diff options
author | Kenny Woodson <kwoodson@redhat.com> | 2015-11-12 10:42:39 -0500 |
---|---|---|
committer | Kenny Woodson <kwoodson@redhat.com> | 2015-11-12 10:42:39 -0500 |
commit | 5ed42612965d72b87638ebe2fa96bec89199c4fa (patch) | |
tree | c58b76b1a9b42ee4adba2a07d5917e519fef9691 /playbooks/common/openshift-cluster | |
parent | 7869fb8c26a96c1e0ee74b930fd0da8a9952cb52 (diff) | |
parent | 597ba24415d6b7faa7ca02d200c8aed3c08b925a (diff) | |
download | openshift-5ed42612965d72b87638ebe2fa96bec89199c4fa.tar.gz openshift-5ed42612965d72b87638ebe2fa96bec89199c4fa.tar.bz2 openshift-5ed42612965d72b87638ebe2fa96bec89199c4fa.tar.xz openshift-5ed42612965d72b87638ebe2fa96bec89199c4fa.zip |
Merge pull request #878 from openshift/master
Master to prod
Diffstat (limited to 'playbooks/common/openshift-cluster')
19 files changed, 929 insertions, 4 deletions
diff --git a/playbooks/common/openshift-cluster/config.yml b/playbooks/common/openshift-cluster/config.yml index 57de7130b..a8bd634d3 100644 --- a/playbooks/common/openshift-cluster/config.yml +++ b/playbooks/common/openshift-cluster/config.yml @@ -8,4 +8,4 @@ - include: ../openshift-node/config.yml vars: osn_cluster_dns_domain: "{{ hostvars[groups.oo_first_master.0].openshift.dns.domain }}" - osn_cluster_dns_ip: "{{ hostvars[groups.oo_first_master.0].openshift.dns.ip }}" + osn_cluster_dns_ip: "{{ hostvars[groups.oo_first_master.0].cluster_dns_ip }}" diff --git a/playbooks/common/openshift-cluster/evaluate_groups.yml b/playbooks/common/openshift-cluster/evaluate_groups.yml index 1919660dd..2bb69614f 100644 --- a/playbooks/common/openshift-cluster/evaluate_groups.yml +++ b/playbooks/common/openshift-cluster/evaluate_groups.yml @@ -4,17 +4,21 @@ gather_facts: no tasks: - fail: - msg: This playbook rquires g_etcd_group to be set + msg: This playbook requires g_etcd_group to be set when: g_etcd_group is not defined - fail: - msg: This playbook rquires g_masters_group to be set + msg: This playbook requires g_masters_group to be set when: g_masters_group is not defined - fail: - msg: This playbook rquires g_nodes_group to be set + msg: This playbook requires g_nodes_group to be set when: g_nodes_group is not defined + - fail: + msg: This playbook requires g_lb_group to be set + when: g_lb_group is not defined + - name: Evaluate oo_etcd_to_config add_host: name: "{{ item }}" @@ -62,3 +66,11 @@ ansible_ssh_user: "{{ g_ssh_user | default(omit) }}" ansible_sudo: "{{ g_sudo | default(omit) }}" when: g_masters_group in groups and (groups[g_masters_group] | length) > 0 + + - name: Evaluate oo_lb_to_config + add_host: + name: "{{ item }}" + groups: oo_lb_to_config + ansible_ssh_user: "{{ g_ssh_user | default(omit) }}" + ansible_sudo: "{{ g_sudo | default(omit) }}" + with_items: groups[g_lb_group] | default([]) diff --git a/playbooks/common/openshift-cluster/scaleup.yml b/playbooks/common/openshift-cluster/scaleup.yml index 201320de8..6d2777732 100644 --- a/playbooks/common/openshift-cluster/scaleup.yml +++ b/playbooks/common/openshift-cluster/scaleup.yml @@ -1,7 +1,16 @@ --- - include: evaluate_groups.yml + vars: + g_etcd_group: "{{ 'etcd' }}" + g_masters_group: "{{ 'masters' }}" + g_nodes_group: "{{ 'nodes' }}" + g_lb_group: "{{ 'lb' }}" + openshift_cluster_id: "{{ cluster_id | default('default') }}" + openshift_debug_level: 2 + openshift_deployment_type: "{{ deployment_type }}" - include: ../openshift-node/config.yml vars: osn_cluster_dns_domain: "{{ hostvars[groups.oo_first_master.0].openshift.dns.domain }}" osn_cluster_dns_ip: "{{ hostvars[groups.oo_first_master.0].openshift.dns.ip }}" + openshift_deployment_type: "{{ deployment_type }}" diff --git a/playbooks/common/openshift-cluster/upgrades/files/pre-upgrade-check b/playbooks/common/openshift-cluster/upgrades/files/pre-upgrade-check new file mode 100644 index 000000000..ed4ab6d1b --- /dev/null +++ b/playbooks/common/openshift-cluster/upgrades/files/pre-upgrade-check @@ -0,0 +1,188 @@ +#!