#!/usr/bin/env python2
# vim: expandtab:tabstop=4:shiftwidth=4
import argparse
import ConfigParser
import os
import sys
import subprocess
import traceback
class Cluster(object):
"""
Provide Command, Control and Configuration (c3) Interface for OpenShift Clusters
"""
def __init__(self):
# setup ansible ssh environment
if 'ANSIBLE_SSH_ARGS' not in os.environ:
os.environ['ANSIBLE_SSH_ARGS'] = (
'-o ForwardAgent=yes '
'-o StrictHostKeyChecking=no '
'-o UserKnownHostsFile=/dev/null '
'-o ControlMaster=auto '
'-o ControlPersist=600s '
)
# Because of `UserKnownHostsFile=/dev/null`
# our `.ssh/known_hosts` file most probably misses the ssh host public keys
# of our servers.
# In that case, ansible serializes the execution of ansible modules
# because we might be interactively prompted to accept the ssh host public keys.
# Because of `StrictHostKeyChecking=no` we know that we won't be prompted
# So, we don't want our modules execution to be serialized.
os.environ['ANSIBLE_HOST_KEY_CHECKING'] = 'False'
# TODO: A more secure way to proceed would consist in dynamically
# retrieving the ssh host public keys from the IaaS interface
if 'ANSIBLE_SSH_PIPELINING' not in os.environ:
os.environ['ANSIBLE_SSH_PIPELINING'] = 'True'
def get_deployment_type(self, args):
"""
Get the deployment_type based on the environment variables and the
command line arguments
:param args: command line arguments provided by the user
:return: string representing the deployment type
"""
deployment_type = 'origin'
if args.deployment_type:
deployment_type = args.deployment_type
elif 'OS_DEPLOYMENT_TYPE' in os.environ:
deployment_type = os.environ['OS_DEPLOYMENT_TYPE']
return deployment_type
def create(self, args):
"""
Create an OpenShift cluster for given provider
:param args: command line arguments provided by user
"""
cluster = {'cluster_id': args.cluster_id,
'deployment_type': self.get_deployment_type(args)}
playbook = "playbooks/{0}/openshift-cluster/launch.yml".format(args.provider)
inventory = self.setup_provider(args.provider)
cluster['num_masters'] = args.masters
cluster['num_nodes'] = args.nodes
cluster['num_infra'] = args.infra
cluster['num_etcd'] = args.etcd
cluster['cluster_env'] = args.env
if args.cloudprovider and args.provider == 'openstack':
cluster['openshift_cloudprovider_kind'] = 'openstack'
cluster['openshift_cloudprovider_openstack_auth_url'] = os.getenv('OS_AUTH_URL')
cluster['openshift_cloudprovider_openstack_username'] = os.getenv('OS_USERNAME')
cluster['openshift_cloudprovider_openstack_password'] = os.getenv('OS_PASSWORD')
if 'OS_USER_DOMAIN_ID' in os.environ:
cluster['openshift_cloudprovider_openstack_domain_id'] = os.getenv('OS_USER_DOMAIN_ID')
if 'OS_USER_DOMAIN_NAME' in os.environ:
cluster['openshift_cloudprovider_openstack_domain_name'] = os.getenv('OS_USER_DOMAIN_NAME')
if 'OS_PROJECT_ID' in os.environ or 'OS_TENANT_ID' in os.environ:
cluster['openshift_cloudprovider_openstack_tenant_id'] = os.getenv('OS_PROJECT_ID',os.getenv('OS_TENANT_ID'))
if 'OS_PROJECT_NAME' is os.environ or 'OS_TENANT_NAME' in os.environ:
cluster['openshift_cloudprovider_openstack_tenant_name'] = os.getenv('OS_PROJECT_NAME',os.getenv('OS_TENANT_NAME'))
if 'OS_REGION_NAME' in os.environ:
cluster['openshift_cloudprovider_openstack_region'] = os.getenv('OS_REGION_NAME')
self.action(args, inventory, cluster, playbook)
def add_nodes(self, args):
"""
Add nodes to an existing cluster for given provider
:param args: command line arguments provided by user
"""
cluster = {'cluster_id': args.cluster_id,
'deployment_type': self.get_deployment_type(args),
}
playbook = "playbooks/{0}/openshift-cluster/add_nodes.yml".format(args.provider)
inventory = self.setup_provider(args.provider)
cluster['num_nodes'] = args.nodes
cluster['num_infra'] = args.infra
cluster['cluster_env'] = args.env
self.action(args, inventory, cluster, playbook)
def terminate(self, args):
"""
Destroy OpenShift cluster
:param args: command line arguments provided by user
"""
cluster = {'cluster_id': args.cluster_id,
'deployment_type': self.get_deployment_type(args),
'cluster_env': args.env,
}
playbook = "playbooks/{0}/openshift-cluster/terminate.yml".format(args.provider)
