From a5f4561d8cdfd974705105275f718cee9bb249ad Mon Sep 17 00:00:00 2001 From: Scott Dodson Date: Mon, 31 Oct 2016 10:30:18 -0400 Subject: Add etcd upgrade for RHEL and Fedora On Fedora we just blindly upgrade to the latest. On RHEL we do stepwise upgrades 2.0,2.1,2.2,2.3,3.0 --- .../openshift-cluster/upgrades/etcd/backup.yml | 73 ++++++++++++++++++ .../upgrades/etcd/fedora_tasks.yml | 23 ++++++ .../openshift-cluster/upgrades/etcd/filter_plugins | 1 + .../openshift-cluster/upgrades/etcd/lookup_plugins | 1 + .../openshift-cluster/upgrades/etcd/main.yml | 88 ++++++++++++++++++++++ .../openshift-cluster/upgrades/etcd/rhel_tasks.yml | 23 ++++++ .../common/openshift-cluster/upgrades/etcd/roles | 1 + .../openshift-cluster/upgrades/etcd_backup.yml | 73 ------------------ .../upgrades/upgrade_control_plane.yml | 2 +- 9 files changed, 211 insertions(+), 74 deletions(-) create mode 100644 playbooks/common/openshift-cluster/upgrades/etcd/backup.yml create mode 100644 playbooks/common/openshift-cluster/upgrades/etcd/fedora_tasks.yml create mode 120000 playbooks/common/openshift-cluster/upgrades/etcd/filter_plugins create mode 120000 playbooks/common/openshift-cluster/upgrades/etcd/lookup_plugins create mode 100644 playbooks/common/openshift-cluster/upgrades/etcd/main.yml create mode 100644 playbooks/common/openshift-cluster/upgrades/etcd/rhel_tasks.yml create mode 120000 playbooks/common/openshift-cluster/upgrades/etcd/roles delete mode 100644 playbooks/common/openshift-cluster/upgrades/etcd_backup.yml (limited to 'playbooks/common') diff --git a/playbooks/common/openshift-cluster/upgrades/etcd/backup.yml b/playbooks/common/openshift-cluster/upgrades/etcd/backup.yml new file mode 100644 index 000000000..57b156b1c --- /dev/null +++ b/playbooks/common/openshift-cluster/upgrades/etcd/backup.yml @@ -0,0 +1,73 @@ +- name: Backup etcd + hosts: etcd_hosts_to_backup + vars: + embedded_etcd: "{{ hostvars[groups.oo_first_master.0].openshift.master.embedded_etcd }}" + 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" + + - stat: path=/var/lib/openshift + register: var_lib_openshift + + - stat: path=/var/lib/origin + register: var_lib_origin + + - name: Create origin symlink if necessary + file: src=/var/lib/openshift/ dest=/var/lib/origin state=link + when: var_lib_openshift.stat.exists == True and var_lib_origin.stat.exists == False + + # TODO: replace shell module with command and update later checks + # We assume to be using the data dir for all backups. + - name: Check available disk space for etcd backup + shell: df --output=avail -k {{ openshift.common.data_dir }} | tail -n 1 + register: avail_disk + + # TODO: replace shell module with command and update later checks + - name: Check current embedded etcd disk usage + shell: du -k {{ openshift.etcd.etcd_data_dir }} | tail -n 1 | cut -f1 + register: etcd_disk_usage + when: embedded_etcd | bool + + - 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) + + - name: Install etcd (for etcdctl) + action: "{{ ansible_pkg_mgr }} name=etcd state=present" + when: not openshift.common.is_atomic | bool + + - name: Generate etcd backup + command: > + etcdctl backup --data-dir={{ openshift.etcd.etcd_data_dir }} + --backup-dir={{ openshift.common.data_dir }}/etcd-backup-{{ backup_tag | default('') }}{{ timestamp }} + + - set_fact: + etcd_backup_complete: True + + - name: Display location of etcd backup + debug: + msg: "Etcd backup created in {{ openshift.common.data_dir }}/etcd-backup-{{ backup_tag | default('') }}{{ timestamp }}" + +- name: Gate on etcd backup + hosts: localhost + connection: local + become: no + tasks: + - set_fact: + etcd_backup_completed: "{{ hostvars + | oo_select_keys(groups.etcd_hosts_to_backup) + | oo_collect('inventory_hostname', {'etcd_backup_complete': true}) }}" + - set_fact: + etcd_backup_failed: "{{ groups.etcd_hosts_to_backup | difference(etcd_backup_completed) }}" + - fail: + msg: "Upgrade cannot continue. The following hosts did not complete etcd