "{{ etcd_conf_dir }}/etcd.conf" etcd_ca_file: "{{ etcd_conf_dir }}/ca.crt" @@ -38,3 +41,9 @@ etcd_is_thirdparty: False # etcd dir vars etcd_data_dir: /var/lib/etcd/ + +# etcd ports and protocols +etcd_client_port: 2379 +etcd_peer_port: 2380 +etcd_url_scheme: http +etcd_peer_url_scheme: http diff --git a/roles/etcd_common/vars/main.yml b/roles/etcd_common/vars/main.yml new file mode 100644 index 000000000..00d697776 --- /dev/null +++ b/roles/etcd_common/vars/main.yml @@ -0,0 +1,4 @@ +--- +etcd_service: "{{ 'etcd_container' if r_etcd_common_etcd_runtime == 'docker' else 'etcd' }}" +# Location of the service file is fixed and not meant to be changed +etcd_service_file: "/etc/systemd/system/{{ etcd_service }}.service" diff --git a/roles/etcd_upgrade/defaults/main.yml b/roles/etcd_upgrade/defaults/main.yml new file mode 100644 index 000000000..01ad8a268 --- /dev/null +++ b/roles/etcd_upgrade/defaults/main.yml @@ -0,0 +1,9 @@ +--- +r_etcd_upgrade_action: upgrade +r_etcd_upgrade_mechanism: rpm +r_etcd_upgrade_embedded_etcd: False + +# etcd run on a host => use etcdctl command directly +# etcd run as a docker container => use docker exec +# etcd run as a runc container => use runc exec +etcdctl_command: "{{ 'etcdctl' if r_etcd_common_etcd_runtime == 'host' or r_etcd_upgrade_embedded_etcd | bool else 'docker exec etcd_container etcdctl' if r_etcd_common_etcd_runtime == 'docker' else 'runc exec etcd etcdctl' }}" diff --git a/roles/etcd_upgrade/meta/main.yml b/roles/etcd_upgrade/meta/main.yml new file mode 100644 index 000000000..018bdc8d7 --- /dev/null +++ b/roles/etcd_upgrade/meta/main.yml @@ -0,0 +1,16 @@ +--- +galaxy_info: + author: Jan Chaloupka + description: + company: Red Hat, Inc. + license: Apache License, Version 2.0 + min_ansible_version: 1.9 + platforms: + - name: EL + versions: + - 7 + categories: + - cloud + - system +dependencies: +- role: etcd_common diff --git a/roles/etcd_upgrade/tasks/backup.yml b/roles/etcd_upgrade/tasks/backup.yml new file mode 100644 index 000000000..1ea6fc59f --- /dev/null +++ b/roles/etcd_upgrade/tasks/backup.yml @@ -0,0 +1,71 @@ +--- +# INPUT r_etcd_backup_sufix_name +# INPUT r_etcd_backup_tag +# OUTPUT r_etcd_upgrade_backup_complete +- set_fact: + # ORIGIN etcd_data_dir etcd_common.defaults + l_etcd_backup_dir: "{{ etcd_data_dir }}/openshift-backup-{{ r_etcd_backup_tag | default('') }}{{ r_etcd_backup_sufix_name }}" + +# TODO: replace shell module with command and update later checks +- name: Check available disk space for etcd backup + shell: df --output=avail -k {{ etcd_data_dir }} | tail -n 1 + register: avail_disk + # AUDIT:changed_when: `false` because we are only inspecting + # state, not manipulating anything + changed_when: false + +# TODO: replace shell module with command and update later checks +- name: Check current etcd disk usage + shell: du --exclude='*openshift-backup*' -k {{ etcd_data_dir }} | tail -n 1 | cut -f1 + register: etcd_disk_usage + when: r_etcd_upgrade_embedded_etcd | bool + # AUDIT:changed_when: `false` because we are only inspecting + # state, not manipulating anything + changed_when: false + +- name: Abort if insufficient disk space for etcd backup + fail: + msg: > + {{ etcd_disk_usage.stdout }} Kb disk space required for etcd backup, + {{ avail_disk.stdout }} Kb available. + when: (r_etcd_upgrade_embedded_etcd | bool) and (etcd_disk_usage.stdout|int > avail_disk.stdout|int) + +# For non containerized and non embedded we should have the correct version of +# etcd installed already. So don't do anything. +# +# For containerized installs we now exec into etcd_container +# +# For embedded non containerized we need to ensure we have the latest version +# etcd on the host. +- name: Install latest etcd for embedded + package: + name: etcd + state: latest + when: + - r_etcd_upgrade_embedded_etcd | bool + - not l_ostree_booted.stat.exists | bool + +- name: Generate etcd backup + command: > + {{ etcdctl_command }} backup --data-dir={{ etcd_data_dir }} + --backup-dir={{ l_etcd_backup_dir }} + +# According to the docs change you can simply copy snap/db +# https://github.com/openshift/openshift-docs/commit/b38042de02d9780842dce95cfa0ef45d53b58bc6 +- name: Check for v3 data store + stat: + path: "{{ etcd_data_dir }}/member/snap/db" + register: v3_db + +- name: Copy etcd v3 data store + command: > + cp -a {{ etcd_data_dir }}/member/snap/db + {{ l_etcd_backup_dir }}/member/snap/ + when: v3_db.stat.exists + +- set_fact: + r_etcd_upgrade_backup_complete: True + +- name: Display location of etcd backup + debug: + msg: "Etcd backup created in {{ l_etcd_backup_dir }}" diff --git a/roles/etcd_upgrade/tasks/main.yml b/roles/etcd_upgrade/tasks/main.yml new file mode 100644 index 000000000..5178c14e3 --- /dev/null +++ b/roles/etcd_upgrade/tasks/main.yml @@ -0,0 +1,14 @@ +--- +# INPUT r_etcd_upgrade_action +- name: Fail if invalid etcd_upgrade_action provided + fail: + msg: "etcd_upgrade role can only be called with 'upgrade' or 'backup'" + when: + - r_etcd_upgrade_action not in ['upgrade', 'backup'] + +- name: Detecting Atomic Host Operating System + stat: + path: /run/ostree-booted + register: l_ostree_booted + +- include: "{{ r_etcd_upgrade_action }}.yml" diff --git a/roles/etcd_upgrade/tasks/upgrade.yml b/roles/etcd_upgrade/tasks/upgrade.yml new file mode 100644 index 000000000..420c9638e --- /dev/null +++ b/roles/etcd_upgrade/tasks/upgrade.yml @@ -0,0 +1,11 @@ +--- +# INPUT r_etcd_upgrade_version +# INPUT r_etcd_upgrade_mechanism +- name: Failt if r_etcd_upgrade_mechanism is not set during upgrade + fail: + msg: "r_etcd_upgrade_mechanism can be only set to 'rpm' or 'image'" + when: + - r_etcd_upgrade_mechanism not in ['rpm', 'image'] + +- name: "Upgrade {{ r_etcd_upgrade_mechanism }} based etcd" + include: upgrade_{{ r_etcd_upgrade_mechanism }}.yml diff --git a/roles/etcd_upgrade/tasks/upgrade_image.yml b/roles/etcd_upgrade/tasks/upgrade_image.yml new file mode 100644 index 000000000..136ec1142 --- /dev/null +++ b/roles/etcd_upgrade/tasks/upgrade_image.yml @@ -0,0 +1,48 @@ +--- +# INPUT r_etcd_upgrade_version +- name: Verify cluster is healthy pre-upgrade + command: "{{ etcdctlv2 }} cluster-health" + +- name: Get current image + shell: "grep 'ExecStart=' {{ etcd_service_file }} | awk '{print $NF}'" + register: current_image + +- name: Set new_etcd_image + set_fact: + new_etcd_image: "{{ current_image.stdout | regex_replace('/etcd.