/usr/bin/env python +""" +Pre-upgrade checks that must be run on a master before proceeding with upgrade. +""" +# This is a script not a python module: +# pylint: disable=invalid-name + +# NOTE: This script should not require any python libs other than what is +# in the standard library. + +__license__ = "ASL 2.0" + +import json +import os +import subprocess +import re + +# The maximum length of container.ports.name +ALLOWED_LENGTH = 15 +# The valid structure of container.ports.name +ALLOWED_CHARS = re.compile('^[a-z0-9][a-z0-9\\-]*[a-z0-9]$') +AT_LEAST_ONE_LETTER = re.compile('[a-z]') +# look at OS_PATH for the full path. Default ot 'oc' +OC_PATH = os.getenv('OC_PATH', 'oc') + + +def validate(value): + """ + validate verifies that value matches required conventions + + Rules of container.ports.name validation: + + * must be less that 16 chars + * at least one letter + * only a-z0-9- + * hyphens can not be leading or trailing or next to each other + + :Parameters: + - `value`: Value to validate + """ + if len(value) > ALLOWED_LENGTH: + return False + + if '--' in value: + return False + + # We search since it can be anywhere + if not AT_LEAST_ONE_LETTER.search(value): + return False + + # We match because it must start at the beginning + if not ALLOWED_CHARS.match(value): + return False + return True + + +def list_items(kind): + """ + list_items returns a list of items from the api + + :Parameters: + - `kind`: Kind of item to access + """ + response = subprocess.check_output([OC_PATH, 'get', '--all-namespaces', '-o', 'json', kind]) + items = json.loads(response) + return items.get("items", []) + + +def get(obj, *paths): + """ + Gets an object + + :Parameters: + - `obj`: A dictionary structure + - `path`: All other non-keyword arguments + """ + ret_obj = obj + for path in paths: + if ret_obj.get(path, None) is None: + return [] + ret_obj = ret_obj[path] + return ret_obj + + +# pylint: disable=too-many-arguments +def pretty_print_errors(namespace, kind, item_name, container_name, port_name, valid): + """ + Prints out results in human friendly way. + + :Parameters: + - `namespace`: Namespace of the resource + - `kind`: Kind of the resource + - `item_name`: Name of the resource + - `container_name`: Name of the container. May be "" when kind=Service. + - `port_name`: Name of the port + - `valid`: True if the port is valid + """ + if not valid: + if len(container_name) > 0: + print('%s/%s -n %s (Container="%s" Port="%s")' % ( + kind, item_name, namespace, container_name, port_name)) + else: + print('%s/%s -n %s (Port="%s")' % ( + kind, item_name, namespace, port_name)) + + +def print_validation_header(): + """ + Prints the error header. Should run on the first error to avoid + overwhelming the user. + """ + print """\ +At least one port name does not validate. Valid port names: + + * must be less that 16 chars + * have at least one letter + * only a-z0-9- + * do not start or end with - + * Dashes may not be next to eachother ('--') +""" + + +def main(): + """ + main is the main entry point to this script + """ + try: + # the comma at the end suppresses the newline + print "Checking for oc ...", + subprocess.check_output([OC_PATH, 'whoami']) + print "found" + except: + print( + 'Unable to run "%s whoami"\n' + 'Please ensure OpenShift is running, and "oc" is on your system ' + 'path.\n' + 'You can override the path with the OC_PATH environment variable.' + % OC_PATH) + raise SystemExit(1) + + # Where the magic happens + first_error = True + for kind, path in [ + ('replicationcontrollers', ("spec", "template", "spec", "containers")), + ('pods', ("spec", "containers")), + ('deploymentconfigs', ("spec", "template", "spec", "containers"))]: + for item in list_items(kind): + namespace = item["metadata"]["namespace"] + item_name = item["metadata"]["name"] + for container in get(item, *path): + container_name = container["name"] + for port in get(container, "ports"): + port_name = port.get("name", None) + if not port_name: + # Unnamed ports are OK + continue + valid = validate(port_name) + if not valid and first_error: + first_error = False + print_validation_header() + pretty_print_errors( + namespace, kind, item_name, + container_name, port_name, valid) + + # Services follow a different flow + for item in