backup: {{ etcd_backup_failed | join(',') }}" + when: etcd_backup_failed | length > 0 diff --git a/playbooks/common/openshift-cluster/upgrades/etcd/fedora_tasks.yml b/playbooks/common/openshift-cluster/upgrades/etcd/fedora_tasks.yml new file mode 100644 index 000000000..30232110e --- /dev/null +++ b/playbooks/common/openshift-cluster/upgrades/etcd/fedora_tasks.yml @@ -0,0 +1,23 @@ +--- +# 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/filter_plugins b/playbooks/common/openshift-cluster/upgrades/etcd/filter_plugins new file mode 120000 index 000000000..27ddaa18b --- /dev/null +++ b/playbooks/common/openshift-cluster/upgrades/etcd/filter_plugins @@ -0,0 +1 @@ +../../../../../filter_plugins \ No newline at end of file diff --git a/playbooks/common/openshift-cluster/upgrades/etcd/lookup_plugins b/playbooks/common/openshift-cluster/upgrades/etcd/lookup_plugins new file mode 120000 index 000000000..cf407f69b --- /dev/null +++ b/playbooks/common/openshift-cluster/upgrades/etcd/lookup_plugins @@ -0,0 +1 @@ +../../../../../lookup_plugins \ No newline at end of file diff --git a/playbooks/common/openshift-cluster/upgrades/etcd/main.yml b/playbooks/common/openshift-cluster/upgrades/etcd/main.yml new file mode 100644 index 000000000..e1c092dbe --- /dev/null +++ b/playbooks/common/openshift-cluster/upgrades/etcd/main.yml @@ -0,0 +1,88 @@ +--- +# For 1.4/3.4 we want to upgrade everyone to etcd-3.0. etcd docs say to +# upgrade from 2.0.x to 2.1.x to 2.2.x to 2.3.x to 3.0.x. While this is a tedius +# task for RHEL and CENTOS it's simply not possible in Fedora unless you've +# mirrored packages on your own because only the GA and latest versions are +# available in the repos. So for Fedora we'll simply skip this, sorry. + +- include: ../../evaluate_groups.yml + tags: + - always + +- name: Evaluate additional groups for upgrade + hosts: localhost + connection: local + become: no + tasks: + - name: Evaluate etcd_hosts_to_upgrade + add_host: + name: "{{ item }}" + groups: etcd_hosts_to_upgrade, etcd_hosts_to_backup + with_items: "{{ groups.oo_etcd_to_config if groups.oo_etcd_to_config is defined and groups.oo_etcd_to_config | length > 0 else groups.oo_first_master }}" + +- name: Backup etcd before upgrading anything + include: backup.yml + vars: + backup_tag: "pre-upgrade-" + +- name: Drop etcdctl profiles + hosts: etcd_hosts_to_upgrade + tasks: + - include: roles/etcd/tasks/etcdctl.yml + +- name: Determine etcd version + hosts: etcd_hosts_to_upgrade + tasks: + - name: Record etcd version + command: rpm -q --qf '%{version}' etcd + register: etcd_installed_version + +# 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: etcd_hosts_to_upgrade + serial: 1 + vars: + upgrade_version: '2.1' + tasks: + - include: rhel_tasks.yml + when: etcd_installed_version.stdout | version_compare('2.1','<') and ansible_distribution == 'RedHat' + +- name: Upgrade to 2.2 + hosts: etcd_hosts_to_upgrade + serial: 1 + vars: + upgrade_version: '2.2' + tasks: + - include: rhel_tasks.yml + when: etcd_installed_version.stdout | version_compare('2.2','<') and ansible_distribution == 'RedHat' + +- name: Upgrade to 2.3 + hosts: etcd_hosts_to_upgrade + serial: 1 + vars: + upgrade_version: '2.3' + tasks: + - include: rhel_tasks.yml + when: etcd_installed_version.stdout | version_compare('2.3','<') and ansible_distribution == 'RedHat' + +- name: Upgrade to 3.0 + hosts: etcd_hosts_to_upgrade + serial: 1 + vars: + upgrade_version: '3.0' + tasks: + - include: rhel_tasks.yml + when: etcd_installed_version.stdout | version_compare('3.0','<') and ansible_distribution == 'RedHat' + +- name: Upgrade fedora to latest + hosts: etcd_hosts_to_upgrade + serial: 1 + tasks: + - include: fedora_tasks.yml + when: ansible_distribution == 'Fedora' + +- name: Backup etcd + include: backup.yml + vars: + backup_tag: "post-3.0-" diff --git a/playbooks/common/openshift-cluster/upgrades/etcd/rhel_tasks.yml b/playbooks/common/openshift-cluster/upgrades/etcd/rhel_tasks.yml