*$','/etcd:' ~ r_etcd_upgrade_version ) }}" + +- name: Pull new etcd image + command: "docker pull {{ new_etcd_image }}" + +- name: Update to latest etcd image + replace: + dest: "{{ etcd_service_file }}" + regexp: "{{ current_image.stdout }}$" + replace: "{{ new_etcd_image }}" + +- name: Restart etcd_container + systemd: + name: "{{ etcd_service }}" + daemon_reload: yes + state: restarted + +## TODO: probably should just move this into the backup playbooks, also this +## will fail on atomic host. We need to revisit how to do etcd backups there as +## the container may be newer than etcdctl on the host. Assumes etcd3 obsoletes etcd (7.3.1) +- name: Upgrade etcd for etcdctl when not atomic + package: name=etcd state=latest + when: not l_ostree_booted.stat.exists | bool + +- name: Verify cluster is healthy + command: "{{ etcdctlv2 }} cluster-health" + register: etcdctl + until: etcdctl.rc == 0 + retries: 3 + delay: 10 + +- name: Store new etcd_image + # DEPENDENCY openshift_facts + openshift_facts: + role: etcd + local_facts: + etcd_image: "{{ new_etcd_image }}" diff --git a/roles/etcd_upgrade/tasks/upgrade_rpm.yml b/roles/etcd_upgrade/tasks/upgrade_rpm.yml new file mode 100644 index 000000000..324b69605 --- /dev/null +++ b/roles/etcd_upgrade/tasks/upgrade_rpm.yml @@ -0,0 +1,32 @@ +--- +# INPUT r_etcd_upgrade_version? + +# F23 GA'd with etcd 2.0, currently has 2.2 in updates +# F24 GA'd with etcd-2.2, currently has 2.2 in updates +# F25 Beta currently has etcd 3.0 +# RHEL 7.3.4 with etcd-3.1.3-1.el7 +# RHEL 7.3.3 with etcd-3.1.0-2.el7 +# RHEL 7.3.2 with etcd-3.0.15-1.el7 + +- name: Verify cluster is healthy pre-upgrade + command: "{{ etcdctlv2 }} cluster-health" + +- set_fact: + l_etcd_target_package: "{{ 'etcd' if r_etcd_upgrade_version is not defined else 'etcd-'+r_etcd_upgrade_version+'*' }}" + +- name: Update etcd RPM to {{ l_etcd_target_package }} + package: + name: "{{ l_etcd_target_package }}" + state: latest + +- name: Restart etcd + service: + name: "{{ etcd_service }}" + state: restarted + +- name: Verify cluster is healthy + command: "{{ etcdctlv2 }} cluster-health" + register: etcdctl + until: etcdctl.rc == 0 + retries: 3 + delay: 10 diff --git a/roles/etcd_upgrade/vars/main.yml b/roles/etcd_upgrade/vars/main.yml new file mode 100644 index 000000000..5ed919d42 --- /dev/null +++ b/roles/etcd_upgrade/vars/main.yml @@ -0,0 +1,3 @@ +--- +# EXPECTS etcd_peer +etcdctlv2: "etcdctl --cert-file {{ etcd_peer_cert_file }} --key-file {{ etcd_peer_key_file }} --ca-file {{ etcd_peer_ca_file }} -C https://{{ etcd_peer }}:{{ etcd_client_port }}" diff --git a/roles/openshift_facts/tasks/main.yml b/roles/openshift_facts/tasks/main.yml index f657d86cf..1b9bda67e 100644 --- a/roles/openshift_facts/tasks/main.yml +++ b/roles/openshift_facts/tasks/main.yml @@ -15,6 +15,9 @@ l_is_etcd_system_container: "{{ (use_etcd_system_container | default(use_system_containers) | bool) }}" - set_fact: l_any_system_container: "{{ l_is_etcd_system_container or l_is_openvswitch_system_container or l_is_node_system_container or l_is_master_system_container }}" +- set_fact: + l_etcd_runtime: "{{ 'runc' if l_is_etcd_system_container else 'docker' if l_is_containerized else 'host' }}" + - name: Validate python version fail: @@ -80,6 +83,7 @@ is_node_system_container: "{{ l_is_node_system_container | default(false) }}" is_master_system_container: "{{ l_is_master_system_container | default(false) }}" is_etcd_system_container: "{{ l_is_etcd_system_container | default(false) }}" + etcd_runtime: "{{ l_etcd_runtime }}" system_images_registry: "{{ system_images_registry | default('') }}" public_hostname: "{{ openshift_public_hostname | default(None) }}" public_ip: "{{ openshift_public_ip | default(None) }}" -- cgit v1.2.3