list_items('services'): + namespace = item["metadata"]["namespace"] + item_name = item["metadata"]["name"] + for port in get(item, "spec", "ports"): + port_name = port.get("targetPort", None) + if isinstance(port_name, int) or port_name is None: + # Integer only or unnamed ports are OK + continue + valid = validate(port_name) + if not valid and first_error: + first_error = False + print_validation_header() + pretty_print_errors( + namespace, "services", item_name, "", port_name, valid) + + # If we had at least 1 error then exit with 1 + if not first_error: + raise SystemExit(1) + + +if __name__ == '__main__': + main() + diff --git a/playbooks/common/openshift-cluster/upgrades/files/versions.sh b/playbooks/common/openshift-cluster/upgrades/files/versions.sh new file mode 100644 index 000000000..f90719cab --- /dev/null +++ b/playbooks/common/openshift-cluster/upgrades/files/versions.sh @@ -0,0 +1,10 @@ +#!/bin/bash + +yum_installed=$(yum list installed "$@" 2>&1 | tail -n +2 | grep -v 'Installed Packages' | grep -v 'Red Hat Subscription Management' | grep -v 'Error:' | awk '{ print $2 }' | tr '\n' ' ') + +yum_available=$(yum list available "$@" 2>&1 | tail -n +2 | grep -v 'Available Packages' | grep -v 'Red Hat Subscription Management' | grep -v 'el7ose' | grep -v 'Error:' | awk '{ print $2 }' | tr '\n' ' ') + + +echo "---" +echo "curr_version: ${yum_installed}" +echo "avail_version: ${yum_available}" diff --git a/playbooks/common/openshift-cluster/upgrades/filter_plugins b/playbooks/common/openshift-cluster/upgrades/filter_plugins new file mode 120000 index 000000000..b1213dedb --- /dev/null +++ b/playbooks/common/openshift-cluster/upgrades/filter_plugins @@ -0,0 +1 @@ +../../../../filter_plugins
\ No newline at end of file diff --git a/playbooks/common/openshift-cluster/upgrades/library/openshift_upgrade_config.py b/playbooks/common/openshift-cluster/upgrades/library/openshift_upgrade_config.py new file mode 100755 index 000000000..a6721bb92 --- /dev/null +++ b/playbooks/common/openshift-cluster/upgrades/library/openshift_upgrade_config.py @@ -0,0 +1,154 @@ +#!/usr/bin/python +# -*- coding: utf-8 -*- +# vim: expandtab:tabstop=4:shiftwidth=4 + +"""Ansible module for modifying OpenShift configs during an upgrade""" + +import os +import yaml + +DOCUMENTATION = ''' +--- +module: openshift_upgrade_config +short_description: OpenShift Upgrade Config +author: Jason DeTiberus +requirements: [ ] +''' +EXAMPLES = ''' +''' + +def modify_api_levels(level_list, remove, ensure, msg_prepend='', + msg_append=''): + """ modify_api_levels """ + changed = False + changes = [] + + if not isinstance(remove, list): + remove = [] + + if not isinstance(ensure, list): + ensure = [] + + if not isinstance(level_list, list): + new_list = [] + changed = True + changes.append("%s created missing %s" % (msg_prepend, msg_append)) + else: + new_list = level_list + for level in remove: + if level in new_list: + new_list.remove(level) + changed = True + changes.append("%s removed %s %s" % (msg_prepend, level, msg_append)) + + for level in ensure: + if level not in new_list: + new_list.append(level) + changed = True + changes.append("%s added %s %s" % (msg_prepend, level, msg_append)) + + return {'new_list': new_list, 'changed': changed, 'changes': changes} + + +def upgrade_master_3_0_to_3_1(ansible_module, config_base, backup): + """Main upgrade method for 3.0 to 3.1.""" + changes = [] + + # Facts do not get transferred to the hosts where custom modules run, + # need to make some assumptions here. + master_config = os.path.join(config_base, 'master/master-config.yaml') + + master_cfg_file = open(master_config, 'r') + config = yaml.safe_load(master_cfg_file.read()) + master_cfg_file.close() + + + # Remove unsupported api versions and ensure supported api versions from + # master config + unsupported_levels = ['v1beta1', 'v1beta2', 'v1beta3'] + supported_levels = ['v1'] + + result = modify_api_levels(config.get('apiLevels'), unsupported_levels, + supported_levels, 'master-config.yaml:', 'from apiLevels') + if result['changed']: + config['apiLevels'] = result['new_list'] + changes.append(result['changes']) + + if 