new file mode 100644 index 000000000..8e7dc9d9b --- /dev/null +++ b/playbooks/common/openshift-cluster/upgrades/etcd/rhel_tasks.yml @@ -0,0 +1,23 @@ +--- +- 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 but exclude etcd3 + command: "{{ ansible_pkg_mgr }} install -y etcd-{{ upgrade_version }}\\* --exclude etcd3" + when: upgrade_version | version_compare('3.0','<') + +- name: Update etcd package not excluding etcd3 + command: "{{ ansible_pkg_mgr }} install -y etcd3-{{ upgrade_version }}\\*" + when: not upgrade_version | version_compare('3.0','<') + +- 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/roles b/playbooks/common/openshift-cluster/upgrades/etcd/roles new file mode 120000 index 000000000..6bc1a7aef --- /dev/null +++ b/playbooks/common/openshift-cluster/upgrades/etcd/roles @@ -0,0 +1 @@ +../../../../../roles \ No newline at end of file diff --git a/playbooks/common/openshift-cluster/upgrades/etcd_backup.yml b/playbooks/common/openshift-cluster/upgrades/etcd_backup.yml deleted file mode 100644 index 57b156b1c..000000000 --- a/playbooks/common/openshift-cluster/upgrades/etcd_backup.yml +++ /dev/null @@ -1,73 +0,0 @@ -- name: Backup etcd - hosts: etcd_hosts_to_backup - vars: - embedded_etcd: "{{ hostvars[groups.oo_first_master.0].openshift.master.embedded_etcd }}" - 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" - - - stat: path=/var/lib/openshift - register: var_lib_openshift - - - stat: path=/var/lib/origin - register: var_lib_origin - - - name: Create origin symlink if necessary - file: src=/var/lib/openshift/ dest=/var/lib/origin state=link - when: var_lib_openshift.stat.exists == True and var_lib_origin.stat.exists == False - - # TODO: replace shell module with command and update later checks - # We assume to be using the data dir for all backups. - - name: Check available disk space for etcd backup - shell: df --output=avail -k {{ openshift.common.data_dir }} | tail -n 1 - register: avail_disk - - # TODO: replace shell module with command and update later checks - - name: Check current embedded etcd disk usage - shell: du -k {{ openshift.etcd.etcd_data_dir }} | tail -n 1 | cut -f1 - register: etcd_disk_usage - when: embedded_etcd | bool - - - 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) - - - name: Install etcd (for etcdctl) - action: "{{ ansible_pkg_mgr }} name=etcd state=present" - when: not openshift.common.is_atomic | bool - - - name: Generate etcd backup - command: > - etcdctl backup --data-dir={{ openshift.etcd.etcd_data_dir }} - --backup-dir={{ openshift.common.data_dir }}/etcd-backup-{{ backup_tag | default('') }}{{ timestamp }} - - - set_fact: - etcd_backup_complete: True - - - name: Display location of etcd backup - debug: - msg: "Etcd backup created in {{ openshift.common.data_dir }}/etcd-backup-{{ backup_tag | default('') }}{{ timestamp }}" - -- name: Gate on etcd backup - hosts: localhost - connection: local - become: no - tasks: - - set_fact: - etcd_backup_completed: "{{ hostvars - | oo_select_keys(groups.etcd_hosts_to_backup) - | oo_collect('inventory_hostname', {'etcd_backup_complete': true}) }}" - - set_fact: - etcd_backup_failed: "{{ groups.etcd_hosts_to_backup | difference(etcd_backup_completed) }}" - - fail: - msg: "Upgrade cannot continue. The following hosts did not complete etcd backup: {{ etcd_backup_failed | join(',') }}" - when: etcd_backup_failed | length > 0 diff --git a/playbooks/common/openshift-cluster/upgrades/upgrade_control_plane.yml b/playbooks/common/openshift-cluster/upgrades/upgrade_control_plane.yml index e1342f76b..339af3d0a 100644 --- a/playbooks/common/openshift-cluster/upgrades/upgrade_control_plane.yml +++ b/playbooks/common/openshift-cluster/upgrades/upgrade_control_plane.yml @@ -27,7 +27,7 @@ embedded_etcd: "{{ groups.oo_etcd_to_config | default([]) | length == 0 }}" - name: Backup etcd - include: etcd_backup.yml + include: ./etcd/backup.yml - name: Upgrade master packages hosts: oo_masters_to_config -- cgit v1.2.3