diff options
61 files changed, 1702 insertions, 207 deletions
diff --git a/.papr.inventory b/.papr.inventory index 878d434e2..aa4324c21 100644 --- a/.papr.inventory +++ b/.papr.inventory @@ -11,6 +11,9 @@ openshift_image_tag="{{ lookup('env', 'OPENSHIFT_IMAGE_TAG') }}" openshift_master_default_subdomain="{{ lookup('env', 'RHCI_ocp_node1_IP') }}.xip.io" openshift_check_min_host_disk_gb=1.5 openshift_check_min_host_memory_gb=1.9 +osm_cluster_network_cidr=10.128.0.0/14 +openshift_portal_net=172.30.0.0/16 +osm_host_subnet_length=9 [masters] ocp-master diff --git a/images/installer/README_INVENTORY_GENERATOR.md b/images/installer/README_INVENTORY_GENERATOR.md new file mode 100644 index 000000000..9c10e4b71 --- /dev/null +++ b/images/installer/README_INVENTORY_GENERATOR.md @@ -0,0 +1,85 @@ +Dynamic Inventory Generation +============================ + +Script within the openshift-ansible image that can dynamically +generate an Ansible inventory file from an existing cluster. + +## Configure + +User configuration helps to provide additional details when creating an inventory file. +The default location of this file is in `/etc/inventory-generator-config.yaml`. The +following configuration values are either expected or default to the given values when omitted: + +- `master_config_path`: + - specifies where to look for the bind-mounted `master-config.yaml` file in the container + - if omitted or a `null` value is given, its value is defaulted to `/opt/app-root/src/master-config.yaml` + +- `admin_kubeconfig_path`: + - specifies where to look for the bind-mounted `admin.kubeconfig` file in the container + - if omitted or a `null` value is given, its value is defaulted to `/opt/app-root/src/.kube/config` + +- `ansible_ssh_user`: + - specifies the ssh user to be used by Ansible when running the specified `PLAYBOOK_FILE` (see `README_CONTAINER_IMAGE.md` for additional information on this environment variable). + - if omitted, its value is defaulted to `root` + +- `ansible_become_user`: + - specifies a user to "become" on the remote host. Used for privilege escalation. + - If a non-null value is specified, `ansible_become` is implicitly set to `yes` in the resulting inventory file. + +See the supplied sample user configuration file in [`root/etc/inventory-generator-config.yaml`](./root/etc/inventory-generator-config.yaml) for additional optional inventory variables that may be specified. + +## Build + +See `README_CONTAINER_IMAGE.md` for information on building this image. + +## Run + +Given a master node's `master-config.yaml` file, a user configuration file (see "Configure" section), and an `admin.kubeconfig` file, the command below will: + +1. Use `oc` to query the host about additional node information (using the supplied `kubeconfig` file) +2. Generate an inventory file based on information retrieved from `oc get nodes` and the given `master-config.yaml` file. +3. run the specified [openshift-ansible](https://github.com/openshift/openshift-ansible) `health.yml` playbook using the generated inventory file from the previous step + +``` +docker run -u `id -u` \ + -v $HOME/.ssh/id_rsa:/opt/app-root/src/.ssh/id_rsa:Z,ro \ + -v /tmp/origin/master/admin.kubeconfig:/opt/app-root/src/.kube/config:Z \ + -v /tmp/origin/master/master-config.yaml:/opt/app-root/src/master-config.yaml:Z \ + -e OPTS="-v --become-user root" \ + -e PLAYBOOK_FILE=playbooks/byo/openshift-checks/health.yml \ + -e GENERATE_INVENTORY=true \ + -e USER=`whoami` \ + openshift/origin-ansible + +``` + +**Note** In the command above, specifying the `GENERATE_INVENTORY` environment variable will automatically generate the inventory file in an expected location. +An `INVENTORY_FILE` variable (or any other inventory location) does not need to be supplied when generating an inventory. + +## Debug + +To debug the `generate` script, run the above script interactively +and manually execute `/usr/local/bin/generate`: + +``` +... +docker run -u `id -u` \ + -v ... + ... + -it openshift/origin-ansible /bin/bash + +--- + +bash-4.2$ cd $HOME +bash-4.2$ ls +master-config.yaml +bash-4.2$ /usr/local/bin/generate $HOME/generated_hosts +bash-4.2$ ls +generated_hosts master-config.yaml +bash-4.2$ less generated_hosts +... +``` + +## Notes + +See `README_CONTAINER_IMAGE.md` for additional information about this image. diff --git a/images/installer/root/etc/inventory-generator-config.yaml b/images/installer/root/etc/inventory-generator-config.yaml new file mode 100644 index 000000000..d56e3f4d2 --- /dev/null +++ b/images/installer/root/etc/inventory-generator-config.yaml @@ -0,0 +1,20 @@ +--- +# meta config +master_config_path: "/opt/app-root/src/master-config.yaml" +admin_kubeconfig_path: "/opt/app-root/src/.kube/config" + +# default user configuration +ansible_ssh_user: ec2-user +ansible_become: "yes" +ansible_become_user: "root" + +# openshift-ansible inventory vars +openshift_uninstall_images: false +openshift_install_examples: true +openshift_deployment_type: origin + +openshift_release: 3.6 +openshift_image_tag: v3.6.0 +openshift_hosted_logging_deploy: null # defaults to "true" if loggingPublicURL is set in master-config.yaml +openshift_logging_image_version: v3.6.0 +openshift_disable_check: "" diff --git a/images/installer/root/exports/config.json.template b/images/installer/root/exports/config.json.template index 739c0080f..1a009fa7b 100644 --- a/images/installer/root/exports/config.json.template +++ b/images/installer/root/exports/config.json.template @@ -24,7 +24,7 @@ "PLAYBOOK_FILE=$PLAYBOOK_FILE", "ANSIBLE_CONFIG=$ANSIBLE_CONFIG" ], - "cwd": "/opt/app-root/src/", + "cwd": "/usr/share/ansible/openshift-ansible", "rlimits": [ { "type": "RLIMIT_NOFILE", diff --git a/images/installer/root/usr/local/bin/generate b/images/installer/root/usr/local/bin/generate new file mode 100755 index 000000000..3db7a3ee8 --- /dev/null +++ b/images/installer/root/usr/local/bin/generate @@ -0,0 +1,397 @@ +#!/bin/env python + +""" +Attempts to read 'master-config.yaml' and extract remote +host information to dynamically create an inventory file +in order to run Ansible playbooks against that host. +""" + +import os +import re +import shlex +import shutil +import subprocess +import sys +import yaml + +try: + HOME = os.environ['HOME'] +except KeyError: + print 'A required environment variable "$HOME" has not been set' + exit(1) + +DEFAULT_USER_CONFIG_PATH = '/etc/inventory-generator-config.yaml' +DEFAULT_MASTER_CONFIG_PATH = HOME + '/master-config.yaml' +DEFAULT_ADMIN_KUBECONFIG_PATH = HOME + '/.kube/config' + +INVENTORY_FULL_PATH = HOME + '/generated_hosts' +USE_STDOUT = True + +if len(sys.argv) > 1: + INVENTORY_FULL_PATH = sys.argv[1] + USE_STDOUT = False + + +class OpenShiftClientError(Exception): + """Base exception class for OpenShift CLI wrapper""" + pass + + +class InvalidHost(Exception): + """Base exception class for host creation problems.""" + pass + + +class InvalidHostGroup(Exception): + """Base exception class for host-group creation problems.""" + pass + + +class OpenShiftClient: + oc = None + kubeconfig = None + + def __init__(self, kubeconfig=DEFAULT_ADMIN_KUBECONFIG_PATH): + """Find and store path to oc binary""" + # https://github.com/openshift/openshift-ansible/issues/3410 + # oc can be in /usr/local/bin in some cases, but that may not + # be in $PATH due to ansible/sudo + paths = os.environ.get("PATH", os.defpath).split(os.pathsep) + ['/usr/local/bin', os.path.expanduser('~/bin')] + + oc_binary_name = 'oc' + oc_binary = None + + # Use shutil.which if it is available, otherwise fallback to a naive path search + try: + which_result = shutil.which(oc_binary_name, path=os.pathsep.join(paths)) + if which_result is not None: + oc_binary = which_result + except AttributeError: + for path in paths: + if os.path.exists(os.path.join(path, oc_binary_name)): + oc_binary = os.path.join(path, oc_binary_name) + break + + if oc_binary is None: + raise OpenShiftClientError('Unable to locate `oc` binary. Not present in PATH.') + + self.oc = oc_binary + self.kubeconfig = kubeconfig + + def call(self, cmd_str): + """Execute a remote call using `oc`""" + cmd = [ + self.oc, + '--config', + self.kubeconfig + ] + shlex.split(cmd_str) + try: + out = subprocess.check_output(list(cmd), stderr=subprocess.STDOUT) + except subprocess.CalledProcessError as err: + raise OpenShiftClientError('[rc {}] {}\n{}'.format(err.returncode, ' '.join(err.cmd), err.output)) + return out + + def whoami(self): + """Retrieve information about the current user in the given kubeconfig""" + return self.call('whoami') + + def get_nodes(self): + """Retrieve remote node information as a yaml object""" + return self.call('get nodes -o yaml') + + +class HostGroup: + groupname = "" + hosts = list() + + def __init__(self, hosts): + if not hosts: + return + first = hosts[0].get_group_name() + for h in hosts: + if h.get_group_name() != first: + raise InvalidHostGroup("Attempt to create HostGroup with hosts of varying groups.") + + self.hosts = hosts + self.groupname = first + + def add_host(self, host): + """Add a new host to this group.""" + self.hosts.append(host) + + def get_group_name(self): + """Return the groupname associated with each aggregated host.""" + return self.groupname + + def get_hosts(self): + """Return aggregated hosts""" + return self.hosts + + def string(self): + """Call the print method for each aggregated host; separated by newlines.""" + infos = "" + for host in self.hosts: + infos += host.string() + "\n" + return infos + + +class Host: + group = "masters" + alias = "" + hostname = "" + public_hostname = "" + ip_addr = "" + public_ip_addr = "" + + def __init__(self, groupname): + if not groupname: + raise InvalidHost("Attempt to create Host with no group name provided.") + self.group = groupname + + def get_group_name(self): + return self.group + + def get_openshift_hostname(self): + return self.hostname + + def host_alias(self, hostalias): + """Set an alias for this host.""" + self.alias = hostalias + + def address(self, ip): + """Set the ip address for this host.""" + self.ip_addr = ip + + def public_address(self, ip): + """Set the external ip address for this host.""" + self.public_ip_addr = ip + + def host_name(self, hname): + self.hostname = parse_hostname(hname) + + def public_host_name(self, phname): + self.public_hostname = parse_hostname(phname) + + def string(self): + """Print an inventory-file compatible string with host information""" + info = "" + if self.alias: + info += self.alias + " " + elif self.hostname: + info += self.hostname + " " + elif self.ip_addr: + info += self.ip_addr + " " + if self.ip_addr: + info += "openshift_ip=" + self.ip_addr + " " + if self.public_ip_addr: + info += "openshift_public_ip=" + self.public_ip_addr + " " + if self.hostname: + info += "openshift_hostname=" + self.hostname + " " + if self.public_hostname: + info += "openshift_public_hostname=" + self.public_hostname + + return info + + +def parse_hostname(host): + """Remove protocol and port from given hostname. + Return parsed string""" + no_proto = re.split('^http(s)?\:\/\/', host) + if no_proto: + host = no_proto[-1] + + no_port = re.split('\:[0-9]+(/)?$', host) + if no_port: + host = no_port[0] + + return host + + +def main(): + """Parse master-config file and populate inventory file.""" + # set default values + USER_CONFIG = os.environ.get('CONFIG') + if not USER_CONFIG: + USER_CONFIG = DEFAULT_USER_CONFIG_PATH + + # read user configuration + try: + config_file_obj = open(USER_CONFIG, 'r') + raw_config_file = config_file_obj.read() + user_config = yaml.load(raw_config_file) + if not user_config: + user_config = dict() + except IOError as err: + print "Unable to find or read user configuration file '{}': {}".format(USER_CONFIG, err) + exit(1) + + master_config_path = user_config.get('master_config_path', DEFAULT_MASTER_CONFIG_PATH) + if not master_config_path: + master_config_path = DEFAULT_MASTER_CONFIG_PATH + + admin_kubeconfig_path = user_config.get('admin_kubeconfig_path', DEFAULT_ADMIN_KUBECONFIG_PATH) + if not admin_kubeconfig_path: + admin_kubeconfig_path = DEFAULT_ADMIN_KUBECONFIG_PATH + + try: + file_obj = open(master_config_path, 'r') + except IOError as err: + print "Unable to find or read host master configuration file '{}': {}".format(master_config_path, err) + exit(1) + + raw_text = file_obj.read() + + y = yaml.load(raw_text) + if y.get("kind", "") != "MasterConfig": + print "Bind-mounted host master configuration file is not of 'kind' MasterConfig. Aborting..." + exit(1) + + # finish reading config file and begin gathering + # cluster information for inventory file + file_obj.close() + + # set inventory values based on user configuration + ansible_ssh_user = user_config.get('ansible_ssh_user', 'root') + ansible_become_user = user_config.get('ansible_become_user') + + openshift_uninstall_images = user_config.get('openshift_uninstall_images', False) + openshift_install_examples = user_config.get('openshift_install_examples', True) + openshift_deployment_type = user_config.get('openshift_deployment_type', 'origin') + + openshift_release = user_config.get('openshift_release') + openshift_image_tag = user_config.get('openshift_image_tag') + openshift_logging_image_version = user_config.get('openshift_logging_image_version') + openshift_disable_check = user_config.get('openshift_disable_check') + + # extract host config info from parsed yaml file + asset_config = y.get("assetConfig") + master_config = y.get("kubernetesMasterConfig") + etcd_config = y.get("etcdClientInfo") + + # if master_config is missing, error out; we expect to be running on a master to be able to + # gather enough information to generate the rest of the inventory file. + if not master_config: + msg = "'kubernetesMasterConfig' missing from '{}'; unable to gather all necessary host information..." + print msg.format(master_config_path) + exit(1) + + master_public_url = y.get("masterPublicURL") + if not master_public_url: + msg = "'kubernetesMasterConfig.masterPublicURL' missing from '{}'; Unable to connect to master host..." + print msg.format(master_config_path) + exit(1) + + oc = OpenShiftClient(admin_kubeconfig_path) + + # ensure kubeconfig is logged in with provided user, or fail with a friendly message otherwise + try: + oc.whoami() + except OpenShiftClientError as err: + msg = ("Unable to obtain user information using the provided kubeconfig file. " + "Current context does not appear to be able to authenticate to the server. " + "Error returned from server:\n\n{}") + print msg.format(str(err)) + exit(1) + + # connect to remote host using the provided config and extract all possible node information + nodes_config = yaml.load(oc.get_nodes()) + + # contains host types (e.g. masters, nodes, etcd) + host_groups = dict() + openshift_hosted_logging_deploy = False + is_etcd_deployed = master_config.get("storage-backend", "") in ["etcd3", "etcd2", "etcd"] + + if asset_config and asset_config.get('loggingPublicURL'): + openshift_hosted_logging_deploy = True + + openshift_hosted_logging_deploy = user_config.get("openshift_hosted_logging_deploy", openshift_hosted_logging_deploy) + + m = Host("masters") + m.address(master_config["masterIP"]) + m.public_host_name(master_public_url) + host_groups["masters"] = HostGroup([m]) + + if nodes_config: + node_hosts = list() + for node in nodes_config.get("items", []): + if node["kind"] != "Node": + continue + + n = Host("nodes") + + address = "" + internal_hostname = "" + for item in node["status"].get("addresses", []): + if not address and item['type'] in ['InternalIP', 'LegacyHostIP']: + address = item['address'] + + if item['type'] == 'Hostname': + internal_hostname = item['address'] + + n.address(address) + n.host_name(internal_hostname) + node_hosts.append(n) + + host_groups["nodes"] = HostGroup(node_hosts) + + if etcd_config: + etcd_hosts = list() + for url in etcd_config.get("urls", []): + e = Host("etcd") + e.host_name(url) + etcd_hosts.append(e) + + host_groups["etcd"] = HostGroup(etcd_hosts) + + # open new inventory file for writing + if USE_STDOUT: + inv_file_obj = sys.stdout + else: + try: + inv_file_obj = open(INVENTORY_FULL_PATH, 'w+') + except IOError as err: + print "Unable to create or open generated inventory file: {}".format(err) + exit(1) + + inv_file_obj.write("[OSEv3:children]\n") + for group in host_groups: + inv_file_obj.write("{}\n".format(group)) + inv_file_obj.write("\n") + + inv_file_obj.write("[OSEv3:vars]\n") + if ansible_ssh_user: + inv_file_obj.write("ansible_ssh_user={}\n".format(ansible_ssh_user)) + if ansible_become_user: + inv_file_obj.write("ansible_become_user={}\n".format(ansible_become_user)) + inv_file_obj.write("ansible_become=yes\n") + + if openshift_uninstall_images: + inv_file_obj.write("openshift_uninstall_images={}\n".format(str(openshift_uninstall_images))) + if openshift_deployment_type: + inv_file_obj.write("openshift_deployment_type={}\n".format(openshift_deployment_type)) + if openshift_install_examples: + inv_file_obj.write("openshift_install_examples={}\n".format(str(openshift_install_examples))) + + if openshift_release: + inv_file_obj.write("openshift_release={}\n".format(str(openshift_release))) + if openshift_image_tag: + inv_file_obj.write("openshift_image_tag={}\n".format(str(openshift_image_tag))) + if openshift_logging_image_version: + inv_file_obj.write("openshift_logging_image_version={}\n".format(str(openshift_logging_image_version))) + if openshift_disable_check: + inv_file_obj.write("openshift_disable_check={}\n".format(str(openshift_disable_check))) + inv_file_obj.write("\n") + + inv_file_obj.write("openshift_hosted_logging_deploy={}\n".format(str(openshift_hosted_logging_deploy))) + inv_file_obj.write("\n") + + for group in host_groups: + inv_file_obj.write("[{}]\n".format(host_groups[group].get_group_name())) + inv_file_obj.write(host_groups[group].string()) + inv_file_obj.write("\n") + + inv_file_obj.close() + + +if __name__ == '__main__': + main() diff --git a/images/installer/root/usr/local/bin/run b/images/installer/root/usr/local/bin/run index 9401ea118..70aa0bac3 100755 --- a/images/installer/root/usr/local/bin/run +++ b/images/installer/root/usr/local/bin/run @@ -24,9 +24,12 @@ elif [[ -v INVENTORY_URL ]]; then elif [[ -v DYNAMIC_SCRIPT_URL ]]; then curl -o ${INVENTORY} ${DYNAMIC_SCRIPT_URL} chmod 755 ${INVENTORY} +elif [[ -v GENERATE_INVENTORY ]]; then + # dynamically generate inventory file using bind-mounted info + /usr/local/bin/generate ${INVENTORY} else echo - echo "One of INVENTORY_FILE, INVENTORY_URL or DYNAMIC_SCRIPT_URL must be provided." + echo "One of INVENTORY_FILE, INVENTORY_URL, GENERATE_INVENTORY, or DYNAMIC_SCRIPT_URL must be provided." exec /usr/local/bin/usage fi INVENTORY_ARG="-i ${INVENTORY}" @@ -36,7 +39,7 @@ if [[ "$ALLOW_ANSIBLE_CONNECTION_LOCAL" = false ]]; then fi if [[ -v VAULT_PASS ]]; then - VAULT_PASS_FILE=.vaultpass + VAULT_PASS_FILE="$(mktemp)" echo ${VAULT_PASS} > ${VAULT_PASS_FILE} VAULT_PASS_ARG="--vault-password-file ${VAULT_PASS_FILE}" fi diff --git a/inventory/byo/hosts.origin.example b/inventory/byo/hosts.origin.example index dbe57bbd2..486fe56a0 100644 --- a/inventory/byo/hosts.origin.example +++ b/inventory/byo/hosts.origin.example @@ -34,17 +34,17 @@ openshift_deployment_type=origin # use this to lookup the latest exact version of the container images, which is the tag actually used to configure # the cluster. For RPM installations we just verify the version detected in your configured repos matches this # release. -openshift_release=v3.6 +openshift_release=v3.7 # Specify an exact container image tag to install or configure. # WARNING: This value will be used for all hosts in containerized environments, even those that have another version installed. # This could potentially trigger an upgrade and downtime, so be careful with modifying this value after the cluster is set up. -#openshift_image_tag=v3.6.0 +#openshift_image_tag=v3.7.0 # Specify an exact rpm version to install or configure. # WARNING: This value will be used for all hosts in RPM based environments, even those that have another version installed. # This could potentially trigger an upgrade and downtime, so be careful with modifying this value after the cluster is set up. -#openshift_pkg_version=-3.6.0 +#openshift_pkg_version=-3.7.0 # This enables all the system containers except for docker: #openshift_use_system_containers=False @@ -538,7 +538,7 @@ openshift_master_identity_providers=[{'name': 'htpasswd_auth', 'login': 'true', #openshift_hosted_metrics_public_url=https://hawkular-metrics.example.com/hawkular/metrics # Configure the prefix and version for the component images #openshift_hosted_metrics_deployer_prefix=docker.io/openshift/origin- -#openshift_hosted_metrics_deployer_version=v3.6.0 +#openshift_hosted_metrics_deployer_version=v3.7.0 # # StorageClass # openshift_storageclass_name=gp2 @@ -593,7 +593,7 @@ openshift_master_identity_providers=[{'name': 'htpasswd_auth', 'login': 'true', #openshift_hosted_logging_elasticsearch_cluster_size=1 # Configure the prefix and version for the component images #openshift_hosted_logging_deployer_prefix=docker.io/openshift/origin- -#openshift_hosted_logging_deployer_version=v3.6.0 +#openshift_hosted_logging_deployer_version=v3.7.0 # Configure the multi-tenant SDN plugin (default is 'redhat/openshift-ovs-subnet') # os_sdn_network_plugin_name='redhat/openshift-ovs-multitenant' @@ -613,7 +613,12 @@ openshift_master_identity_providers=[{'name': 'htpasswd_auth', 'login': 'true', # WORKAROUND : If you must use an overlapping subnet, you can configure a non conflicting # docker0 CIDR range by adding '--bip=192.168.2.1/24' to DOCKER_NETWORK_OPTIONS # environment variable located in /etc/sysconfig/docker-network. -# When upgrading these must be specificed! +# When upgrading or scaling up the following must match whats in your master config! +# Inventory: master yaml field +# osm_cluster_network_cidr: clusterNetworkCIDR +# openshift_portal_net: serviceNetworkCIDR +# When installing osm_cluster_network_cidr and openshift_portal_net must be set. +# Sane examples are provided below. #osm_cluster_network_cidr=10.128.0.0/14 #openshift_portal_net=172.30.0.0/16 @@ -635,7 +640,10 @@ openshift_master_identity_providers=[{'name': 'htpasswd_auth', 'login': 'true', # Configure number of bits to allocate to each host’s subnet e.g. 9 # would mean a /23 network on the host. -# When upgrading this must be specificed! +# When upgrading or scaling up the following must match whats in your master config! +# Inventory: master yaml field +# osm_host_subnet_length: hostSubnetLength +# When installing osm_host_subnet_length must be set. A sane example is provided below. #osm_host_subnet_length=9 # Configure master API and console ports. diff --git a/inventory/byo/hosts.ose.example b/inventory/byo/hosts.ose.example index 0d60de6d2..92a0927e5 100644 --- a/inventory/byo/hosts.ose.example +++ b/inventory/byo/hosts.ose.example @@ -34,17 +34,17 @@ openshift_deployment_type=openshift-enterprise # use this to lookup the latest exact version of the container images, which is the tag actually used to configure # the cluster. For RPM installations we just verify the version detected in your configured repos matches this # release. -openshift_release=v3.6 +openshift_release=v3.7 # Specify an exact container image tag to install or configure. # WARNING: This value will be used for all hosts in containerized environments, even those that have another version installed. # This could potentially trigger an upgrade and downtime, so be careful with modifying this value after the cluster is set up. -#openshift_image_tag=v3.6.0 +#openshift_image_tag=v3.7.0 # Specify an exact rpm version to install or configure. # WARNING: This value will be used for all hosts in RPM based environments, even those that have another version installed. # This could potentially trigger an upgrade and downtime, so be careful with modifying this value after the cluster is set up. -#openshift_pkg_version=-3.6.0 +#openshift_pkg_version=-3.7.0 # This enables all the system containers except for docker: #openshift_use_system_containers=False @@ -546,7 +546,7 @@ openshift_master_identity_providers=[{'name': 'htpasswd_auth', 'login': 'true', #openshift_hosted_metrics_public_url=https://hawkular-metrics.example.com/hawkular/metrics # Configure the prefix and version for the component images #openshift_hosted_metrics_deployer_prefix=registry.example.com:8888/openshift3/ -#openshift_hosted_metrics_deployer_version=3.6.0 +#openshift_hosted_metrics_deployer_version=3.7.0 # # StorageClass # openshift_storageclass_name=gp2 @@ -601,7 +601,7 @@ openshift_master_identity_providers=[{'name': 'htpasswd_auth', 'login': 'true', #openshift_hosted_logging_elasticsearch_cluster_size=1 # Configure the prefix and version for the component images #openshift_hosted_logging_deployer_prefix=registry.example.com:8888/openshift3/ -#openshift_hosted_logging_deployer_version=3.6.0 +#openshift_hosted_logging_deployer_version=3.7.0 # Configure the multi-tenant SDN plugin (default is 'redhat/openshift-ovs-subnet') # os_sdn_network_plugin_name='redhat/openshift-ovs-multitenant' @@ -621,7 +621,12 @@ openshift_master_identity_providers=[{'name': 'htpasswd_auth', 'login': 'true', # WORKAROUND : If you must use an overlapping subnet, you can configure a non conflicting # docker0 CIDR range by adding '--bip=192.168.2.1/24' to DOCKER_NETWORK_OPTIONS # environment variable located in /etc/sysconfig/docker-network. -# When upgrading these must be specificed! +# When upgrading or scaling up the following must match whats in your master config! +# Inventory: master yaml field +# osm_cluster_network_cidr: clusterNetworkCIDR +# openshift_portal_net: serviceNetworkCIDR +# When installing osm_cluster_network_cidr and openshift_portal_net must be set. +# Sane examples are provided below. #osm_cluster_network_cidr=10.128.0.0/14 #openshift_portal_net=172.30.0.0/16 @@ -643,7 +648,10 @@ openshift_master_identity_providers=[{'name': 'htpasswd_auth', 'login': 'true', # Configure number of bits to allocate to each host’s subnet e.g. 9 # would mean a /23 network on the host. -# When upgrading this must be specificed! +# When upgrading or scaling up the following must match whats in your master config! +# Inventory: master yaml field +# osm_host_subnet_length: hostSubnetLength +# When installing osm_host_subnet_length must be set. A sane example is provided below. #osm_host_subnet_length=9 # Configure master API and console ports. diff --git a/playbooks/byo/openshift-loadbalancer/config.yml b/playbooks/byo/openshift-loadbalancer/config.yml new file mode 100644 index 000000000..32c828f97 --- /dev/null +++ b/playbooks/byo/openshift-loadbalancer/config.yml @@ -0,0 +1,6 @@ +--- +- include: ../openshift-cluster/initialize_groups.yml + +- include: ../../common/openshift-cluster/std_include.yml + +- include: ../../common/openshift-loadbalancer/config.yml diff --git a/playbooks/byo/openshift-nfs/config.yml b/playbooks/byo/openshift-nfs/config.yml new file mode 100644 index 000000000..93b24411e --- /dev/null +++ b/playbooks/byo/openshift-nfs/config.yml @@ -0,0 +1,6 @@ +--- +- include: ../openshift-cluster/initialize_groups.yml + +- include: ../../common/openshift-cluster/std_include.yml + +- include: ../../common/openshift-nfs/config.yml diff --git a/playbooks/common/openshift-cluster/evaluate_groups.yml b/playbooks/common/openshift-cluster/evaluate_groups.yml index 16a733899..e55b2f964 100644 --- a/playbooks/common/openshift-cluster/evaluate_groups.yml +++ b/playbooks/common/openshift-cluster/evaluate_groups.yml @@ -43,11 +43,14 @@ - name: Evaluate groups - Fail if no etcd hosts group is defined fail: msg: > - No etcd hosts defined. Running an all-in-one master is deprecated and - will no longer be supported in a future upgrade. + Running etcd as an embedded service is no longer supported. If this is a + new install please define an 'etcd' group with either one or three + hosts. These hosts may be the same hosts as your masters. If this is an + upgrade you may set openshift_master_unsupported_embedded_etcd=true + until a migration playbook becomes available. when: - - g_etcd_hosts | default([]) | length == 0 - - not openshift_master_unsupported_all_in_one | default(False) + - g_etcd_hosts | default([]) | length not in [3,1] + - not openshift_master_unsupported_embedded_etcd | default(False) - not openshift_node_bootstrap | default(False) - name: Evaluate oo_all_hosts diff --git a/playbooks/common/openshift-cluster/upgrades/pre/verify_inventory_vars.yml b/playbooks/common/openshift-cluster/upgrades/pre/verify_inventory_vars.yml index 4c345dbe8..9a959a959 100644 --- a/playbooks/common/openshift-cluster/upgrades/pre/verify_inventory_vars.yml +++ b/playbooks/common/openshift-cluster/upgrades/pre/verify_inventory_vars.yml @@ -9,21 +9,6 @@ deployment types when: deployment_type not in ['origin','openshift-enterprise', 'online'] - # osm_cluster_network_cidr, osm_host_subnet_length and openshift_portal_net are - # required when upgrading to avoid changes that may occur between releases - # Reference: https://bugzilla.redhat.com/show_bug.cgi?id=1451023 - - assert: - that: - - "osm_cluster_network_cidr is defined" - - "osm_host_subnet_length is defined" - - "openshift_portal_net is defined" - msg: > - osm_cluster_network_cidr, osm_host_subnet_length, and openshift_portal_net are required inventory - variables when upgrading. These variables should match what is currently used in the cluster. If - you don't remember what these values are you can find them in /etc/origin/master/master-config.yaml - on a master with the names clusterNetworkCIDR (osm_cluster_network_cidr), - hostSubnetLength (osm_host_subnet_length), and serviceNetworkCIDR (openshift_portal_net). - # Error out in situations where the user has older versions specified in their # inventory in any of the openshift_release, openshift_image_tag, and # openshift_pkg_version variables. These must be removed or updated to proceed diff --git a/playbooks/common/openshift-cluster/upgrades/upgrade_control_plane.yml b/playbooks/common/openshift-cluster/upgrades/upgrade_control_plane.yml index 18f10437d..b75aae589 100644 --- a/playbooks/common/openshift-cluster/upgrades/upgrade_control_plane.yml +++ b/playbooks/common/openshift-cluster/upgrades/upgrade_control_plane.yml @@ -13,11 +13,11 @@ {{ openshift.common.client_binary }} adm --config={{ openshift.common.config_base }}/master/admin.kubeconfig migrate storage --include=* --confirm register: l_pb_upgrade_control_plane_pre_upgrade_storage - when: openshift_upgrade_pre_storage_migration_enabled | default(true,true) | bool + when: openshift_upgrade_pre_storage_migration_enabled | default(true) | bool failed_when: - - openshift_upgrade_pre_storage_migration_enabled | default(true,true) | bool + - openshift_upgrade_pre_storage_migration_enabled | default(true) | bool - l_pb_upgrade_control_plane_pre_upgrade_storage.rc != 0 - - openshift_upgrade_pre_storage_migration_fatal | default(true,true) | bool + - openshift_upgrade_pre_storage_migration_fatal | default(true) | bool # If facts cache were for some reason deleted, this fact may not be set, and if not set # it will always default to true. This causes problems for the etcd data dir fact detection @@ -151,11 +151,11 @@ {{ openshift.common.client_binary }} adm --config={{ openshift.common.config_base }}/master/admin.kubeconfig migrate storage --include=clusterpolicies --confirm register: l_pb_upgrade_control_plane_post_upgrade_storage - when: openshift_upgrade_post_storage_migration_enabled | default(true,true) | bool + when: openshift_upgrade_post_storage_migration_enabled | default(true) | bool failed_when: - - openshift_upgrade_post_storage_migration_enabled | default(true,true) | bool + - openshift_upgrade_post_storage_migration_enabled | default(true) | bool - l_pb_upgrade_control_plane_post_upgrade_storage.rc != 0 - - openshift_upgrade_post_storage_migration_fatal | default(false,true) | bool + - openshift_upgrade_post_storage_migration_fatal | default(false) | bool run_once: true delegate_to: "{{ groups.oo_first_master.0 }}" @@ -247,11 +247,11 @@ migrate storage --include=* --confirm run_once: true register: l_pb_upgrade_control_plane_post_upgrade_storage - when: openshift_upgrade_post_storage_migration_enabled | default(true,true) | bool + when: openshift_upgrade_post_storage_migration_enabled | default(true) | bool failed_when: - - openshift_upgrade_post_storage_migration_enabled | default(true,true) | bool + - openshift_upgrade_post_storage_migration_enabled | default(true) | bool - l_pb_upgrade_control_plane_post_upgrade_storage.rc != 0 - - openshift_upgrade_post_storage_migration_fatal | default(false,true) | bool + - openshift_upgrade_post_storage_migration_fatal | default(false) | bool - set_fact: reconcile_complete: True diff --git a/playbooks/common/openshift-etcd/migrate.yml b/playbooks/common/openshift-etcd/migrate.yml index a2af7bb21..e4ab0aa41 100644 --- a/playbooks/common/openshift-etcd/migrate.yml +++ b/playbooks/common/openshift-etcd/migrate.yml @@ -69,7 +69,7 @@ - role: etcd_migrate r_etcd_migrate_action: migrate r_etcd_common_embedded_etcd: "{{ groups.oo_etcd_to_config | default([]) | length == 0 }}" - etcd_peer: "{{ ansible_default_ipv4.address }}" + etcd_peer: "{{ openshift.common.ip }}" etcd_url_scheme: "https" etcd_peer_url_scheme: "https" @@ -80,7 +80,7 @@ - role: etcd_migrate r_etcd_migrate_action: clean_data r_etcd_common_embedded_etcd: "{{ groups.oo_etcd_to_config | default([]) | length == 0 }}" - etcd_peer: "{{ ansible_default_ipv4.address }}" + etcd_peer: "{{ openshift.common.ip }}" etcd_url_scheme: "https" etcd_peer_url_scheme: "https" post_tasks: @@ -115,7 +115,7 @@ roles: - role: etcd_migrate r_etcd_migrate_action: add_ttls - etcd_peer: "{{ hostvars[groups.oo_etcd_to_migrate.0].ansible_default_ipv4.address }}" + etcd_peer: "{{ hostvars[groups.oo_etcd_to_migrate.0].openshift.common.ip }}" etcd_url_scheme: "https" etcd_peer_url_scheme: "https" when: etcd_migration_failed | length == 0 diff --git a/playbooks/common/openshift-etcd/scaleup.yml b/playbooks/common/openshift-etcd/scaleup.yml index 5f8bb1c7a..d3fa48bad 100644 --- a/playbooks/common/openshift-etcd/scaleup.yml +++ b/playbooks/common/openshift-etcd/scaleup.yml @@ -23,6 +23,9 @@ -C {{ etcd_peer_url_scheme }}://{{ hostvars[etcd_ca_host].etcd_hostname }}:{{ etcd_client_port }} member add {{ etcd_hostname }} {{ etcd_peer_url_scheme }}://{{ etcd_ip }}:{{ etcd_peer_port }} delegate_to: "{{ etcd_ca_host }}" + failed_when: + - etcd_add_check.rc == 1 + - ("peerURL exists" not in etcd_add_check.stderr) register: etcd_add_check retries: 3 delay: 10 @@ -53,3 +56,19 @@ retries: 3 delay: 30 until: scaleup_health.rc == 0 + +- name: Update master etcd client urls + hosts: oo_masters_to_config + serial: 1 + tasks: + - include_role: + name: openshift_master + tasks_from: update_etcd_client_urls + vars: + etcd_ca_host: "{{ groups.oo_etcd_to_config.0 }}" + openshift_ca_host: "{{ groups.oo_first_master.0 }}" + openshift_master_etcd_hosts: "{{ hostvars + | oo_select_keys(groups['oo_etcd_to_config'] | union(groups['oo_new_etcd_to_config'])) + | oo_collect('openshift.common.hostname') + | default(none, true) }}" + openshift_master_etcd_port: "{{ (etcd_client_port | default('2379')) if (groups.oo_etcd_to_config is defined and groups.oo_etcd_to_config) else none }}" diff --git a/playbooks/common/openshift-master/config.yml b/playbooks/common/openshift-master/config.yml index c77d7bb87..e1b9a4964 100644 --- a/playbooks/common/openshift-master/config.yml +++ b/playbooks/common/openshift-master/config.yml @@ -179,28 +179,36 @@ openshift_master_count: "{{ openshift.master.master_count }}" openshift_master_session_auth_secrets: "{{ hostvars[groups.oo_first_master.0].openshift.master.session_auth_secrets }}" openshift_master_session_encryption_secrets: "{{ hostvars[groups.oo_first_master.0].openshift.master.session_encryption_secrets }}" - openshift_no_proxy_internal_hostnames: "{{ hostvars | oo_select_keys(groups['oo_nodes_to_config'] - | union(groups['oo_masters_to_config']) - | union(groups['oo_etcd_to_config'] | default([]))) - | oo_collect('openshift.common.hostname') | default([]) | join (',') - }}" - openshift_no_proxy_etcd_host_ips: "{{ hostvars | oo_select_keys(groups['oo_etcd_to_config'] | default([])) - | oo_collect('openshift.common.ip') | default([]) | join(',') - }}" - roles: - - role: os_firewall - - role: openshift_master openshift_ca_host: "{{ groups.oo_first_master.0 }}" openshift_master_etcd_hosts: "{{ hostvars | oo_select_keys(groups['oo_etcd_to_config'] | default([])) | oo_collect('openshift.common.hostname') | default(none, true) }}" - openshift_master_hosts: "{{ groups.oo_masters_to_config }}" - r_etcd_common_etcd_runtime: "{{ openshift.common.etcd_runtime }}" - etcd_ca_host: "{{ groups.oo_etcd_to_config.0 }}" + openshift_no_proxy_etcd_host_ips: "{{ hostvars | oo_select_keys(groups['oo_etcd_to_config'] | default([])) + | oo_collect('openshift.common.ip') | default([]) | join(',') + }}" + roles: + - role: os_firewall + - role: openshift_master_facts + - role: openshift_hosted_facts + - role: openshift_master_certificates + - role: openshift_etcd_client_certificates etcd_cert_subdir: "openshift-master-{{ openshift.common.hostname }}" etcd_cert_config_dir: "{{ openshift.common.config_base }}/master" etcd_cert_prefix: "master.etcd-" + r_etcd_common_etcd_runtime: "{{ openshift.common.etcd_runtime }}" + etcd_ca_host: "{{ groups.oo_etcd_to_config.0 }}" + when: groups.oo_etcd_to_config | default([]) | length != 0 + - role: openshift_clock + - role: openshift_cloud_provider + - role: openshift_builddefaults + - role: openshift_buildoverrides + - role: nickhammond.logrotate + - role: contiv + contiv_role: netmaster + when: openshift_use_contiv | default(False) | bool + - role: openshift_master + openshift_master_hosts: "{{ groups.oo_masters_to_config }}" r_openshift_master_clean_install: "{{ hostvars[groups.oo_first_master.0].l_clean_install }}" r_openshift_master_etcd3_storage: "{{ hostvars[groups.oo_first_master.0].l_etcd3_enabled }}" openshift_master_is_scaleup_host: "{{ g_openshift_master_is_scaleup | default(false) }}" @@ -211,7 +219,6 @@ when: openshift_use_nuage | default(false) | bool - role: calico_master when: openshift_use_calico | default(false) | bool - post_tasks: - name: Create group for deployment type group_by: key=oo_masters_deployment_type_{{ openshift.common.deployment_type }} diff --git a/playbooks/common/openshift-nfs/config.yml b/playbooks/common/openshift-nfs/config.yml index 000e46e80..64ea0d3c4 100644 --- a/playbooks/common/openshift-nfs/config.yml +++ b/playbooks/common/openshift-nfs/config.yml @@ -2,5 +2,5 @@ - name: Configure nfs hosts: oo_nfs_to_config roles: - - role: openshift_facts + - role: os_firewall - role: openshift_storage_nfs diff --git a/playbooks/gcp/openshift-cluster/provision.yml b/playbooks/gcp/openshift-cluster/provision.yml new file mode 100644 index 000000000..a3d1d46a6 --- /dev/null +++ b/playbooks/gcp/openshift-cluster/provision.yml @@ -0,0 +1,19 @@ +--- +- name: Ensure all cloud resources necessary for the cluster, including instances, have been started + hosts: localhost + connection: local + gather_facts: no + tasks: + + - name: provision a GCP cluster in the specified project + include_role: + name: openshift_gcp + +- name: normalize groups + include: ../../byo/openshift-cluster/initialize_groups.yml + +- name: run the std_include + include: ../../common/openshift-cluster/std_include.yml + +- name: run the config + include: ../../common/openshift-cluster/config.yml diff --git a/roles/docker/tasks/package_docker.yml b/roles/docker/tasks/package_docker.yml index 145b552a6..d685d77f2 100644 --- a/roles/docker/tasks/package_docker.yml +++ b/roles/docker/tasks/package_docker.yml @@ -3,6 +3,8 @@ command: "{{ repoquery_cmd }} --installed --qf '%{version}' docker" when: not openshift.common.is_atomic | bool register: curr_docker_version + retries: 4 + until: not curr_docker_version | failed changed_when: false - name: Error out if Docker pre-installed but too old diff --git a/roles/etcd_common/tasks/backup.yml b/roles/etcd_common/tasks/backup.yml index c1580640f..42d27c081 100644 --- a/roles/etcd_common/tasks/backup.yml +++ b/roles/etcd_common/tasks/backup.yml @@ -36,7 +36,7 @@ - name: Abort if insufficient disk space for etcd backup fail: msg: > - {{ l_etcd_disk_usage.stdout }} Kb disk space required for etcd backup, + {{ l_etcd_disk_usage.stdout|int*2 }} Kb disk space required for etcd backup, {{ l_avail_disk.stdout }} Kb available. when: l_etcd_disk_usage.stdout|int*2 > l_avail_disk.stdout|int diff --git a/roles/lib_utils/library/repoquery.py b/roles/lib_utils/library/repoquery.py index 95a305b58..e5ac1f74f 100644 --- a/roles/lib_utils/library/repoquery.py +++ b/roles/lib_utils/library/repoquery.py @@ -35,6 +35,7 @@ import os # noqa: F401 import re # noqa: F401 import shutil # noqa: F401 import tempfile # noqa: F401 +import time # noqa: F401 try: import ruamel.yaml as yaml # noqa: F401 @@ -618,17 +619,22 @@ def main(): show_duplicates=dict(default=False, required=False, type='bool'), match_version=dict(default=None, required=False, type='str'), ignore_excluders=dict(default=False, required=False, type='bool'), + retries=dict(default=4, required=False, type='int'), + retry_interval=dict(default=5, required=False, type='int'), ), supports_check_mode=False, required_if=[('show_duplicates', True, ['name'])], ) - rval = Repoquery.run_ansible(module.params, module.check_mode) - - if 'failed' in rval: - module.fail_json(**rval) - - module.exit_json(**rval) + tries = 1 + while True: + rval = Repoquery.run_ansible(module.params, module.check_mode) + if 'failed' not in rval: + module.exit_json(**rval) + elif tries > module.params['retries']: + module.fail_json(**rval) + tries += 1 + time.sleep(module.params['retry_interval']) if __name__ == "__main__": diff --git a/roles/lib_utils/library/yedit.py b/roles/lib_utils/library/yedit.py index baf72fe47..921bca074 100644 --- a/roles/lib_utils/library/yedit.py +++ b/roles/lib_utils/library/yedit.py @@ -35,6 +35,7 @@ import os # noqa: F401 import re # noqa: F401 import shutil # noqa: F401 import tempfile # noqa: F401 +import time # noqa: F401 try: import ruamel.yaml as yaml # noqa: F401 diff --git a/roles/lib_utils/src/ansible/repoquery.py b/roles/lib_utils/src/ansible/repoquery.py index 40773b1c1..5f5b93639 100644 --- a/roles/lib_utils/src/ansible/repoquery.py +++ b/roles/lib_utils/src/ansible/repoquery.py @@ -19,17 +19,22 @@ def main(): show_duplicates=dict(default=False, required=False, type='bool'), match_version=dict(default=None, required=False, type='str'), ignore_excluders=dict(default=False, required=False, type='bool'), + retries=dict(default=4, required=False, type='int'), + retry_interval=dict(default=5, required=False, type='int'), ), supports_check_mode=False, required_if=[('show_duplicates', True, ['name'])], ) - rval = Repoquery.run_ansible(module.params, module.check_mode) - - if 'failed' in rval: - module.fail_json(**rval) - - module.exit_json(**rval) + tries = 1 + while True: + rval = Repoquery.run_ansible(module.params, module.check_mode) + if 'failed' not in rval: + module.exit_json(**rval) + elif tries > module.params['retries']: + module.fail_json(**rval) + tries += 1 + time.sleep(module.params['retry_interval']) if __name__ == "__main__": diff --git a/roles/lib_utils/src/lib/import.py b/roles/lib_utils/src/lib/import.py index 567f8c9e0..07a04b7ae 100644 --- a/roles/lib_utils/src/lib/import.py +++ b/roles/lib_utils/src/lib/import.py @@ -10,6 +10,7 @@ import os # noqa: F401 import re # noqa: F401 import shutil # noqa: F401 import tempfile # noqa: F401 +import time # noqa: F401 try: import ruamel.yaml as yaml # noqa: F401 diff --git a/roles/openshift_facts/library/openshift_facts.py b/roles/openshift_facts/library/openshift_facts.py index ebfa6bb8f..517e0231d 100755 --- a/roles/openshift_facts/library/openshift_facts.py +++ b/roles/openshift_facts/library/openshift_facts.py @@ -1602,11 +1602,13 @@ def set_builddefaults_facts(facts): builddefaults['git_no_proxy'] = builddefaults['no_proxy'] # If we're actually defining a builddefaults config then create admission_plugin_config # then merge builddefaults[config] structure into admission_plugin_config + + # 'config' is the 'openshift_builddefaults_json' inventory variable if 'config' in builddefaults: if 'admission_plugin_config' not in facts['master']: - facts['master']['admission_plugin_config'] = dict() + # Scaffold out the full expected datastructure + facts['master']['admission_plugin_config'] = {'BuildDefaults': {'configuration': {'env': {}}}} facts['master']['admission_plugin_config'].update(builddefaults['config']) - # if the user didn't actually provide proxy values, delete the proxy env variable defaults. delete_empty_keys(facts['master']['admission_plugin_config']['BuildDefaults']['configuration']['env']) return facts diff --git a/roles/openshift_gcp/tasks/main.yaml b/roles/openshift_gcp/tasks/main.yaml new file mode 100644 index 000000000..ad205ba33 --- /dev/null +++ b/roles/openshift_gcp/tasks/main.yaml @@ -0,0 +1,43 @@ +# +# This role relies on gcloud invoked via templated bash in order to +# provide a high performance deployment option. The next logical step +# is to transition to a deployment manager template which is then instantiated. +# TODO: use a formal set of role parameters consistent with openshift_aws +# +--- +- name: Templatize DNS script + template: src=dns.j2.sh dest=/tmp/openshift_gcp_provision_dns.sh mode=u+rx +- name: Templatize provision script + template: src=provision.j2.sh dest=/tmp/openshift_gcp_provision.sh mode=u+rx +- name: Templatize de-provision script + template: src=remove.j2.sh dest=/tmp/openshift_gcp_provision_remove.sh mode=u+rx + when: + - state | default('present') == 'absent' + +- name: Provision GCP DNS domain + command: /tmp/openshift_gcp_provision_dns.sh + args: + chdir: "{{ playbook_dir }}/files" + register: dns_provision + when: + - state | default('present') == 'present' + +- name: Ensure that DNS resolves to the hosted zone + assert: + that: + - "lookup('dig', public_hosted_zone, 'qtype=NS', wantlist=True) | sort | join(',') == dns_provision.stdout" + msg: "The DNS domain {{ public_hosted_zone }} defined in 'public_hosted_zone' must have NS records pointing to the Google nameservers: '{{ dns_provision.stdout }}' instead of '{{ lookup('dig', public_hosted_zone, 'qtype=NS') }}'." + when: + - state | default('present') == 'present' + +- name: Provision GCP resources + command: /tmp/openshift_gcp_provision.sh + args: + chdir: "{{ playbook_dir }}/files" + when: + - state | default('present') == 'present' + +- name: De-provision GCP resources + command: /tmp/openshift_gcp_provision_remove.sh + when: + - state | default('present') == 'absent' diff --git a/roles/openshift_gcp/templates/dns.j2.sh b/roles/openshift_gcp/templates/dns.j2.sh new file mode 100644 index 000000000..eacf84b4d --- /dev/null +++ b/roles/openshift_gcp/templates/dns.j2.sh @@ -0,0 +1,13 @@ +#!/bin/bash + +set -euo pipefail + +dns_zone="{{ dns_managed_zone | default(provision_prefix + 'managed-zone') }}" + +# Check the DNS managed zone in Google Cloud DNS, create it if it doesn't exist +if ! gcloud --project "{{ gce_project_id }}" dns managed-zones describe "${dns_zone}" &>/dev/null; then + gcloud --project "{{ gce_project_id }}" dns managed-zones create "${dns_zone}" --dns-name "{{ public_hosted_zone }}" --description "{{ public_hosted_zone }} domain" >/dev/null +fi + +# Always output the expected nameservers as a comma delimited list +gcloud --project "{{ gce_project_id }}" dns managed-zones describe "${dns_zone}" --format='value(nameServers)' | tr ';' ',' diff --git a/roles/openshift_gcp/templates/provision.j2.sh b/roles/openshift_gcp/templates/provision.j2.sh new file mode 100644 index 000000000..e68e9683f --- /dev/null +++ b/roles/openshift_gcp/templates/provision.j2.sh @@ -0,0 +1,318 @@ +#!