'kubernetesMasterConfig' in config and 'apiLevels' in config['kubernetesMasterConfig']: + config['kubernetesMasterConfig'].pop('apiLevels') + changes.append('master-config.yaml: removed kubernetesMasterConfig.apiLevels') + + # Add proxyClientInfo to master-config + if 'proxyClientInfo' not in config['kubernetesMasterConfig']: + config['kubernetesMasterConfig']['proxyClientInfo'] = { + 'certFile': 'master.proxy-client.crt', + 'keyFile': 'master.proxy-client.key' + } + changes.append("master-config.yaml: added proxyClientInfo") + + if len(changes) > 0: + if backup: + # TODO: Check success: + ansible_module.backup_local(master_config) + + # Write the modified config: + out_file = open(master_config, 'w') + out_file.write(yaml.safe_dump(config, default_flow_style=False)) + out_file.close() + + return changes + + +def upgrade_master(ansible_module, config_base, from_version, to_version, backup): + """Upgrade entry point.""" + if from_version == '3.0': + if to_version == '3.1': + return upgrade_master_3_0_to_3_1(ansible_module, config_base, backup) + + +def main(): + """ main """ + # disabling pylint errors for global-variable-undefined and invalid-name + # for 'global module' usage, since it is required to use ansible_facts + # pylint: disable=global-variable-undefined, invalid-name, + # redefined-outer-name + global module + + module = AnsibleModule( + argument_spec=dict( + config_base=dict(required=True), + from_version=dict(required=True, choices=['3.0']), + to_version=dict(required=True, choices=['3.1']), + role=dict(required=True, choices=['master']), + backup=dict(required=False, default=True, type='bool') + ), + supports_check_mode=True, + ) + + from_version = module.params['from_version'] + to_version = module.params['to_version'] + role = module.params['role'] + backup = module.params['backup'] + config_base = module.params['config_base'] + + try: + changes = [] + if role == 'master': + changes = upgrade_master(module, config_base, from_version, + to_version, backup) + + changed = len(changes) > 0 + return module.exit_json(changed=changed, changes=changes) + + # ignore broad-except error to avoid stack trace to ansible user + # pylint: disable=broad-except + except Exception, e: + return module.fail_json(msg=str(e)) + +# ignore pylint errors related to the module_utils import +# pylint: disable=redefined-builtin, unused-wildcard-import, wildcard-import +# import module snippets +from ansible.module_utils.basic import * + +if __name__ == '__main__': + main() diff --git a/playbooks/common/openshift-cluster/upgrades/lookup_plugins b/playbooks/common/openshift-cluster/upgrades/lookup_plugins new file mode 120000 index 000000000..aff753026 --- /dev/null +++ b/playbooks/common/openshift-cluster/upgrades/lookup_plugins @@ -0,0 +1 @@ +../../../../lookup_plugins
\ No newline at end of file diff --git a/playbooks/common/openshift-cluster/upgrades/roles b/playbooks/common/openshift-cluster/upgrades/roles new file mode 120000 index 000000000..4bdbcbad3 --- /dev/null +++ b/playbooks/common/openshift-cluster/upgrades/roles @@ -0,0 +1 @@ +../../../../roles
\ No newline at end of file diff --git a/playbooks/common/openshift-cluster/upgrades/v3_0_minor/filter_plugins b/playbooks/common/openshift-cluster/upgrades/v3_0_minor/filter_plugins new file mode 120000 index 000000000..27ddaa18b --- /dev/null +++ b/playbooks/common/openshift-cluster/upgrades/v3_0_minor/filter_plugins @@ -0,0 +1 @@ +../../../../../filter_plugins
\ No newline at end of file diff --git a/playbooks/common/openshift-cluster/upgrades/v3_0_minor/library b/playbooks/common/openshift-cluster/upgrades/v3_0_minor/library new file mode 120000 index 000000000..53bed9684 --- /dev/null +++ b/playbooks/common/openshift-cluster/upgrades/v3_0_minor/library @@ -0,0 +1 @@ +../library
\ No newline at end of file diff --git a/playbooks/common/openshift-cluster/upgrades/v3_0_minor/lookup_plugins b/playbooks/common/openshift-cluster/upgrades/v3_0_minor/lookup_plugins new file mode 120000 index 000000000..cf407f69b --- /dev/null +++ b/playbooks/common/openshift-cluster/upgrades/v3_0_minor/lookup_plugins @@ -0,0 +1 @@ +../../../../../lookup_plugins
\ No newline at end of file diff --git a/playbooks/common/openshift-cluster/upgrades/v3_0_minor/roles b/playbooks/common/openshift-cluster/upgrades/v3_0_minor/roles new file mode 120000 index 000000000..6bc1a7aef --- /dev/null +++ b/playbooks/common/openshift-cluster/upgrades/v3_0_minor/roles @@ -0,0 +1 @@ +../../../../../roles
\ No newline at end of file diff --git a/playbooks/common/openshift-cluster/upgrades/v3_0_minor/upgrade.yml b/playbooks/common/openshift-cluster/upgrades/v3_0_minor/upgrade.yml new file mode 100644 index 000000000..9f7e49b93 --- /dev/null +++ b/playbooks/common/openshift-cluster/upgrades/v3_0_minor/upgrade.yml @@ -0,0 +1,112 @@ +--- +- name: Evaluate groups + include: ../../evaluate_groups.yml + +- name: Re-Run cluster configuration to apply latest configuration changes + include: ../../config.yml + +- name: Upgrade masters + hosts: oo_masters_to_config + vars: + openshift_version: "{{ openshift_pkg_version | default('') }}" + tasks: + - name: Upgrade master packages + yum: pkg={{ openshift.common.service_type }}-master{{ openshift_version }} state=latest + - name: Restart master services + service: name="{{ openshift.common.service_type}}-master" state=restarted + +- name: Upgrade nodes + hosts: oo_nodes_to_config + vars: + openshift_version: "{{ openshift_pkg_version | default('') }}" + tasks: + - name: Upgrade node packages + yum: pkg={{ openshift.common.service_type }}-node{{ openshift_version }} state=latest + - name: Restart node services + service: name="{{ openshift.common.service_type }}-node" state=restarted + +- name: Determine new master version + hosts: oo_first_master + tasks: + - name: Determine new version + command: > + rpm -q --queryformat '%{version}' {{ openshift.common.service_type }}-master + register: _new_version + +- name: Ensure AOS 3.0.2 or Origin 1.0.6 + hosts: oo_first_master + tasks: + fail: This playbook requires Origin 1.0.6 or Atomic OpenShift 3.0.2 or later + when: _new_version.stdout | version_compare('1.0.6','<') or ( _new_version.stdout | version_compare('3.0','>=' and _new_version.stdout | version_compare('3.0.2','<') ) + +- name: Update cluster policy + hosts: oo_first_master + tasks: + - name: oadm policy reconcile-cluster-roles --confirm + command: > + {{ openshift.common.admin_binary}} --config={{ openshift.common.config_base }}/master/admin.kubeconfig + policy reconcile-cluster-roles --confirm + +- name: Upgrade default router + hosts: oo_first_master + vars: + - router_image: "{{ openshift.master.registry_url | replace( '${component}', 'haproxy-router' ) | replace ( '${version}', 'v' + _new_version.stdout ) }}" + - oc_cmd: "{{ openshift.common.client_binary }} --config={{ openshift.common.config_base }}/master/admin.kubeconfig" + tasks: + - name: Check for default router + command: > + {{ oc_cmd }} get -n default dc/router + register: _default_router + failed_when: false + changed_when: false + - name: Check for allowHostNetwork and allowHostPorts + when: _default_router.rc == 0 + shell: > + {{ oc_cmd }} get -o yaml scc/privileged | /usr/bin/grep -e allowHostPorts -e allowHostNetwork + register: _scc + - name: Grant allowHostNetwork and allowHostPorts + when: + - _default_router.rc == 0 + - "'false' in _scc.stdout" + command: > + {{ oc_cmd }} patch scc/privileged -p '{"allowHostPorts":true,"allowHostNetwork":true}' --loglevel=9 + - name: Update deployment config to 1.0.4/3.0.1 spec + when: _default_router.rc == 0 + command: > + {{ oc_cmd }} patch dc/router -p + '{"spec":{"strategy":{"rollingParams":{"updatePercent":-10},"spec":{"serviceAccount":"router","serviceAccountName":"router"}}}}' + - name: Switch to hostNetwork=true + when: _default_router.rc == 0 + command: > + {{ oc_cmd }} patch dc/router -p '{"spec":{"template":{"spec":{"hostNetwork":true}}}}' + - name: Update router image to current version + when: _default_router.rc == 0 + command: > + {{ oc_cmd }} patch dc/router -p + '{"spec":{"template":{"spec":{"containers":[{"name":"router","image":"{{ router_image }}"}]}}}}' + +- name: Upgrade