/bin/bash + +set -euo pipefail + +# Create SSH key for GCE +if [ ! -f "{{ gce_ssh_private_key }}" ]; then + ssh-keygen -t rsa -f "{{ gce_ssh_private_key }}" -C gce-provision-cloud-user -N '' + ssh-add "{{ gce_ssh_private_key }}" || true +fi + +# Check if the ~/.ssh/google_compute_engine.pub key is in the project metadata, and if not, add it there +pub_key=$(cut -d ' ' -f 2 < "{{ gce_ssh_private_key }}.pub") +key_tmp_file='/tmp/ocp-gce-keys' +if ! gcloud --project "{{ gce_project_id }}" compute project-info describe | grep -q "$pub_key"; then + if gcloud --project "{{ gce_project_id }}" compute project-info describe | grep -q ssh-rsa; then + gcloud --project "{{ gce_project_id }}" compute project-info describe | grep ssh-rsa | sed -e 's/^[[:space:]]*//' -e 's/[[:space:]]*$//' -e 's/value: //' > "$key_tmp_file" + fi + echo -n 'cloud-user:' >> "$key_tmp_file" + cat "{{ gce_ssh_private_key }}.pub" >> "$key_tmp_file" + gcloud --project "{{ gce_project_id }}" compute project-info add-metadata --metadata-from-file "sshKeys=${key_tmp_file}" + rm -f "$key_tmp_file" +fi + +metadata="" +if [[ -n "{{ provision_gce_startup_script_file }}" ]]; then + if [[ ! -f "{{ provision_gce_startup_script_file }}" ]]; then + echo "Startup script file missing at {{ provision_gce_startup_script_file }} from=$(pwd)" + exit 1 + fi + metadata+="--metadata-from-file=startup-script={{ provision_gce_startup_script_file }}" +fi +if [[ -n "{{ provision_gce_user_data_file }}" ]]; then + if [[ ! -f "{{ provision_gce_user_data_file }}" ]]; then + echo "User data file missing at {{ provision_gce_user_data_file }}" + exit 1 + fi + if [[ -n "${metadata}" ]]; then + metadata+="," + else + metadata="--metadata-from-file=" + fi + metadata+="user-data={{ provision_gce_user_data_file }}" +fi + +# Select image or image family +image="{{ provision_gce_registered_image }}" +if ! gcloud --project "{{ gce_project_id }}" compute images describe "${image}" &>/dev/null; then + if ! gcloud --project "{{ gce_project_id }}" compute images describe-from-family "${image}" &>/dev/null; then + echo "No compute image or image-family found, create an image named '{{ provision_gce_registered_image }}' to continue'" + exit 1 + fi + image="family/${image}" +fi + +### PROVISION THE INFRASTRUCTURE ### + +dns_zone="{{ dns_managed_zone | default(provision_prefix + 'managed-zone') }}" + +# Check the DNS managed zone in Google Cloud DNS, create it if it doesn't exist and exit after printing NS servers +if ! gcloud --project "{{ gce_project_id }}" dns managed-zones describe "${dns_zone}" &>/dev/null; then + echo "DNS zone '${dns_zone}' doesn't exist. Must be configured prior to running this script" + exit 1 +fi + +# Create network +if ! gcloud --project "{{ gce_project_id }}" compute networks describe "{{ gce_network_name }}" &>/dev/null; then + gcloud --project "{{ gce_project_id }}" compute networks create "{{ gce_network_name }}" --mode "auto" +else + echo "Network '{{ gce_network_name }}' already exists" +fi + +# Firewall rules in a form: +# ['name']='parameters for "gcloud compute firewall-rules create"' +# For all possible parameters see: gcloud compute firewall-rules create --help +range="" +if [[ -n "{{ openshift_node_port_range }}" ]]; then + range=",tcp:{{ openshift_node_port_range }},udp:{{ openshift_node_port_range }}" +fi +declare -A FW_RULES=( + ['icmp']='--allow icmp' + ['ssh-external']='--allow tcp:22' + ['ssh-internal']='--allow tcp:22 --source-tags bastion' + ['master-internal']="--allow tcp:2224,tcp:2379,tcp:2380,tcp:4001,udp:4789,udp:5404,udp:5405,tcp:8053,udp:8053,tcp:8444,tcp:10250,tcp:10255,udp:10255,tcp:24224,udp:24224 --source-tags ocp --target-tags ocp-master" + ['master-external']="--allow tcp:80,tcp:443,tcp:1936,tcp:8080,tcp:8443${range} --target-tags ocp-master" + ['node-internal']="--allow udp:4789,tcp:10250,tcp:10255,udp:10255 --source-tags ocp --target-tags ocp-node,ocp-infra-node" + ['infra-node-internal']="--allow tcp:5000 --source-tags ocp --target-tags ocp-infra-node" + ['infra-node-external']="--allow tcp:80,tcp:443,tcp:1936${range} --target-tags ocp-infra-node" +) +for rule in "${!FW_RULES[@]}"; do + ( if ! gcloud --project "{{ gce_project_id }}" compute firewall-rules describe "{{ provision_prefix }}$rule" &>/dev/null; then + gcloud --project "{{ gce_project_id }}" compute firewall-rules create "{{ provision_prefix }}$rule" --network "{{ gce_network_name }}" ${FW_RULES[$rule]} + else + echo "Firewall rule '{{ provision_prefix }}${rule}' already exists" + fi ) & +done + + +# Master IP +( if ! gcloud --project "{{ gce_project_id }}" compute addresses describe "{{ provision_prefix }}master-ssl-lb-ip" --global &>/dev/null; then + gcloud --project "{{ gce_project_id }}" compute addresses create "{{ provision_prefix }}master-ssl-lb-ip" --global +else + echo "IP '{{ provision_prefix }}master-ssl-lb-ip' already exists" +fi ) & + +# Internal master IP +( if ! gcloud --project "{{ gce_project_id }}" compute addresses describe "{{ provision_prefix }}master-network-lb-ip" --region "{{ gce_region_name }}" &>/dev/null; then + gcloud --project "{{ gce_project_id }}" compute addresses create "{{ provision_prefix }}master-network-lb-ip" --region "{{ gce_region_name }}" +else + echo "IP '{{ provision_prefix }}master-network-lb-ip' already exists" +fi ) & + +# Router IP +( if ! gcloud --project "{{ gce_project_id }}" compute addresses describe "{{ provision_prefix }}router-network-lb-ip" --region "{{ gce_region_name }}" &>/dev/null; then + gcloud --project "{{ gce_project_id }}" compute addresses create "{{ provision_prefix }}router-network-lb-ip" --region "{{ gce_region_name }}" +else + echo "IP '{{ provision_prefix }}router-network-lb-ip' already exists" +fi ) & + + +{% for node_group in provision_gce_node_groups %} +# configure {{ node_group.name }} +( + if ! gcloud --project "{{ gce_project_id }}" compute instance-templates describe "{{ provision_prefix }}instance-template-{{ node_group.name }}" &>/dev/null; then + gcloud --project "{{ gce_project_id }}" compute instance-templates create "{{ provision_prefix }}instance-template-{{ node_group.name }}" \ + --machine-type "{{ node_group.machine_type }}" --network "{{ gce_network_name }}" \ + --tags "{{ provision_prefix }}ocp,ocp,{{ node_group.tags }}" \ + --boot-disk-size "{{ node_group.boot_disk_size }}" --boot-disk-type "pd-ssd" \ + --scopes "logging-write,monitoring-write,useraccounts-ro,service-control,service-management,storage-ro,compute-rw" \ + --image "${image}" ${metadata} + else + echo "Instance template '{{ provision_prefix }}instance-template-{{ node_group.name }}' already exists" + fi + + # Create instance group + if ! gcloud --project "{{ gce_project_id }}" compute instance-groups managed describe "{{ provision_prefix }}ig-{{ node_group.suffix }}" --zone "{{ gce_zone_name }}" &>/dev/null; then + gcloud --project "{{ gce_project_id }}" compute instance-groups managed create "{{ provision_prefix }}ig-{{ node_group.suffix }}" \ + --zone "{{ gce_zone_name }}" --template "{{ provision_prefix }}instance-template-{{ node_group.name }}" --size "{{ node_group.scale }}" + else + echo "Instance group '{{ provision_prefix }}ig-{{ node_group.suffix }}' already exists" + fi +) & +{% endfor %} + +for i in `jobs -p`; do wait $i; done + + +# Configure the master external LB rules +( +# Master health check +if ! gcloud --project "{{ gce_project_id }}" compute health-checks describe "{{ provision_prefix }}master-ssl-lb-health-check" &>/dev/null; then + gcloud --project "{{ gce_project_id }}" compute health-checks create https "{{ provision_prefix }}master-ssl-lb-health-check" --port "{{ internal_console_port }}" --request-path "/healthz" +else + echo "Health check '{{ provision_prefix }}master-ssl-lb-health-check' already exists" +fi + +gcloud --project "{{ gce_project_id }}" compute instance-groups managed set-named-ports "{{ provision_prefix }}ig-m" \ + --zone "{{ gce_zone_name }}" --named-ports "{{ provision_prefix }}port-name-master:{{ internal_console_port }}" + +# Master backend service +if ! gcloud --project "{{ gce_project_id }}" compute backend-services describe "{{ provision_prefix }}master-ssl-lb-backend" --global &>/dev/null; then + gcloud --project "{{ gce_project_id }}" compute backend-services create "{{ provision_prefix }}master-ssl-lb-backend" --health-checks "{{ provision_prefix }}master-ssl-lb-health-check" --port-name "{{ provision_prefix }}port-name-master" --protocol "TCP" --global --timeout="{{ provision_gce_master_https_timeout | default('2m') }}" + gcloud --project "{{ gce_project_id }}" compute backend-services add-backend "{{ provision_prefix }}master-ssl-lb-backend" --instance-group "{{ provision_prefix }}ig-m" --global --instance-group-zone "{{ gce_zone_name }}" +else + echo "Backend service '{{ provision_prefix }}master-ssl-lb-backend' already exists" +fi + +# Master tcp proxy target +if ! gcloud --project "{{ gce_project_id }}" compute target-tcp-proxies describe "{{ provision_prefix }}master-ssl-lb-target" &>/dev/null; then + gcloud --project "{{ gce_project_id }}" compute target-tcp-proxies create "{{ provision_prefix }}master-ssl-lb-target" --backend-service "{{ provision_prefix }}master-ssl-lb-backend" +else + echo "Proxy target '{{ provision_prefix }}master-ssl-lb-target' already exists" +fi + +# Master forwarding rule +if ! gcloud --project "{{ gce_project_id }}" compute forwarding-rules describe "{{ provision_prefix }}master-ssl-lb-rule" --global &>/dev/null; then + IP=$(gcloud --project "{{ gce_project_id }}" compute addresses describe "{{ provision_prefix }}master-ssl-lb-ip" --global --format='value(address)') + gcloud --project "{{ gce_project_id }}" compute forwarding-rules create "{{ provision_prefix }}master-ssl-lb-rule" --address "$IP" --global --ports "{{ console_port }}" --target-tcp-proxy "{{ provision_prefix }}master-ssl-lb-target" +else + echo "Forwarding rule '{{ provision_prefix }}master-ssl-lb-rule' already exists" +fi +) & + + +# Configure the master internal LB rules +( +# Internal master health check +if ! gcloud --project "{{ gce_project_id }}" compute http-health-checks describe "{{ provision_prefix }}master-network-lb-health-check" &>/dev/null; then + gcloud --project "{{ gce_project_id }}" compute http-health-checks create "{{ provision_prefix }}master-network-lb-health-check" --port "8080" --request-path "/healthz" +else + echo "Health check '{{ provision_prefix }}master-network-lb-health-check' already exists" +fi + +# Internal master target pool +if ! gcloud --project "{{ gce_project_id }}" compute target-pools describe "{{ provision_prefix }}master-network-lb-pool" --region "{{ gce_region_name }}" &>/dev/null; then + gcloud --project "{{ gce_project_id }}" compute target-pools create "{{ provision_prefix }}master-network-lb-pool" --http-health-check "{{ provision_prefix }}master-network-lb-health-check" --region "{{ gce_region_name }}" +else + echo "Target pool '{{ provision_prefix }}master-network-lb-pool' already exists" +fi + +# Internal master forwarding rule +if ! gcloud --project "{{ gce_project_id }}" compute forwarding-rules describe "{{ provision_prefix }}master-network-lb-rule" --region "{{ gce_region_name }}" &>/dev/null; then + IP=$(gcloud --project "{{ gce_project_id }}" compute addresses describe "{{ provision_prefix }}master-network-lb-ip" --region "{{ gce_region_name }}" --format='value(address)') + gcloud --project "{{ gce_project_id }}" compute forwarding-rules create "{{ provision_prefix }}master-network-lb-rule" --address "$IP" --region "{{ gce_region_name }}" --target-pool "{{ provision_prefix }}master-network-lb-pool" +else + echo "Forwarding rule '{{ provision_prefix }}master-network-lb-rule' already exists" +fi +) & + + +# Configure the infra node rules +( +# Router health check +if ! gcloud --project "{{ gce_project_id }}" compute http-health-checks describe "{{ provision_prefix }}router-network-lb-health-check" &>/dev/null; then + gcloud --project "{{ gce_project_id }}" compute http-health-checks create "{{ provision_prefix }}router-network-lb-health-check" --port "1936" --request-path "/healthz" +else + echo "Health check '{{ provision_prefix }}router-network-lb-health-check' already exists" +fi + +# Router target pool +if ! gcloud --project "{{ gce_project_id }}" compute target-pools describe "{{ provision_prefix }}router-network-lb-pool" --region "{{ gce_region_name }}" &>/dev/null; then + gcloud --project "{{ gce_project_id }}" compute target-pools create "{{ provision_prefix }}router-network-lb-pool" --http-health-check "{{ provision_prefix }}router-network-lb-health-check" --region "{{ gce_region_name }}" +else + echo "Target pool '{{ provision_prefix }}router-network-lb-pool' already exists" +fi + +# Router forwarding rule +if ! gcloud --project "{{ gce_project_id }}" compute forwarding-rules describe "{{ provision_prefix }}router-network-lb-rule" --region "{{ gce_region_name }}" &>/dev/null; then + IP=$(gcloud --project "{{ gce_project_id }}" compute addresses describe "{{ provision_prefix }}router-network-lb-ip" --region "{{ gce_region_name }}" --format='value(address)') + gcloud --project "{{ gce_project_id }}" compute forwarding-rules create "{{ provision_prefix }}router-network-lb-rule" --address "$IP" --region "{{ gce_region_name }}" --target-pool "{{ provision_prefix }}router-network-lb-pool" +else + echo "Forwarding rule '{{ provision_prefix }}router-network-lb-rule' already exists" +fi +) & + +for i in `jobs -p`; do wait $i; done + +# set the target pools +( +if [[ "ig-m" == "{{ provision_gce_router_network_instance_group }}" ]]; then + gcloud --project "{{ gce_project_id }}" compute instance-groups managed set-target-pools "{{ provision_prefix }}ig-m" --target-pools "{{ provision_prefix }}master-network-lb-pool,{{ provision_prefix }}router-network-lb-pool" --zone "{{ gce_zone_name }}" +else + gcloud --project "{{ gce_project_id }}" compute instance-groups managed set-target-pools "{{ provision_prefix }}ig-m" --target-pools "{{ provision_prefix }}master-network-lb-pool" --zone "{{ gce_zone_name }}" + gcloud --project "{{ gce_project_id }}" compute instance-groups managed set-target-pools "{{ provision_prefix }}{{ provision_gce_router_network_instance_group }}" --target-pools "{{ provision_prefix }}router-network-lb-pool" --zone "{{ gce_zone_name }}" +fi +) & + +# configure DNS +( +# Retry DNS changes until they succeed since this may be a shared resource +while true; do + dns="${TMPDIR:-/tmp}/dns.yaml" + rm -f $dns + + # DNS record for master lb + if ! gcloud --project "{{ gce_project_id }}" dns record-sets list -z "${dns_zone}" --name "{{ openshift_master_cluster_public_hostname }}" 2>/dev/null | grep -q "{{ openshift_master_cluster_public_hostname }}"; then + IP=$(gcloud --project "{{ gce_project_id }}" compute addresses describe "{{ provision_prefix }}master-ssl-lb-ip" --global --format='value(address)') + if [[ ! -f $dns ]]; then + gcloud --project "{{ gce_project_id }}" dns record-sets transaction --transaction-file=$dns start -z "${dns_zone}" + fi + gcloud --project "{{ gce_project_id }}" dns record-sets transaction --transaction-file=$dns add -z "${dns_zone}" --ttl 3600 --name "{{ openshift_master_cluster_public_hostname }}." --type A "$IP" + else + echo "DNS record for '{{ openshift_master_cluster_public_hostname }}' already exists" + fi + + # DNS record for internal master lb + if ! gcloud --project "{{ gce_project_id }}" dns record-sets list -z "${dns_zone}" --name "{{ openshift_master_cluster_hostname }}" 2>/dev/null | grep -q "{{ openshift_master_cluster_hostname }}"; then + IP=$(gcloud --project "{{ gce_project_id }}" compute addresses describe "{{ provision_prefix }}master-network-lb-ip" --region "{{ gce_region_name }}" --format='value(address)') + if [[ ! -f $dns ]]; then + gcloud --project "{{ gce_project_id }}" dns record-sets transaction --transaction-file=$dns start -z "${dns_zone}" + fi + gcloud --project "{{ gce_project_id }}" dns record-sets transaction --transaction-file=$dns add -z "${dns_zone}" --ttl 3600 --name "{{ openshift_master_cluster_hostname }}." --type A "$IP" + else + echo "DNS record for '{{ openshift_master_cluster_hostname }}' already exists" + fi + + # DNS record for router lb + if ! gcloud --project "{{ gce_project_id }}" dns record-sets list -z "${dns_zone}" --name "{{ wildcard_zone }}" 2>/dev/null | grep -q "{{ wildcard_zone }}"; then + IP=$(gcloud --project "{{ gce_project_id }}" compute addresses describe "{{ provision_prefix }}router-network-lb-ip" --region "{{ gce_region_name }}" --format='value(address)') + if [[ ! -f $dns ]]; then + gcloud --project "{{ gce_project_id }}" dns record-sets transaction --transaction-file=$dns start -z "${dns_zone}" + fi + gcloud --project "{{ gce_project_id }}" dns record-sets transaction --transaction-file=$dns add -z "${dns_zone}" --ttl 3600 --name "{{ wildcard_zone }}." --type A "$IP" + gcloud --project "{{ gce_project_id }}" dns record-sets transaction --transaction-file=$dns add -z "${dns_zone}" --ttl 3600 --name "*.