default + hosts: oo_first_master + vars: + - registry_image: "{{ openshift.master.registry_url | replace( '${component}', 'docker-registry' ) | replace ( '${version}', 'v' + _new_version.stdout ) }}" + - oc_cmd: "{{ openshift.common.client_binary }} --config={{ openshift.common.config_base }}/master/admin.kubeconfig" + tasks: + - 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 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 }}"}]}}}}' + +- name: Update image streams and templates + hosts: oo_first_master + vars: + openshift_examples_import_command: "update" + openshift_deployment_type: "{{ deployment_type }}" + roles: + - openshift_examples diff --git a/playbooks/common/openshift-cluster/upgrades/v3_0_to_v3_1/filter_plugins b/playbooks/common/openshift-cluster/upgrades/v3_0_to_v3_1/filter_plugins new file mode 120000 index 000000000..27ddaa18b --- /dev/null +++ b/playbooks/common/openshift-cluster/upgrades/v3_0_to_v3_1/filter_plugins @@ -0,0 +1 @@ +../../../../../filter_plugins
\ No newline at end of file diff --git a/playbooks/common/openshift-cluster/upgrades/v3_0_to_v3_1/library b/playbooks/common/openshift-cluster/upgrades/v3_0_to_v3_1/library new file mode 120000 index 000000000..53bed9684 --- /dev/null +++ b/playbooks/common/openshift-cluster/upgrades/v3_0_to_v3_1/library @@ -0,0 +1 @@ +../library
\ No newline at end of file diff --git a/playbooks/common/openshift-cluster/upgrades/v3_0_to_v3_1/lookup_plugins b/playbooks/common/openshift-cluster/upgrades/v3_0_to_v3_1/lookup_plugins new file mode 120000 index 000000000..cf407f69b --- /dev/null +++ b/playbooks/common/openshift-cluster/upgrades/v3_0_to_v3_1/lookup_plugins @@ -0,0 +1 @@ +../../../../../lookup_plugins
\ No newline at end of file diff --git a/playbooks/common/openshift-cluster/upgrades/v3_0_to_v3_1/roles b/playbooks/common/openshift-cluster/upgrades/v3_0_to_v3_1/roles new file mode 120000 index 000000000..6bc1a7aef --- /dev/null +++ b/playbooks/common/openshift-cluster/upgrades/v3_0_to_v3_1/roles @@ -0,0 +1 @@ +../../../../../roles
\ No newline at end of file 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 new file mode 100644 index 000000000..78797f8b8 --- /dev/null +++ b/playbooks/common/openshift-cluster/upgrades/v3_0_to_v3_1/upgrade.yml @@ -0,0 +1,429 @@ +--- +############################################################################### +# Evaluate host groups and gather facts +############################################################################### +- name: Evaluate host groups + include: ../../evaluate_groups.yml + +- name: Load openshift_facts + hosts: oo_masters_to_config:oo_nodes_to_config:oo_etcd_to_config:oo_lb_to_config + roles: + - openshift_facts + +- name: Evaluate etcd_hosts_to_backup + hosts: localhost + tasks: + - name: Evaluate etcd_hosts_to_backup + add_host: + name: "{{ item }}" + groups: 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 + + +############################################################################### +# Pre-upgrade checks +############################################################################### +- name: Verify upgrade can proceed + hosts: oo_first_master + vars: + openshift_master_ha: "{{ groups.oo_masters_to_config | length > 1 }}" + gather_facts: no + tasks: + # Pacemaker is currently the only supported upgrade path for multiple masters + - fail: + msg: "openshift_master_cluster_method must be set to 'pacemaker'" + when: openshift_master_ha | bool and ((openshift_master_cluster_method is not defined) or (openshift_master_cluster_method is defined and openshift_master_cluster_method != "pacemaker")) + + - fail: + msg: > + This upgrade is only supported for origin and openshift-enterprise + deployment types + when: deployment_type not in ['origin','openshift-enterprise'] + + - fail: + msg: > + openshift_pkg_version is {{ openshift_pkg_version }} which is not a + valid version for a 3.1 upgrade + when: openshift_pkg_version is defined and openshift_pkg_version.split('-',1).1 | version_compare('3.0.2.900','<') + + # If this script errors out ansible will show the default stdout/stderr + # which contains