{{ wildcard_zone }}." --type CNAME "{{ wildcard_zone }}." + else + echo "DNS record for '{{ wildcard_zone }}' already exists" + fi + + # Commit all DNS changes, retrying if preconditions are not met + if [[ -f $dns ]]; then + if ! out="$( gcloud --project "{{ gce_project_id }}" dns record-sets transaction --transaction-file=$dns execute -z "${dns_zone}" 2>&1 )"; then + rc=$? + if [[ "${out}" == *"HTTPError 412: Precondition not met"* ]]; then + continue + fi + exit $rc + fi + fi + break +done +) & + +# Create bucket for registry +( +if ! gsutil ls -p "{{ gce_project_id }}" "gs://{{ openshift_hosted_registry_storage_gcs_bucket }}" &>/dev/null; then + gsutil mb -p "{{ gce_project_id }}" -l "{{ gce_region_name }}" "gs://{{ openshift_hosted_registry_storage_gcs_bucket }}" +else + echo "Bucket '{{ openshift_hosted_registry_storage_gcs_bucket }}' already exists" +fi +) & + +# wait until all node groups are stable +{% for node_group in provision_gce_node_groups %} +# wait for stable {{ node_group.name }} +( gcloud --project "{{ gce_project_id }}" compute instance-groups managed wait-until-stable "{{ provision_prefix }}ig-{{ node_group.suffix }}" --zone "{{ gce_zone_name }}" --timeout=300) & +{% endfor %} + + +for i in `jobs -p`; do wait $i; done diff --git a/roles/openshift_gcp/templates/remove.j2.sh b/roles/openshift_gcp/templates/remove.j2.sh new file mode 100644 index 000000000..41ceab2b5 --- /dev/null +++ b/roles/openshift_gcp/templates/remove.j2.sh @@ -0,0 +1,156 @@ +#!/bin/bash + +set -euo pipefail + +function teardown_cmd() { + a=( $@ ) + local name=$1 + a=( "${a[@]:1}" ) + local flag=0 + local found= + for i in ${a[@]}; do + if [[ "$i" == "--"* ]]; then + found=true + break + fi + flag=$((flag+1)) + done + if [[ -z "${found}" ]]; then + flag=$((flag+1)) + fi + if gcloud --project "{{ gce_project_id }}" ${a[@]::$flag} describe "${name}" ${a[@]:$flag} &>/dev/null; then + gcloud --project "{{ gce_project_id }}" ${a[@]::$flag} delete -q "${name}" ${a[@]:$flag} + fi +} + +function teardown() { + for i in `seq 1 20`; do + if teardown_cmd $@; then + break + fi + sleep 0.5 + done +} + +# Preemptively spin down the instances +{% for node_group in provision_gce_node_groups %} +# scale down {{ node_group.name }} +( + # performs a delete and scale down as one operation to ensure maximum parallelism + if ! instances=$( gcloud --project "{{ gce_project_id }}" compute instance-groups managed list-instances "{{ provision_prefix }}ig-{{ node_group.suffix }}" --zone "{{ gce_zone_name }}" --format='value[terminator=","](instance)' ); then + exit 0 + fi + instances="${instances%?}" + if [[ -z "${instances}" ]]; then + echo "warning: No instances in {{ node_group.name }}" 1>&2 + exit 0 + fi + if ! gcloud --project "{{ gce_project_id }}" compute instance-groups managed delete-instances "{{ provision_prefix }}ig-{{ node_group.suffix }}" --zone "{{ gce_zone_name }}" --instances "${instances}"; then + echo "warning: Unable to scale down the node group {{ node_group.name }}" 1>&2 + exit 0 + fi +) & +{% endfor %} + +# Bucket for registry +( +if gsutil ls -p "{{ gce_project_id }}" "gs://{{ openshift_hosted_registry_storage_gcs_bucket }}" &>/dev/null; then + gsutil -m rm -r "gs://{{ openshift_hosted_registry_storage_gcs_bucket }}" +fi +) & + +# DNS +( +dns_zone="{{ dns_managed_zone | default(provision_prefix + 'managed-zone') }}" +if gcloud --project "{{ gce_project_id }}" dns managed-zones describe "${dns_zone}" &>/dev/null; then + # Retry DNS changes until they succeed since this may be a shared resource + while true; do + dns="${TMPDIR:-/tmp}/dns.yaml" + rm -f "${dns}" + + # export all dns records that match into a zone format, and turn each line into a set of args for + # record-sets transaction. + gcloud dns record-sets export --project "{{ gce_project_id }}" -z "${dns_zone}" --zone-file-format "${dns}" + if grep -F -e '{{ openshift_master_cluster_hostname }}' -e '{{ openshift_master_cluster_public_hostname }}' -e '{{ wildcard_zone }}' "${dns}" | \ + awk '{ print "--name", $1, "--ttl", $2, "--type", $4, $5; }' > "${dns}.input" + then + rm -f "${dns}" + gcloud --project "{{ gce_project_id }}" dns record-sets transaction --transaction-file=$dns start -z "${dns_zone}" + cat "${dns}.input" | xargs -L1 gcloud --project "{{ gce_project_id }}" dns record-sets transaction --transaction-file="${dns}" remove -z "${dns_zone}" + + # Commit all DNS changes, retrying if preconditions are not met + if ! out="$( gcloud --project "{{ gce_project_id }}" dns record-sets transaction --transaction-file=$dns execute -z "${dns_zone}" 2>&1 )"; then + rc=$? + if [[ "${out}" == *"HTTPError 412: Precondition not met"* ]]; then + continue + fi + exit $rc + fi + fi + rm "${dns}.input" + break + done +fi +) & + +( +# Router network rules +teardown "{{ provision_prefix }}router-network-lb-rule" compute forwarding-rules --region "{{ gce_region_name }}" +teardown "{{ provision_prefix }}router-network-lb-pool" compute target-pools --region "{{ gce_region_name }}" +teardown "{{ provision_prefix }}router-network-lb-health-check" compute http-health-checks +teardown "{{ provision_prefix }}router-network-lb-ip" compute addresses --region "{{ gce_region_name }}" + +# Internal master network rules +teardown "{{ provision_prefix }}master-network-lb-rule" compute forwarding-rules --region "{{ gce_region_name }}" +teardown "{{ provision_prefix }}master-network-lb-pool" compute target-pools --region "{{ gce_region_name }}" +teardown "{{ provision_prefix }}master-network-lb-health-check" compute http-health-checks +teardown "{{ provision_prefix }}master-network-lb-ip" compute addresses --region "{{ gce_region_name }}" +) & + +( +# Master SSL network rules +teardown "{{ provision_prefix }}master-ssl-lb-rule" compute forwarding-rules --global +teardown "{{ provision_prefix }}master-ssl-lb-target" compute target-tcp-proxies +teardown "{{ provision_prefix }}master-ssl-lb-ip" compute addresses --global +teardown "{{ provision_prefix }}master-ssl-lb-backend" compute backend-services --global +teardown "{{ provision_prefix }}master-ssl-lb-health-check" compute health-checks +) & + +#Firewall rules +#['name']='parameters for "gcloud compute firewall-rules create"' +#For all possible parameters see: gcloud compute firewall-rules create --help +declare -A FW_RULES=( + ['icmp']="" + ['ssh-external']="" + ['ssh-internal']="" + ['master-internal']="" + ['master-external']="" + ['node-internal']="" + ['infra-node-internal']="" + ['infra-node-external']="" +) +for rule in "${!FW_RULES[@]}"; do + ( if gcloud --project "{{ gce_project_id }}" compute firewall-rules describe "{{ provision_prefix }}$rule" &>/dev/null; then + # retry a few times because this call can be flaky + for i in `seq 1 3`; do + if gcloud -q --project "{{ gce_project_id }}" compute firewall-rules delete "{{ provision_prefix }}$rule"; then + break + fi + done + fi ) & +done + +for i in `jobs -p`; do wait $i; done + +{% for node_group in provision_gce_node_groups %} +# teardown {{ node_group.name }} - any load balancers referencing these groups must be removed +( + teardown "{{ provision_prefix }}ig-{{ node_group.suffix }}" compute instance-groups managed --zone "{{ gce_zone_name }}" + teardown "{{ provision_prefix }}instance-template-{{ node_group.name }}" compute instance-templates +) & +{% endfor %} + +for i in `jobs -p`; do wait $i; done + +# Network +teardown "{{ gce_network_name }}" compute networks diff --git a/roles/openshift_gcp_image_prep/files/partition.conf b/roles/openshift_gcp_image_prep/files/partition.conf new file mode 100644 index 000000000..b87e5e0b6 --- /dev/null +++ b/roles/openshift_gcp_image_prep/files/partition.conf @@ -0,0 +1,3 @@ +[Service] +ExecStartPost=-/usr/bin/growpart /dev/sda 1 +ExecStartPost=-/sbin/xfs_growfs / diff --git a/roles/openshift_gcp_image_prep/tasks/main.yaml b/roles/openshift_gcp_image_prep/tasks/main.yaml new file mode 100644 index 000000000..fee5ab618 --- /dev/null +++ b/roles/openshift_gcp_image_prep/tasks/main.yaml @@ -0,0 +1,18 @@ +--- +# GCE instances are starting with xfs AND barrier=1, which is only for extfs. +- name: Remove barrier=1 from XFS fstab entries + lineinfile: + path: /etc/fstab + regexp: '^(.+)xfs(.+?),?barrier=1,?(.*?)$' + line: '\1xfs\2 \4' + backrefs: yes + +- name: Ensure the root filesystem has XFS group quota turned on + lineinfile: + path: /boot/grub2/grub.cfg + regexp: '^(.*)linux16 (.*)$' + line: '\1linux16 \2 rootflags=gquota' + backrefs: yes + +- name: Ensure the root partition grows on startup + copy: src=partition.conf dest=/etc/systemd/system/google-instance-setup.service.d/ diff --git a/roles/openshift_health_checker/action_plugins/openshift_health_check.py b/roles/openshift_health_checker/action_plugins/openshift_health_check.py index d02a43655..326176273 100644 --- a/roles/openshift_health_checker/action_plugins/openshift_health_check.py +++ b/roles/openshift_health_checker/action_plugins/openshift_health_check.py @@ -3,7 +3,10 @@ Ansible action plugin to execute health checks in OpenShift clusters. """ import sys import os +import base64 import traceback +import errno +import json from collections import defaultdict from ansible.plugins.action import ActionBase @@ -38,8 +41,13 @@ class ActionModule(ActionBase): # storing the information we need in the result. result['playbook_context'] = task_vars.get('r_openshift_health_checker_playbook_context') + # if the user wants to write check results to files, they provide this directory: + output_dir = task_vars.get("openshift_checks_output_dir") + if output_dir: + output_dir = os.path.join(output_dir, task_vars["ansible_host"]) + try: - known_checks = self.load_known_checks(tmp, task_vars) + known_checks = self.load_known_checks(tmp, task_vars, output_dir) args = self._task.args requested_checks = normalize(args.get('checks', [])) @@ -65,21 +73,20 @@ class ActionModule(ActionBase): for name in resolved_checks: display.banner("CHECK [{} : {}]".format(name, task_vars["ansible_host"])) - check = known_checks[name] - check_results[name] = run_check(name, check, user_disabled_checks) - if check.changed: - check_results[name]["changed"] = True + check_results[name] = run_check(name, known_checks[name], user_disabled_checks, output_dir) result["changed"] = any(r.get("changed") for r in check_results.values()) if any(r.get("failed") for r in check_results.values()): result["failed"] = True result["msg"] = "One or more checks failed" + write_result_to_output_dir(output_dir, result) return result - def load_known_checks(self, tmp, task_vars): + def load_known_checks(self, tmp, task_vars, output_dir=None): """Find all existing checks and return a mapping of names to instances.""" load_checks() + want_full_results = bool(output_dir) known_checks = {} for cls in OpenShiftCheck.subclasses(): @@ -90,7 +97,12 @@ class ActionModule(ActionBase): "duplicate check name '{}' in: '{}' and '{}'" "".format(name, full_class_name(cls), full_class_name(other_cls)) ) - known_checks[name] = cls(execute_module=self._execute_module, tmp=tmp, task_vars=task_vars) + known_checks[name] = cls( + execute_module=self._execute_module, + tmp=tmp, + task_vars=task_vars, + want_full_results=want_full_results + ) return known_checks @@ -185,8 +197,10 @@ def normalize(checks): return [name.strip() for name in checks if name.strip()] -def run_check(name, check, user_disabled_checks): +def run_check(name, check, user_disabled_checks, output_dir=None): """Run a single check if enabled and return a result dict.""" + + # determine if we're going to run the check (not inactive or disabled) if name in user_disabled_checks or '*' in user_disabled_checks: return dict(skipped=True, skipped_reason="Disabled by user request") @@ -201,12 +215,134 @@ def run_check(name, check, user_disabled_checks): if not is_active: return dict(skipped=True, skipped_reason="Not active for this host") + # run the check + result = {} try: - return check.run() + result = check.run() except OpenShiftCheckException as exc: - return dict(failed=True, msg=str(exc)) + check.register_failure(exc) + except Exception as exc: + check.register_failure("\n".join([str(exc), traceback.format_exc()])) + + # process the check state; compose the result hash, write files as needed + if check.changed: + result["changed"] = True + if check.failures or result.get("failed"): + if "msg" in result: # failure result has msg; combine with any registered failures + check.register_failure(result.get("msg")) + result["failures"] = [(fail.name, str(fail)) for fail in check.failures] + result["failed"] = True + result["msg"] = "\n".join(str(fail) for fail in check.failures) + write_to_output_file(output_dir, name + ".failures.json", result["failures"]) + if check.logs: + write_to_output_file(output_dir, name + ".log.json", check.logs) + if check.files_to_save: + write_files_to_save(output_dir, check) + + return result + + +def prepare_output_dir(dirname): + """Create the directory, including parents. Return bool for success/failure.""" + try: + os.makedirs(dirname) + return True + except OSError as exc: + # trying to create existing dir leads to error; + # that error is fine, but for any other, assume the dir is not there + return exc.errno == errno.EEXIST + + +def copy_remote_file_to_dir(check, file_to_save, output_dir, fname): + """Copy file from remote host to local file in output_dir, if given.""" + if not output_dir or not prepare_output_dir(output_dir): + return + local_file = os.path.join(output_dir, fname) + + # pylint: disable=broad-except; do not need to do anything about failure to write dir/file + # and do not want exceptions to break anything. + try: + # NOTE: it would have been nice to copy the file directly without loading it into + # memory, but there does not seem to be a good way to do this via ansible. + result = check.execute_module("slurp", dict(src=file_to_save), register=False) + if result.get("failed"): + display.warning("Could not retrieve file {}: {}".format(file_to_save, result.get("msg"))) + return + + content = result["content"] + if result.get("encoding") == "base64": + content = base64.b64decode(content) + with open(local_file, "wb") as outfile: + outfile.write(content) + except Exception as exc: + display.warning("Failed writing remote {} to local {}: {}".format(file_to_save, local_file, exc)) + return + + +def _no_fail(obj): + # pylint: disable=broad-except; do not want serialization to fail for any reason + try: + return str(obj) + except Exception: + return "[not serializable]" + + +def write_to_output_file(output_dir, filename, data): + """If output_dir provided, write data to file. Serialize as JSON if data is not a string.""" + + if not output_dir or not prepare_output_dir(output_dir): + return + filename = os.path.join(output_dir, filename) + try: + with open(filename, 'w') as outfile: + if isinstance(data, string_types): + outfile.write(data) + else: + json.dump(data, outfile, sort_keys=True, indent=4, default=_no_fail) + # pylint: disable=broad-except; do not want serialization/write to break for any reason + except Exception as exc: + display.warning("Could not write output file {}: {}".format(filename, exc)) + + +def write_result_to_output_dir(output_dir, result): + """If output_dir provided, write the result as json to result.json. + + Success/failure of the write is recorded as "output_files" in the result hash afterward. + Otherwise this is much like write_to_output_file. + """ + + if not output_dir: + return + if not prepare_output_dir(output_dir): + result["output_files"] = "Error creating output directory " + output_dir + return + + filename = os.path.join(output_dir, "result.json") + try: + with open(filename, 'w') as outfile: + json.dump(result, outfile, sort_keys=True, indent=4, default=_no_fail) + result["output_files"] = "Check results for this host written to " + filename + # pylint: disable=broad-except; do not want serialization/write to break for any reason except Exception as exc: - return dict(failed=True, msg=str(exc), exception=traceback.format_exc()) + result["output_files"] = "Error writing check results to {}:\n{}".format(filename, exc) + + +def write_files_to_save(output_dir, check): + """Write files to check subdir in output dir.""" + if not output_dir: + return + output_dir = os.path.join(output_dir, check.name) + seen_file = defaultdict(lambda: 0) + for file_to_save in check.files_to_save: + fname = file_to_save.filename + while seen_file[fname]: # just to be sure we never re-write a file, append numbers as needed + seen_file[fname] += 1 + fname = "{}.