details for the user: + - script: ../files/pre-upgrade-check + + +- name: Verify upgrade can proceed + hosts: masters:nodes + tasks: + - name: Clean yum cache + command: yum 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 1.0.6 or later + when: deployment_type == 'origin' and g_aos_versions.curr_version | version_compare('1.0.6','<') + + - fail: + msg: Atomic OpenShift 3.1 packages not found + when: g_aos_versions.curr_version | version_compare('3.0.2.900','<') and (g_aos_versions.avail_version is none or g_aos_versions.avail_version | version_compare('3.0.2.900','<')) + + +############################################################################### +# Backup etcd +############################################################################### +- name: Backup etcd + hosts: etcd_hosts_to_backup + vars: + embedded_etcd: "{{ openshift.master.embedded_etcd }}" + timestamp: "{{ lookup('pipe', 'date +%Y%m%d%H%M%S') }}" + roles: + - openshift_facts + tasks: + - 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) + yum: + pkg: etcd + state: latest + + - name: Generate etcd backup + command: > + etcdctl backup --data-dir={{ openshift.etcd.etcd_data_dir }} + --backup-dir={{ openshift.common.data_dir }}/etcd-backup-{{ timestamp }} + + - name: Display location of etcd backup + debug: + msg: "Etcd backup created in {{ openshift.common.data_dir }}/etcd-backup-{{ timestamp }}" + + +############################################################################### +# Upgrade Masters +############################################################################### +- name: Create temp directory for syncing certs + hosts: localhost + gather_facts: no + tasks: + - name: Create local temp directory for syncing certs + local_action: command mktemp -d /tmp/openshift-ansible-XXXXXXX + register: g_master_mktemp + changed_when: False + +- name: Update deployment type + hosts: OSEv3 + roles: + - openshift_facts + post_tasks: + - openshift_facts: + role: common + local_facts: + deployment_type: "{{ deployment_type }}" + +- name: Upgrade master packages and configuration + hosts: oo_masters_to_config + vars: + openshift_version: "{{ openshift_pkg_version | default('') }}" + tasks: + - name: Upgrade to latest available kernel + yum: + pkg: kernel + state: latest + + - name: Upgrade master packages + command: yum update -y {{ openshift.common.service_type }}-master{{ openshift_version }} + + - name: Ensure python-yaml present for config upgrade + yum: + pkg: PyYAML + state: installed + + - 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 }}" + + - set_fact: + master_certs_missing: True + master_cert_subdir: master-{{ openshift.common.hostname }} + master_cert_config_dir: "{{ openshift.common.config_base }}/master" + + +- name: Generate missing master certificates + hosts: oo_first_master + vars: + master_hostnames: "{{ hostvars + | oo_select_keys(groups.oo_masters_to_config) + | oo_collect('openshift.common.all_hostnames') + | 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]) }}" + sync_tmpdir: "{{ hostvars.localhost.g_master_mktemp.stdout }}" + openshift_deployment_type: "{{ deployment_type }}" + roles: + - openshift_master_certificates + post_tasks: + - name: Remove generated etcd client certs when using external etcd + file: + path: "{{ master_generated_certs_dir }}/{{ item.0.master_cert_subdir }}/{{ item.1 }}" + state: absent + when: groups.oo_etcd_to_config is defined and groups.oo_etcd_to_config + with_nested: + - masters_needing_certs + - - master.etcd-client.crt + - master.etcd-client.key + + - name: Create a tarball of the master certs + command: > + tar -czvf {{ master_generated_certs_dir }}/{{ item.master_cert_subdir }}.tgz + -C {{ master_generated_certs_dir }}/{{ item.master_cert_subdir }} . + with_items: masters_needing_certs + + - name: Retrieve the master cert tarball from the master + fetch: + src: "{{ master_generated_certs_dir }}/{{ item.master_cert_subdir }}.tgz" + dest: "{{ sync_tmpdir }}/" + flat: yes + fail_on_missing: yes + validate_checksum: yes + with_items: masters_needing_certs + + +- name: Sync generated certs, update