{}".format(fname, seen_file[fname]) + seen_file[fname] += 1 + if file_to_save.remote_filename: + copy_remote_file_to_dir(check, file_to_save.remote_filename, output_dir, fname) + else: + write_to_output_file(output_dir, fname, file_to_save.contents) def full_class_name(cls): diff --git a/roles/openshift_health_checker/openshift_checks/__init__.py b/roles/openshift_health_checker/openshift_checks/__init__.py index 987c955b6..28cb53cc5 100644 --- a/roles/openshift_health_checker/openshift_checks/__init__.py +++ b/roles/openshift_health_checker/openshift_checks/__init__.py @@ -2,9 +2,11 @@ Health checks for OpenShift clusters. """ +import json import operator import os import time +import collections from abc import ABCMeta, abstractmethod, abstractproperty from importlib import import_module @@ -28,7 +30,7 @@ class OpenShiftCheckException(Exception): class OpenShiftCheckExceptionList(OpenShiftCheckException): - """A container for multiple logging errors that may be detected in one check.""" + """A container for multiple errors that may be detected in one check.""" def __init__(self, errors): self.errors = errors super(OpenShiftCheckExceptionList, self).__init__( @@ -41,29 +43,53 @@ class OpenShiftCheckExceptionList(OpenShiftCheckException): return self.errors[index] +FileToSave = collections.namedtuple("FileToSave", "filename contents remote_filename") + + +# pylint: disable=too-many-instance-attributes; all represent significantly different state. +# Arguably they could be separated into two hashes, one for storing parameters, and one for +# storing result state; but that smells more like clutter than clarity. @six.add_metaclass(ABCMeta) class OpenShiftCheck(object): - """ - A base class for defining checks for an OpenShift cluster environment. + """A base class for defining checks for an OpenShift cluster environment. - Expect optional params: method execute_module, dict task_vars, and string tmp. + Optional init params: method execute_module, dict task_vars, and string tmp execute_module is expected to have a signature compatible with _execute_module from ansible plugins/action/__init__.py, e.g.: def execute_module(module_name=None, module_args=None, tmp=None, task_vars=None, *args): This is stored so that it can be invoked in subclasses via check.execute_module("name", args) which provides the check's stored task_vars and tmp. + + Optional init param: want_full_results + If the check can gather logs, tarballs, etc., do so when True; but no need to spend + the time if they're not wanted (won't be written to output directory). """ - def __init__(self, execute_module=None, task_vars=None, tmp=None): + def __init__(self, execute_module=None, task_vars=None, tmp=None, want_full_results=False): + # store a method for executing ansible modules from the check self._execute_module = execute_module + # the task variables and tmpdir passed into the health checker task self.task_vars = task_vars or {} self.tmp = tmp + # a boolean for disabling the gathering of results (files, computations) that won't + # actually be recorded/used + self.want_full_results = want_full_results + # mainly for testing purposes; see execute_module_with_retries self._module_retries = 3 self._module_retry_interval = 5 # seconds + # state to be recorded for inspection after the check runs: + # # set to True when the check changes the host, for accurate total "changed" count self.changed = False + # list of OpenShiftCheckException for check to report (alternative to returning a failed result) + self.failures = [] + # list of FileToSave - files the check specifies to be written locally if so configured + self.files_to_save = [] + # log messages for the check - tuples of (description, msg) where msg is serializable. + # These are intended to be a sequential record of what the check observed and determined. + self.logs = [] @abstractproperty def name(self): @@ -86,7 +112,13 @@ class OpenShiftCheck(object): @abstractmethod def run(self): - """Executes a check, normally implemented as a module.""" + """Executes a check against a host and returns a result hash similar to Ansible modules. + + Actually the direction ahead is to record state in the attributes and + not bother building a result hash. Instead, return an empty hash and let + the action plugin fill it in. Or raise an OpenShiftCheckException. + Returning a hash may become deprecated if it does not prove necessary. + """ return {} @classmethod @@ -98,7 +130,43 @@ class OpenShiftCheck(object): for subclass in subclass.subclasses(): yield subclass - def execute_module(self, module_name=None, module_args=None): + def register_failure(self, error): + """Record in the check that a failure occurred. + + Recorded failures are merged into the result hash for now. They are also saved to output directory + (if provided) <check>.failures.json and registered as a log entry for context <check>.log.json. + """ + # It should be an exception; make it one if not + if not isinstance(error, OpenShiftCheckException): + error = OpenShiftCheckException(str(error)) + self.failures.append(error) + # duplicate it in the logs so it can be seen in the context of any + # information that led to the failure + self.register_log("failure: " + error.name, str(error)) + + def register_log(self, context, msg): + """Record an entry for the check log. + + Notes are intended to serve as context of the whole sequence of what the check observed. + They are be saved as an ordered list in a local check log file. + They are not to included in the result or in the ansible log; it's just for the record. + """ + self.logs.append([context, msg]) + + def register_file(self, filename, contents=None, remote_filename=""): + """Record a file that a check makes available to be saved individually to output directory. + + Either file contents should be passed in, or a file to be copied from the remote host + should be specified. Contents that are not a string are to be serialized as JSON. + + NOTE: When copying a file from remote host, it is slurped into memory as base64, meaning + you should avoid using this on huge files (more than say 10M). + """ + if contents is None and not remote_filename: + raise OpenShiftCheckException("File data/source not specified; this is a bug in the check.") + self.files_to_save.append(FileToSave(filename, contents, remote_filename)) + + def execute_module(self, module_name=None, module_args=None, save_as_name=None, register=True): """Invoke an Ansible module from a check. Invoke stored _execute_module, normally copied from the action @@ -110,6 +178,12 @@ class OpenShiftCheck(object): Ansible version). So e.g. check.execute_module("foo", dict(arg1=...)) + + save_as_name specifies a file name for saving the result to an output directory, + if needed, and is intended to uniquely identify the result of invoking execute_module. + If not provided, the module name will be used. + If register is set False, then the result won't be registered in logs or files to save. + Return: result hash from module execution. """ if self._execute_module is None: @@ -117,7 +191,20 @@ class OpenShiftCheck(object): self.__class__.__name__ + " invoked execute_module without providing the method at initialization." ) - return self._execute_module(module_name, module_args, self.tmp, self.task_vars) + result = self._execute_module(module_name, module_args, self.tmp, self.task_vars) + if result.get("changed"): + self.changed = True + for output in ["result", "stdout"]: + # output is often JSON; attempt to decode + try: + result[output + "_json"] = json.loads(result[output]) + except (KeyError, ValueError): + pass + + if register: + self.register_log("execute_module: " + module_name, result) + self.register_file(save_as_name or module_name + ".json", result) + return result def execute_module_with_retries(self, module_name, module_args): """Run execute_module and retry on failure.""" @@ -188,8 +275,12 @@ class OpenShiftCheck(object): 'There is a bug in this check. While trying to convert variable \n' ' "{var}={value}"\n' 'the given converter cannot be used or failed unexpectedly:\n' - '{error}'.format(var=".".join(keys), value=value, error=error) - ) + '{type}: {error}'.format( + var=".".join(keys), + value=value, + type=error.__class__.__name__, + error=error + )) @staticmethod def get_major_minor_version(openshift_image_tag): @@ -231,7 +322,9 @@ class OpenShiftCheck(object): mount_point = os.path.dirname(mount_point) try: - return mount_for_path[mount_point] + mount = mount_for_path[mount_point] + self.register_log("mount point for " + path, mount) + return mount except KeyError: known_mounts = ', '.join('"{}"'.format(mount) for mount in sorted(mount_for_path)) raise OpenShiftCheckException( @@ -259,7 +352,7 @@ def load_checks(path=None, subpkg=""): modules = modules + load_checks(os.path.join(path, name), subpkg + "." + name) continue - if name.endswith(".py") and not name.startswith(".") and name not in LOADER_EXCLUDES: + if name.endswith(".py") and name not in LOADER_EXCLUDES: modules.append(import_module(__package__ + subpkg + "." + name[:-3])) return modules diff --git a/roles/openshift_health_checker/openshift_checks/disk_availability.py b/roles/openshift_health_checker/openshift_checks/disk_availability.py index f302fd14b..cdf56e959 100644 --- a/roles/openshift_health_checker/openshift_checks/disk_availability.py +++ b/roles/openshift_health_checker/openshift_checks/disk_availability.py @@ -70,6 +70,10 @@ class DiskAvailability(OpenShiftCheck): # If it is not a number, then it should be a nested dict. pass + self.register_log("recommended thresholds", self.recommended_disk_space_bytes) + if user_config: + self.register_log("user-configured thresholds", user_config) + # TODO: as suggested in # https://github.com/openshift/openshift-ansible/pull/4436#discussion_r122180021, # maybe we could support checking disk availability in paths that are @@ -113,10 +117,7 @@ class DiskAvailability(OpenShiftCheck): 'in your Ansible inventory, and lower the recommended disk space availability\n' 'if necessary for this upgrade.').format(config_bytes) - return { - 'failed': True, - 'msg': msg, - } + self.register_failure(msg) return {} diff --git a/roles/openshift_health_checker/openshift_checks/logging/elasticsearch.py b/roles/openshift_health_checker/openshift_checks/logging/elasticsearch.py index 7fc843fd7..986a01f38 100644 --- a/roles/openshift_health_checker/openshift_checks/logging/elasticsearch.py +++ b/roles/openshift_health_checker/openshift_checks/logging/elasticsearch.py @@ -72,7 +72,7 @@ class Elasticsearch(LoggingCheck): for pod_name in pods_by_name.keys(): # Compare what each ES node reports as master and compare for split brain get_master_cmd = self._build_es_curl_cmd(pod_name, "https://localhost:9200/_cat/master") - master_name_str = self.exec_oc(get_master_cmd, []) + master_name_str = self.exec_oc(get_master_cmd, [], save_as_name="get_master_names.json") master_names = (master_name_str or '').split(' ') if len(master_names) > 1: es_master_names.add(master_names[1]) @@ -113,7 +113,7 @@ class Elasticsearch(LoggingCheck): # get ES cluster nodes node_cmd = self._build_es_curl_cmd(list(pods_by_name.keys())[0], 'https://localhost:9200/_nodes') - cluster_node_data = self.exec_oc(node_cmd, []) + cluster_node_data = self.exec_oc(node_cmd, [], save_as_name="get_es_nodes.json") try: cluster_nodes = json.loads(cluster_node_data)['nodes'] except (ValueError, KeyError): @@ -142,7 +142,7 @@ class Elasticsearch(LoggingCheck): errors = [] for pod_name in pods_by_name.keys(): cluster_health_cmd = self._build_es_curl_cmd(pod_name, 'https://localhost:9200/_cluster/health?pretty=true') - cluster_health_data = self.exec_oc(cluster_health_cmd, []) + cluster_health_data = self.exec_oc(cluster_health_cmd, [], save_as_name='get_es_health.json') try: health_res = json.loads(cluster_health_data) if not health_res or not health_res.get('status'): @@ -171,7 +171,7 @@ class Elasticsearch(LoggingCheck): errors = [] for pod_name in pods_by_name.keys(): df_cmd = 'exec {} -- df --output=ipcent,pcent /elasticsearch/persistent'.format(pod_name) - disk_output = self.exec_oc(df_cmd, []) + disk_output = self.exec_oc(df_cmd, [], save_as_name='get_pv_diskspace.json') lines = disk_output.splitlines() # expecting one header looking like 'IUse% Use%' and one body line body_re = r'\s*(\d+)%?\s+(\d+)%?\s*$' diff --git a/roles/openshift_health_checker/openshift_checks/logging/logging.py b/roles/openshift_health_checker/openshift_checks/logging/logging.py index ecd8adb64..06bdfebf6 100644 --- a/roles/openshift_health_checker/openshift_checks/logging/logging.py +++ b/roles/openshift_health_checker/openshift_checks/logging/logging.py @@ -78,7 +78,7 @@ class LoggingCheck(OpenShiftCheck): """Returns the namespace in which logging is configured to deploy.""" return self.get_var("openshift_logging_namespace", default="logging") - def exec_oc(self, cmd_str="", extra_args=None): + def exec_oc(self, cmd_str="", extra_args=None, save_as_name=None): """ Execute an 'oc' command in the remote host. Returns: output of command and namespace, @@ -92,7 +92,7 @@ class LoggingCheck(OpenShiftCheck): "extra_args": list(extra_args) if extra_args else [], } - result = self.execute_module("ocutil", args) + result = self.execute_module("ocutil", args, save_as_name=save_as_name) if result.get("failed"): if result['result'] == '[Errno 2] No such file or directory': raise CouldNotUseOc( diff --git a/roles/openshift_health_checker/openshift_checks/logging/logging_index_time.py b/roles/openshift_health_checker/openshift_checks/logging/logging_index_time.py index d781db649..cacdf4213 100644 --- a/roles/openshift_health_checker/openshift_checks/logging/logging_index_time.py +++ b/roles/openshift_health_checker/openshift_checks/logging/logging_index_time.py @@ -104,7 +104,7 @@ class LoggingIndexTime(LoggingCheck): "https://logging-es:9200/project.