service config and restart master services + hosts: oo_masters_to_config + vars: + sync_tmpdir: "{{ hostvars.localhost.g_master_mktemp.stdout }}" + openshift_master_ha: "{{ groups.oo_masters_to_config | length > 1 }}" + openshift_deployment_type: "{{ deployment_type }}" + tasks: + - name: Unarchive the tarball on the master + unarchive: + src: "{{ sync_tmpdir }}/{{ master_cert_subdir }}.tgz" + dest: "{{ master_cert_config_dir }}" + when: inventory_hostname != groups.oo_first_master.0 + + - name: Restart master service + service: name="{{ openshift.common.service_type}}-master" state=restarted + when: not openshift_master_ha | bool + + - name: Ensure the master service is enabled + service: name="{{ openshift.common.service_type}}-master" state=started enabled=yes + when: not openshift_master_ha | bool + + - name: Check for configured cluster + stat: + path: /etc/corosync/corosync.conf + register: corosync_conf + when: openshift_master_ha | bool + + - name: Destroy cluster + command: pcs cluster destroy --all + when: openshift_master_ha | bool and corosync_conf.stat.exists == true + run_once: true + + - name: Start pcsd + service: name=pcsd enabled=yes state=started + when: openshift_master_ha | bool + + +- name: Re-create cluster + hosts: oo_first_master + vars: + openshift_master_ha: "{{ groups.oo_masters_to_config | length > 1 }}" + openshift_deployment_type: "{{ deployment_type }}" + omc_cluster_hosts: "{{ groups.oo_masters_to_config | join(' ') }}" + roles: + - role: openshift_master_cluster + when: openshift_master_ha | bool + + +- name: Delete temporary directory on localhost + hosts: localhost + gather_facts: no + tasks: + - file: name={{ g_master_mktemp.stdout }} state=absent + changed_when: False + + +############################################################################### +# 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: yum update -y {{ openshift.common.service_type }}-node{{ openshift_version }} + + - name: Restart node service + service: name="{{ openshift.common.service_type }}-node" state=restarted + + - name: Ensure node service enabled + service: name="{{ openshift.common.service_type }}-node" state=started enabled=yes + + +############################################################################### +# Post upgrade - 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 + + - name: Restart master services + service: name="{{ openshift.common.service_type}}-master" state=restarted + when: not openshift_master_ha | bool + + - name: Restart master cluster + command: pcs resource restart master + when: openshift_master_ha | bool + run_once: true + + - name: Wait for the clustered master service to be available + wait_for: + host: "{{ openshift_master_cluster_vip }}" + port: 8443 + state: started + timeout: 180 + delay: 90 + when: openshift_master_ha | bool + run_once: true + + +############################################################################### +# 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, 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 allowHostNetwork and allowHostPorts + when: _default_router.rc == 0 + shell: > + {{ oc_cmd }} get -o yaml scc/privileged | /usr/bin/grep -e allowHostPorts -e allowHostNetwork + register: _scc + + - name: Grant allowHostNetwork and allowHostPorts + when: + - _default_router.rc == 0 + - "'false' in _scc.stdout" + command: > + {{ oc_cmd }} patch scc/privileged -p '{"allowHostPorts":true,"allowHostNetwork":true}' --loglevel=9 + + - name: Update deployment config to 1.0.4/3.0.1 spec + when: _default_router.rc == 0 + command: > + {{ oc_cmd }} patch dc/router -p + '{"spec":{"strategy":{"rollingParams":{"updatePercent":-10},"spec":{"serviceAccount":"router","serviceAccountName":"router"}}}}' + + - name: Switch to hostNetwork=true + when: _default_router.rc == 0 + command: > + {{ oc_cmd }} patch dc/router -p '{"spec":{"template":{"spec":{"hostNetwork":true}}}}' + + - 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 }}"}]}}}}' + + - 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 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 }}"}]}}}}' |