{namespace}*/_count?q=message:{uuid}" ) exec_cmd = exec_cmd.format(pod_name=pod_name, namespace=self.logging_namespace(), uuid=uuid) - result = self.exec_oc(exec_cmd, []) + result = self.exec_oc(exec_cmd, [], save_as_name="query_for_uuid.json") try: count = json.loads(result)["count"] diff --git a/roles/openshift_health_checker/openshift_checks/mixins.py b/roles/openshift_health_checker/openshift_checks/mixins.py index 24f1d938a..b90ebf6dd 100644 --- a/roles/openshift_health_checker/openshift_checks/mixins.py +++ b/roles/openshift_health_checker/openshift_checks/mixins.py @@ -49,5 +49,4 @@ class DockerHostMixin(object): " {deps}\n{msg}" ).format(deps=',\n '.join(self.dependencies), msg=msg) failed = result.get("failed", False) or result.get("rc", 0) != 0 - self.changed = result.get("changed", False) return msg, failed diff --git a/roles/openshift_health_checker/test/action_plugin_test.py b/roles/openshift_health_checker/test/action_plugin_test.py index 58864da21..f14887303 100644 --- a/roles/openshift_health_checker/test/action_plugin_test.py +++ b/roles/openshift_health_checker/test/action_plugin_test.py @@ -3,10 +3,12 @@ import pytest from ansible.playbook.play_context import PlayContext from openshift_health_check import ActionModule, resolve_checks -from openshift_checks import OpenShiftCheckException +from openshift_health_check import copy_remote_file_to_dir, write_result_to_output_dir, write_to_output_file +from openshift_checks import OpenShiftCheckException, FileToSave -def fake_check(name='fake_check', tags=None, is_active=True, run_return=None, run_exception=None, changed=False): +def fake_check(name='fake_check', tags=None, is_active=True, run_return=None, run_exception=None, + run_logs=None, run_files=None, changed=False, get_var_return=None): """Returns a new class that is compatible with OpenShiftCheck for testing.""" _name, _tags = name, tags @@ -14,12 +16,16 @@ def fake_check(name='fake_check', tags=None, is_active=True, run_return=None, ru class FakeCheck(object): name = _name tags = _tags or [] - changed = False - def __init__(self, execute_module=None, task_vars=None, tmp=None): - pass + def __init__(self, **_): + self.changed = False + self.failures = [] + self.logs = run_logs or [] + self.files_to_save = run_files or [] def is_active(self): + if isinstance(is_active, Exception): + raise is_active return is_active def run(self): @@ -28,6 +34,13 @@ def fake_check(name='fake_check', tags=None, is_active=True, run_return=None, ru raise run_exception return run_return + def get_var(*args, **_): + return get_var_return + + def register_failure(self, exc): + self.failures.append(OpenShiftCheckException(str(exc))) + return + return FakeCheck @@ -98,13 +111,18 @@ def test_action_plugin_cannot_load_checks_with_the_same_name(plugin, task_vars, assert failed(result, msg_has=['duplicate', 'duplicate_name', 'FakeCheck']) -def test_action_plugin_skip_non_active_checks(plugin, task_vars, monkeypatch): - checks = [fake_check(is_active=False)] +@pytest.mark.parametrize('is_active, skipped_reason', [ + (False, "Not active for this host"), + (Exception("borked"), "exception"), +]) +def test_action_plugin_skip_non_active_checks(is_active, skipped_reason, plugin, task_vars, monkeypatch): + checks = [fake_check(is_active=is_active)] monkeypatch.setattr('openshift_checks.OpenShiftCheck.subclasses', classmethod(lambda cls: checks)) result = plugin.run(tmp=None, task_vars=task_vars) - assert result['checks']['fake_check'] == dict(skipped=True, skipped_reason="Not active for this host") + assert result['checks']['fake_check'].get('skipped') + assert skipped_reason in result['checks']['fake_check'].get('skipped_reason') assert not failed(result) assert not changed(result) assert not skipped(result) @@ -128,10 +146,21 @@ def test_action_plugin_skip_disabled_checks(to_disable, plugin, task_vars, monke assert not skipped(result) +def test_action_plugin_run_list_checks(monkeypatch): + task = FakeTask('openshift_health_check', {'checks': []}) + plugin = ActionModule(task, None, PlayContext(), None, None, None) + monkeypatch.setattr(plugin, 'load_known_checks', lambda *_: {}) + result = plugin.run() + + assert failed(result, msg_has="Available checks") + assert not changed(result) + assert not skipped(result) + + def test_action_plugin_run_check_ok(plugin, task_vars, monkeypatch): check_return_value = {'ok': 'test'} - check_class = fake_check(run_return=check_return_value) - monkeypatch.setattr(plugin, 'load_known_checks', lambda tmp, task_vars: {'fake_check': check_class()}) + check_class = fake_check(run_return=check_return_value, run_files=[None]) + monkeypatch.setattr(plugin, 'load_known_checks', lambda *_: {'fake_check': check_class()}) monkeypatch.setattr('openshift_health_check.resolve_checks', lambda *args: ['fake_check']) result = plugin.run(tmp=None, task_vars=task_vars) @@ -145,7 +174,7 @@ def test_action_plugin_run_check_ok(plugin, task_vars, monkeypatch): def test_action_plugin_run_check_changed(plugin, task_vars, monkeypatch): check_return_value = {'ok': 'test'} check_class = fake_check(run_return=check_return_value, changed=True) - monkeypatch.setattr(plugin, 'load_known_checks', lambda tmp, task_vars: {'fake_check': check_class()}) + monkeypatch.setattr(plugin, 'load_known_checks', lambda *_: {'fake_check': check_class()}) monkeypatch.setattr('openshift_health_check.resolve_checks', lambda *args: ['fake_check']) result = plugin.run(tmp=None, task_vars=task_vars) @@ -158,9 +187,9 @@ def test_action_plugin_run_check_changed(plugin, task_vars, monkeypatch): def test_action_plugin_run_check_fail(plugin, task_vars, monkeypatch): - check_return_value = {'failed': True} + check_return_value = {'failed': True, 'msg': 'this is a failure'} check_class = fake_check(run_return=check_return_value) - monkeypatch.setattr(plugin, 'load_known_checks', lambda tmp, task_vars: {'fake_check': check_class()}) + monkeypatch.setattr(plugin, 'load_known_checks', lambda *_: {'fake_check': check_class()}) monkeypatch.setattr('openshift_health_check.resolve_checks', lambda *args: ['fake_check']) result = plugin.run(tmp=None, task_vars=task_vars) @@ -171,24 +200,51 @@ def test_action_plugin_run_check_fail(plugin, task_vars, monkeypatch): assert not skipped(result) -def test_action_plugin_run_check_exception(plugin, task_vars, monkeypatch): +@pytest.mark.parametrize('exc_class, expect_traceback', [ + (OpenShiftCheckException, False), + (Exception, True), +]) +def test_action_plugin_run_check_exception(plugin, task_vars, exc_class, expect_traceback, monkeypatch): exception_msg = 'fake check has an exception' - run_exception = OpenShiftCheckException(exception_msg) + run_exception = exc_class(exception_msg) check_class = fake_check(run_exception=run_exception, changed=True) - monkeypatch.setattr(plugin, 'load_known_checks', lambda tmp, task_vars: {'fake_check': check_class()}) + monkeypatch.setattr(plugin, 'load_known_checks', lambda *_: {'fake_check': check_class()}) monkeypatch.setattr('openshift_health_check.resolve_checks', lambda *args: ['fake_check']) result = plugin.run(tmp=None, task_vars=task_vars) assert failed(result['checks']['fake_check'], msg_has=exception_msg) + assert expect_traceback == ("Traceback" in result['checks']['fake_check']['msg']) assert failed(result, msg_has=['failed']) assert changed(result['checks']['fake_check']) assert changed(result) assert not skipped(result) +def test_action_plugin_run_check_output_dir(plugin, task_vars, tmpdir, monkeypatch): + check_class = fake_check( + run_return={}, + run_logs=[('thing', 'note')], + run_files=[ + FileToSave('save.file', 'contents', None), + FileToSave('save.file', 'duplicate', None), + FileToSave('copy.file', None, 'foo'), # note: copy runs execute_module => exception + ], + ) + task_vars['openshift_checks_output_dir'] = str(tmpdir) + check_class.get_var = lambda self, name, **_: task_vars.get(name) + monkeypatch.setattr(plugin, 'load_known_checks', lambda *_: {'fake_check': check_class()}) + monkeypatch.setattr('openshift_health_check.resolve_checks', lambda *args: ['fake_check']) + + plugin.run(tmp=None, task_vars=task_vars) + assert any(path.basename == task_vars['ansible_host'] for path in tmpdir.listdir()) + assert any(path.basename == 'fake_check.log.json' for path in tmpdir.visit()) + assert any(path.basename == 'save.file' for path in tmpdir.visit()) + assert any(path.basename == 'save.file.2' for path in tmpdir.visit()) + + def test_action_plugin_resolve_checks_exception(plugin, task_vars, monkeypatch): - monkeypatch.setattr(plugin, 'load_known_checks', lambda tmp, task_vars: {}) + monkeypatch.setattr(plugin, 'load_known_checks', lambda *_: {}) result = plugin.run(tmp=None, task_vars=task_vars) @@ -254,3 +310,38 @@ def test_resolve_checks_failure(names, all_checks, words_in_exception): resolve_checks(names, all_checks) for word in words_in_exception: assert word in str(excinfo.value) + + +@pytest.mark.parametrize('give_output_dir, result, expect_file', [ + (False, None, False), + (True, dict(content="c3BhbQo=", encoding="base64"), True), + (True, dict(content="encoding error", encoding="base64"), False), + (True, dict(content="spam", no_encoding=None), True), + (True, dict(failed=True, msg="could not slurp"), False), +]) +def test_copy_remote_file_to_dir(give_output_dir, result, expect_file, tmpdir): + check = fake_check()() + check.execute_module = lambda *args, **_: result + copy_remote_file_to_dir(check, "remote_file", str(tmpdir) if give_output_dir else "", "local_file") + assert expect_file == any(path.basename == "local_file" for path in tmpdir.listdir()) + + +def test_write_to_output_exceptions(tmpdir, monkeypatch, capsys): + + class Spam(object): + def __str__(self): + raise Exception("break str") + + test = {1: object(), 2: Spam()} + test[3] = test + write_result_to_output_dir(str(tmpdir), test) + assert "Error writing" in test["output_files"] + + output_dir = tmpdir.join("eggs") + output_dir.write("spam") # so now it's not a dir + write_to_output_file(str(output_dir), "somefile", "somedata") + assert "Could not write" in capsys.readouterr()[1] + + monkeypatch.setattr("openshift_health_check.prepare_output_dir", lambda *_: False) + write_result_to_output_dir(str(tmpdir), test) + assert "Error creating" in test["output_files"] diff --git a/roles/openshift_health_checker/test/disk_availability_test.py b/roles/openshift_health_checker/test/disk_availability_test.py index f4fd2dfed..9ae679b79 100644 --- a/roles/openshift_health_checker/test/disk_availability_test.py +++ b/roles/openshift_health_checker/test/disk_availability_test.py @@ -183,11 +183,12 @@ def test_fails_with_insufficient_disk_space(name, group_names, configured_min, a ansible_mounts=ansible_mounts, ) - result = DiskAvailability(fake_execute_module, task_vars).run() + check = DiskAvailability(fake_execute_module, task_vars) + check.run() - assert result['failed'] + assert check.failures for chunk in 'below recommended'.split() + expect_chunks: - assert chunk in result.get('msg', '') + assert chunk in str(check.failures[0]) @pytest.mark.parametrize('name,group_names,context,ansible_mounts,failed,extra_words', [ @@ -237,11 +238,11 @@ def test_min_required_space_changes_with_upgrade_context(name, group_names, cont ) check = DiskAvailability(fake_execute_module, task_vars) - result = check.run() + check.run() - assert result.get("failed", False) == failed + assert bool(check.failures) == failed for word in extra_words: - assert word in result.get('msg', '') + assert word in str(check.failures[0]) def fake_execute_module(*args): diff --git a/roles/openshift_health_checker/test/elasticsearch_test.py b/roles/openshift_health_checker/test/elasticsearch_test.py index 09bacd9ac..3fa5e8929 100644 --- a/roles/openshift_health_checker/test/elasticsearch_test.py +++ b/roles/openshift_health_checker/test/elasticsearch_test.py @@ -72,7 +72,7 @@ def test_check_elasticsearch(): assert_error_in_list('NoRunningPods', excinfo.value) # canned oc responses to match so all the checks pass - def exec_oc(cmd, args): + def exec_oc(cmd, args, **_): if '_cat/master' in cmd: return 'name logging-es' elif '/_nodes' in cmd: @@ -97,7 +97,7 @@ def test_check_running_es_pods(): def test_check_elasticsearch_masters(): pods = [plain_es_pod] - check = canned_elasticsearch(task_vars_config_base, lambda *_: plain_es_pod['_test_master_name_str']) + check = canned_elasticsearch(task_vars_config_base, lambda *args, **_: plain_es_pod['_test_master_name_str']) assert not check.check_elasticsearch_masters(pods_by_name(pods)) @@ -117,7 +117,7 @@ def test_check_elasticsearch_masters(): ]) def test_check_elasticsearch_masters_error(pods, expect_error): test_pods = list(pods) - check = canned_elasticsearch(task_vars_config_base, lambda *_: test_pods.pop(0)['_test_master_name_str']) + check = canned_elasticsearch(task_vars_config_base, lambda *args, **_: test_pods.pop(0)['_test_master_name_str']) assert_error_in_list(expect_error, check.check_elasticsearch_masters(pods_by_name(pods))) @@ -129,7 +129,7 @@ es_node_list = { def test_check_elasticsearch_node_list(): - check = canned_elasticsearch(task_vars_config_base, lambda *_: json.dumps(es_node_list)) + check = canned_elasticsearch(task_vars_config_base, lambda *args, **_: json.dumps(es_node_list)) assert not check.check_elasticsearch_node_list(pods_by_name([plain_es_pod])) @@ -151,13 +151,13 @@ def test_check_elasticsearch_node_list(): ), ]) def test_check_elasticsearch_node_list_errors(pods, node_list, expect_error): - check = canned_elasticsearch(task_vars_config_base, lambda cmd, args: json.dumps(node_list)) + check = canned_elasticsearch(task_vars_config_base, lambda cmd, args, **_: json.dumps(node_list)) assert_error_in_list(expect_error, check.check_elasticsearch_node_list(pods_by_name(pods))) def test_check_elasticsearch_cluster_health(): test_health_data = [{"status": "green"}] - check = canned_elasticsearch(exec_oc=lambda *_: json.dumps(test_health_data.pop(0))) + check = canned_elasticsearch(exec_oc=lambda *args, **_: json.dumps(test_health_data.pop(0))) assert not check.check_es_cluster_health(pods_by_name([plain_es_pod])) @@ -175,12 +175,12 @@ def test_check_elasticsearch_cluster_health(): ]) def test_check_elasticsearch_cluster_health_errors(pods, health_data, expect_error): test_health_data = list(health_data) - check = canned_elasticsearch(exec_oc=lambda *_: json.dumps(test_health_data.pop(0))) + check = canned_elasticsearch(exec_oc=lambda *args, **_: json.dumps(test_health_data.pop(0))) assert_error_in_list(expect_error, check.check_es_cluster_health(pods_by_name(pods))) def test_check_elasticsearch_diskspace(): - check = canned_elasticsearch(exec_oc=lambda *_: 'IUse% Use%\n 3% 4%\n') + check = canned_elasticsearch(exec_oc=lambda *args, **_: 'IUse% Use%\n 3% 4%\n') assert not check.check_elasticsearch_diskspace(pods_by_name([plain_es_pod])) @@ -199,5 +199,5 @@ def test_check_elasticsearch_diskspace(): ), ]) def test_check_elasticsearch_diskspace_errors(disk_data, expect_error): - check = canned_elasticsearch(exec_oc=lambda *_: disk_data) + check = canned_elasticsearch(exec_oc=lambda *args, **_: disk_data) assert_error_in_list(expect_error, check.check_elasticsearch_diskspace(pods_by_name([plain_es_pod]))) diff --git a/roles/openshift_health_checker/test/logging_index_time_test.py b/roles/openshift_health_checker/test/logging_index_time_test.py index 22566b295..c48ade9b8 100644 --- a/roles/openshift_health_checker/test/logging_index_time_test.py +++ b/roles/openshift_health_checker/test/logging_index_time_test.py @@ -102,7 +102,7 @@ def test_with_running_pods(): ), ], ids=lambda argval: argval[0]) def test_wait_until_cmd_or_err_succeeds(name, json_response, uuid, timeout): - check = canned_loggingindextime(lambda *_: json.dumps(json_response)) + check = canned_loggingindextime(lambda *args, **_: json.dumps(json_response)) check.wait_until_cmd_or_err(plain_running_elasticsearch_pod, uuid, timeout) @@ -131,7 +131,7 @@ def test_wait_until_cmd_or_err_succeeds(name, json_response, uuid, timeout): ) ], ids=lambda argval: argval[0]) def test_wait_until_cmd_or_err(name, json_response, timeout, expect_error): - check = canned_loggingindextime(lambda *_: json.dumps(json_response)) + check = canned_loggingindextime(lambda *args, **_: json.dumps(json_response)) with pytest.raises(OpenShiftCheckException) as error: check.wait_until_cmd_or_err(plain_running_elasticsearch_pod, SAMPLE_UUID, timeout) @@ -139,7 +139,7 @@ def test_wait_until_cmd_or_err(name, json_response, timeout, expect_error): def test_curl_kibana_with_uuid(): - check = canned_loggingindextime(lambda *_: json.dumps({"statusCode": 404})) + check = canned_loggingindextime(lambda *args, **_: json.dumps({"statusCode": 404})) check.generate_uuid = lambda: SAMPLE_UUID assert SAMPLE_UUID == check.curl_kibana_with_uuid(plain_running_kibana_pod) @@ -161,7 +161,7 @@ def test_curl_kibana_with_uuid(): ), ], ids=lambda argval: argval[0]) def test_failed_curl_kibana_with_uuid(name, json_response, expect_error): - check = canned_loggingindextime(lambda *_: json.dumps(json_response)) + check = canned_loggingindextime(lambda *args, **_: json.dumps(json_response)) check.generate_uuid = lambda: SAMPLE_UUID with pytest.raises(OpenShiftCheckException) as error: diff --git a/roles/openshift_health_checker/test/openshift_check_test.py b/roles/openshift_health_checker/test/openshift_check_test.py index 789784c77..bc0c3b26c 100644 --- a/roles/openshift_health_checker/test/openshift_check_test.py +++ b/roles/openshift_health_checker/test/openshift_check_test.py @@ -106,13 +106,40 @@ def test_get_var_convert(task_vars, keys, convert, expected): assert dummy_check(task_vars).get_var(*keys, convert=convert) == expected -@pytest.mark.parametrize("keys, convert", [ - (("bar", "baz"), int), - (("bar.baz"), float), - (("foo"), "bogus"), - (("foo"), lambda a, b: 1), - (("foo"), lambda a: 1 / 0), +def convert_oscexc(_): + raise OpenShiftCheckException("known failure") + + +def convert_exc(_): + raise Exception("failure unknown") + + +@pytest.mark.parametrize("keys, convert, expect_text", [ + (("bar", "baz"), int, "Cannot convert"), + (("bar.baz",), float, "Cannot convert"), + (("foo",), "bogus", "TypeError"), + (("foo",), lambda a, b: 1, "TypeError"), + (("foo",), lambda a: 1 / 0, "ZeroDivisionError"), + (("foo",), convert_oscexc, "known failure"), + (("foo",), convert_exc, "failure unknown"), ]) -def test_get_var_convert_error(task_vars, keys, convert): - with pytest.raises(OpenShiftCheckException): +def test_get_var_convert_error(task_vars, keys, convert, expect_text): + with pytest.raises(OpenShiftCheckException) as excinfo: dummy_check(task_vars).get_var(*keys, convert=convert) + assert expect_text in str(excinfo.value) + + +def test_register(task_vars): + check = dummy_check(task_vars) + + check.register_failure(OpenShiftCheckException("spam")) + assert "spam" in str(check.failures[0]) + + with pytest.raises(OpenShiftCheckException) as excinfo: + check.register_file("spam") # no file contents specified + assert "not specified" in str(excinfo.value) + + # normally execute_module registers the result file; test disabling that + check._execute_module = lambda *args, **_: dict() + check.execute_module("eggs", module_args={}, register=False) + assert not check.files_to_save diff --git a/roles/openshift_health_checker/test/ovs_version_test.py b/roles/openshift_health_checker/test/ovs_version_test.py index e1bf29d2a..602f32989 100644 --- a/roles/openshift_health_checker/test/ovs_version_test.py +++ b/roles/openshift_health_checker/test/ovs_version_test.py @@ -50,7 +50,7 @@ def test_ovs_package_version(openshift_release, expected_ovs_version): openshift_release=openshift_release, openshift_image_tag='v' + openshift_release, ) - return_value = object() + return_value = {} # note: check.execute_module modifies return hash contents def execute_module(module_name=None, module_args=None, *_): assert module_name == 'rpm_version' diff --git a/roles/openshift_health_checker/test/package_availability_test.py b/roles/openshift_health_checker/test/package_availability_test.py index 8aa87ca59..b34e8fbfc 100644 --- a/roles/openshift_health_checker/test/package_availability_test.py +++ b/roles/openshift_health_checker/test/package_availability_test.py @@ -49,7 +49,7 @@ def test_is_active(pkg_mgr, is_containerized, is_active): ), ]) def test_package_availability(task_vars, must_have_packages, must_not_have_packages): - return_value = object() + return_value = {} def execute_module(module_name=None, module_args=None, *_): assert module_name == 'check_yum_update' diff --git a/roles/openshift_health_checker/test/package_update_test.py b/roles/openshift_health_checker/test/package_update_test.py index 7d9035a36..85d3c9cab 100644 --- a/roles/openshift_health_checker/test/package_update_test.py +++ b/roles/openshift_health_checker/test/package_update_test.py @@ -2,7 +2,7 @@ from openshift_checks.package_update import PackageUpdate def test_package_update(): - return_value = object() + return_value = {} def execute_module(module_name=None, module_args=None, *_): assert module_name == 'check_yum_update' diff --git a/roles/openshift_hosted/tasks/registry/secure.yml b/roles/openshift_hosted/tasks/registry/secure.yml index a8a6f6fc8..434b679df 100644 --- a/roles/openshift_hosted/tasks/registry/secure.yml +++ b/roles/openshift_hosted/tasks/registry/secure.yml @@ -1,7 +1,7 @@ --- - name: Configure facts for docker-registry set_fact: - openshift_hosted_registry_routecertificates: "{{ ('routecertificates' in openshift.hosted.registry.keys()) | ternary(openshift.hosted.registry.routecertificates, {}) }}" + openshift_hosted_registry_routecertificates: "{{ ('routecertificates' in openshift.hosted.registry.keys()) | ternary(openshift_hosted_registry_routecertificates, {}) }}" openshift_hosted_registry_routehost: "{{ ('routehost' in openshift.hosted.registry.keys()) | ternary(openshift.hosted.registry.routehost, False) }}" openshift_hosted_registry_routetermination: "{{ ('routetermination' in openshift.hosted.registry.keys()) | ternary(openshift.hosted.registry.routetermination, 'passthrough') }}" diff --git a/roles/openshift_manageiq/vars/main.yml b/roles/openshift_manageiq/vars/main.yml index 7ccc2fc3b..f142f89f0 100644 --- a/roles/openshift_manageiq/vars/main.yml +++ b/roles/openshift_manageiq/vars/main.yml @@ -3,6 +3,9 @@ manage_iq_tasks: - resource_kind: role resource_name: admin user: management-admin +- resource_kind: role + resource_name: admin + user: system:serviceaccount:management-infra:management-admin - resource_kind: cluster-role resource_name: management-infra-admin user: system:serviceaccount:management-infra:management-admin diff --git a/roles/openshift_master/meta/main.yml b/roles/openshift_master/meta/main.yml index b0237141b..a657668a9 100644 --- a/roles/openshift_master/meta/main.yml +++ b/roles/openshift_master/meta/main.yml @@ -14,19 +14,3 @@ galaxy_info: dependencies: - role: lib_openshift - role: lib_os_firewall -- role: openshift_master_facts -- role: openshift_hosted_facts -- role: openshift_master_certificates -- role: openshift_etcd_client_certificates - etcd_cert_subdir: "openshift-master-{{ openshift.common.hostname }}" - etcd_cert_config_dir: "{{ openshift.common.config_base }}/master" - etcd_cert_prefix: "master.etcd-" - when: groups.oo_etcd_to_config | default([]) | length != 0 -- role: openshift_clock -- role: openshift_cloud_provider -- role: openshift_builddefaults -- role: openshift_buildoverrides -- role: nickhammond.logrotate -- role: contiv - contiv_role: netmaster - when: openshift_use_contiv | default(False) | bool diff --git a/roles/openshift_master/tasks/main.yml b/roles/openshift_master/tasks/main.yml index 894fe8e2b..1fe0e94b1 100644 --- a/roles/openshift_master/tasks/main.yml +++ b/roles/openshift_master/tasks/main.yml @@ -177,6 +177,8 @@ local_facts: no_proxy_etcd_host_ips: "{{ openshift_no_proxy_etcd_host_ips }}" +- include: registry_auth.yml + - name: Install the systemd units include: systemd_units.yml @@ -229,8 +231,6 @@ - restart master controllers when: openshift_master_bootstrap_enabled | default(False) -- include: registry_auth.yml - - include: set_loopback_context.yml when: - openshift.common.version_gte_3_2_or_1_2 diff --git a/roles/openshift_master/tasks/update_etcd_client_urls.yml b/roles/openshift_master/tasks/update_etcd_client_urls.yml new file mode 100644 index 000000000..1ab105808 --- /dev/null +++ b/roles/openshift_master/tasks/update_etcd_client_urls.yml @@ -0,0 +1,8 @@ +--- +- yedit: + src: "{{ openshift.common.config_base }}/master/master-config.yaml" + key: 'etcdClientInfo.urls' + value: "{{ openshift.master.etcd_urls }}" + notify: + - restart master api + - restart master controllers diff --git a/roles/openshift_master_facts/filter_plugins/openshift_master.py b/roles/openshift_master_facts/filter_plugins/openshift_master.py index 5558f55cb..56c864ec7 100644 --- a/roles/openshift_master_facts/filter_plugins/openshift_master.py +++ b/roles/openshift_master_facts/filter_plugins/openshift_master.py @@ -380,11 +380,6 @@ class OpenIDIdentityProvider(IdentityProviderOauthBase): if 'extra_authorize_parameters' in self._idp: self._idp['extraAuthorizeParameters'] = self._idp.pop('extra_authorize_parameters') - if 'extraAuthorizeParameters' in self._idp: - if 'include_granted_scopes' in self._idp['extraAuthorizeParameters']: - val = ansible_bool(self._idp['extraAuthorizeParameters'].pop('include_granted_scopes')) - self._idp['extraAuthorizeParameters']['include_granted_scopes'] = '"true"' if val else '"false"' - def validate(self): ''' validate this idp instance ''' IdentityProviderOauthBase.validate(self) diff --git a/roles/openshift_node/defaults/main.yml b/roles/openshift_node/defaults/main.yml index 5424a64d2..433e92201 100644 --- a/roles/openshift_node/defaults/main.yml +++ b/roles/openshift_node/defaults/main.yml @@ -60,7 +60,7 @@ openshift_deployment_type: origin openshift_node_bootstrap: False r_openshift_node_os_firewall_deny: [] -r_openshift_node_os_firewall_allow: +default_r_openshift_node_os_firewall_allow: - service: Kubernetes kubelet port: 10250/tcp - service: http @@ -79,6 +79,8 @@ r_openshift_node_os_firewall_allow: - service: Kubernetes service NodePort UDP port: "{{ openshift_node_port_range | default('') }}/udp" cond: "{{ openshift_node_port_range is defined }}" +# Allow multiple port ranges to be added to the role +r_openshift_node_os_firewall_allow: "{{ default_r_openshift_node_os_firewall_allow | union(openshift_node_open_ports | default([])) }}" oreg_url: '' oreg_host: "{{ oreg_url.split('/')[0] if '.' in oreg_url.split('/')[0] else '' }}" diff --git a/roles/openshift_node/handlers/main.yml b/roles/openshift_node/handlers/main.yml index 855b0a8d8..25a6fc721 100644 --- a/roles/openshift_node/handlers/main.yml +++ b/roles/openshift_node/handlers/main.yml @@ -29,8 +29,5 @@ - not (node_service_status_changed | default(false) | bool) - not openshift_node_bootstrap -- name: reload sysctl.conf - command: /sbin/sysctl -p - - name: reload systemd units command: systemctl daemon-reload diff --git a/roles/openshift_node/tasks/config.yml b/roles/openshift_node/tasks/config.yml index 1504d01af..2759188f3 100644 --- a/roles/openshift_node/tasks/config.yml +++ b/roles/openshift_node/tasks/config.yml @@ -2,10 +2,6 @@ - name: Install the systemd units include: systemd_units.yml -- name: Setup tuned - include: tuned.yml - static: yes - - name: Start and enable openvswitch service systemd: name: openvswitch.service @@ -99,5 +95,9 @@ msg: Node failed to start please inspect the logs and try again when: node_start_result | failed +- name: Setup tuned + include: tuned.yml + static: yes + - set_fact: node_service_status_changed: "{{ node_start_result | changed }}" diff --git a/roles/openshift_node/tasks/main.yml b/roles/openshift_node/tasks/main.yml index ff8d1942c..ef79b6ac0 100644 --- a/roles/openshift_node/tasks/main.yml +++ b/roles/openshift_node/tasks/main.yml @@ -59,25 +59,22 @@ # The atomic-openshift-node service will set this parameter on # startup, but if the network service is restarted this setting is # lost. Reference: https://bugzilla.redhat.com/show_bug.cgi?id=1372388 -# -# Use lineinfile w/ a handler for this task until -# https://github.com/ansible/ansible/pull/24277 is included in an -# ansible release and we can use the sysctl module. -- name: Persist net.ipv4.ip_forward sysctl entry - lineinfile: dest=/etc/sysctl.conf regexp='^net.ipv4.ip_forward' line='net.ipv4.ip_forward=1' - notify: - - reload sysctl.conf +- sysctl: + name: net.ipv4.ip_forward + value: 1 + sysctl_file: "/etc/sysctl.d/99-openshift.conf" + reload: yes - name: include bootstrap node config include: bootstrap.yml when: openshift_node_bootstrap +- include: registry_auth.yml + - name: include standard node config include: config.yml when: not openshift_node_bootstrap -- include: registry_auth.yml - - name: Configure AWS Cloud Provider Settings lineinfile: dest: /etc/sysconfig/{{ openshift.common.service_type }}-node diff --git a/roles/openshift_prometheus/defaults/main.yaml b/roles/openshift_prometheus/defaults/main.yaml index 18d6a1645..5aa8aecec 100644 --- a/roles/openshift_prometheus/defaults/main.yaml +++ b/roles/openshift_prometheus/defaults/main.yaml @@ -11,7 +11,7 @@ openshift_prometheus_node_selector: {"region":"infra"} openshift_prometheus_image_proxy: "openshift/oauth-proxy:v1.0.0" openshift_prometheus_image_prometheus: "openshift/prometheus:v2.0.0-dev" openshift_prometheus_image_alertmanager: "openshift/prometheus-alertmanager:dev" -openshift_prometheus_image_alertbuffer: "ilackarms/message-buffer" +openshift_prometheus_image_alertbuffer: "openshift/prometheus-alert-buffer:v0.0.1" # additional prometheus rules file openshift_prometheus_additional_rules_file: null diff --git a/roles/openshift_prometheus/tasks/install_prometheus.yaml b/roles/openshift_prometheus/tasks/install_prometheus.yaml index 93bdda3e8..a9bce2fb1 100644 --- a/roles/openshift_prometheus/tasks/install_prometheus.yaml +++ b/roles/openshift_prometheus/tasks/install_prometheus.yaml @@ -107,7 +107,10 @@ - name: annotate prometheus service command: > {{ openshift.common.client_binary }} annotate --overwrite -n {{ openshift_prometheus_namespace }} - service prometheus 'service.alpha.openshift.io/serving-cert-secret-name=prometheus-tls' + service prometheus + prometheus.io/scrape='true' + prometheus.io/scheme=https + service.alpha.openshift.io/serving-cert-secret-name=prometheus-tls - name: annotate alerts service command: > diff --git a/roles/openshift_sanitize_inventory/tasks/main.yml b/roles/openshift_sanitize_inventory/tasks/main.yml index 59ce505d3..47d7be05a 100644 --- a/roles/openshift_sanitize_inventory/tasks/main.yml +++ b/roles/openshift_sanitize_inventory/tasks/main.yml @@ -12,6 +12,27 @@ deployment_type is deprecated in favor of openshift_deployment_type. Please specify only openshift_deployment_type, or make both the same. +# osm_cluster_network_cidr, osm_host_subnet_length and openshift_portal_net are +# now required to avoid changes that may occur between releases +# +# Note: We will skip these checks when some tests run which don't +# actually do any insalling/upgrading/scaling/etc.. +# Reference: https://bugzilla.redhat.com/show_bug.cgi?id=1451023 +- when: + - not testing_skip_some_requirements|default(False)|bool + assert: + that: + - "osm_cluster_network_cidr is defined" + - "osm_host_subnet_length is defined" + - "openshift_portal_net is defined" + msg: > + osm_cluster_network_cidr, osm_host_subnet_length, and openshift_portal_net are required inventory + variables. If you are upgrading or scaling up these variables should match what is currently used + in the cluster. If you don't remember what these values are you can find them in + /etc/origin/master/master-config.yaml on a master with the names clusterNetworkCIDR + (osm_cluster_network_cidr), hostSubnetLength (osm_host_subnet_length), + and serviceNetworkCIDR (openshift_portal_net). + - name: Standardize on latest variable names set_fact: # goal is to deprecate deployment_type in favor of openshift_deployment_type. diff --git a/roles/os_firewall/tasks/iptables.yml b/roles/os_firewall/tasks/iptables.yml index 0af5abf38..2d74f2e48 100644 --- a/roles/os_firewall/tasks/iptables.yml +++ b/roles/os_firewall/tasks/iptables.yml @@ -33,7 +33,7 @@ register: result delegate_to: "{{item}}" run_once: true - with_items: "{{ ansible_play_hosts }}" + with_items: "{{ ansible_play_batch }}" - name: need to pause here, otherwise the iptables service starting can sometimes cause ssh to fail pause: diff --git a/test/integration/openshift_health_checker/common.go b/test/integration/openshift_health_checker/common.go index a92d6861d..8b79c48cb 100644 --- a/test/integration/openshift_health_checker/common.go +++ b/test/integration/openshift_health_checker/common.go @@ -25,7 +25,7 @@ func (p PlaybookTest) Run(t *testing.T) { // A PlaybookTest is intended to be run in parallel with other tests. t.Parallel() - cmd := exec.Command("ansible-playbook", "-i", "/dev/null", p.Path) + cmd := exec.Command("ansible-playbook", "-e", "testing_skip_some_requirements=1", "-i", "/dev/null", p.Path) cmd.Env = append(os.Environ(), "ANSIBLE_FORCE_COLOR=1") b, err := cmd.CombinedOutput() |