inventory = self.setup_provider(args.provider)
self.action(args, inventory, cluster, playbook)
def list(self, args):
"""
List VMs in cluster
:param args: command line arguments provided by user
"""
cluster = {'cluster_id': args.cluster_id,
'deployment_type': self.get_deployment_type(args),
'cluster_env': args.env,
}
playbook = "playbooks/{0}/openshift-cluster/list.yml".format(args.provider)
inventory = self.setup_provider(args.provider)
self.action(args, inventory, cluster, playbook)
def config(self, args):
"""
Configure or reconfigure OpenShift across clustered VMs
:param args: command line arguments provided by user
"""
cluster = {'cluster_id': args.cluster_id,
'deployment_type': self.get_deployment_type(args),
'cluster_env': args.env,
}
playbook = "playbooks/{0}/openshift-cluster/config.yml".format(args.provider)
inventory = self.setup_provider(args.provider)
self.action(args, inventory, cluster, playbook)
def update(self, args):
"""
Update to latest OpenShift across clustered VMs
:param args: command line arguments provided by user
"""
cluster = {'cluster_id': args.cluster_id,
'deployment_type': self.get_deployment_type(args),
'cluster_env': args.env,
}
playbook = "playbooks/{0}/openshift-cluster/update.yml".format(args.provider)
inventory = self.setup_provider(args.provider)
self.action(args, inventory, cluster, playbook)
def service(self, args):
"""
Make the same service call across all nodes in the cluster
:param args: command line arguments provided by user
"""
cluster = {'cluster_id': args.cluster_id,
'deployment_type': self.get_deployment_type(args),
'new_cluster_state': args.state,
'cluster_env': args.env,
}
playbook = "playbooks/{0}/openshift-cluster/service.yml".format(args.provider)
inventory = self.setup_provider(args.provider)
self.action(args, inventory, cluster, playbook)
def setup_provider(self, provider):
"""
Setup ansible playbook environment
:param provider: command line arguments provided by user
:return: path to inventory for given provider
"""
config = ConfigParser.ConfigParser()
if 'gce' == provider:
gce_ini_default_path = os.path.join('inventory/gce/hosts/gce.ini')
gce_ini_path = os.environ.get('GCE_INI_PATH', gce_ini_default_path)
if os.path.exists(gce_ini_path):
config.readfp(open(gce_ini_path))
for key in config.options('gce'):
os.environ[key] = config.get('gce', key)
inventory = '-i inventory/gce/hosts'
elif 'aws' == provider:
config.readfp(open('inventory/aws/hosts/ec2.ini'))
for key in config.options('ec2'):
os.environ[key] = config.get('ec2', key)
inventory = '-i inventory/aws/hosts'
key_vars = ['AWS_ACCESS_KEY_ID', 'AWS_SECRET_ACCESS_KEY']
key_missing = [key for key in key_vars if key not in os.environ]
boto_conf_files = ['~/.aws/credentials', '~/.boto']
conf_exists = lambda conf: os.path.isfile(os.path.expanduser(conf))
boto_configs = [conf for conf in boto_conf_files if conf_exists(conf)]
if len(key_missing) > 0 and len(boto_configs) == 0:
raise ValueError("PROVIDER aws requires {0} environment variable(s). See README_AWS.md".format(key_missing))
elif 'libvirt' == provider:
inventory = '-i inventory/libvirt/hosts'
elif 'openstack' == provider:
inventory = '-i inventory/openstack/hosts'
else:
# this code should never be reached
raise ValueError("invalid PROVIDER {0}".format(provider))
return inventory
def action(self, args, inventory, cluster, playbook):
"""
Build ansible-playbook command line and execute
:param args: command line arguments provided by user
:param inventory: derived provider library
:param cluster: cluster variables for kubernetes
:param playbook: ansible playbook to execute
"""
verbose = ''
if args.verbose > 0:
verbose = '-{0}'.format('v' * args.verbose)
if args.option:
for opt in args.option:
k, v = opt.split('=', 1)
cluster['cli_' + k] = v
ansible_extra_vars = '-e \'{0}\''.format(
' '.join(['%s=%s' % (key, value) for (key, value) in cluster.items()])
)
command = 'ansible-playbook {0} {1} {2} {3}'.format(
verbose, inventory, ansible_extra_vars, playbook
)
if args.profile:
command = 'ANSIBLE_CALLBACK_PLUGINS=ansible-profile/callback_plugins ' + command
if args.verbose > 1:
command = 'time {0}'.format(command)
if args.verbose > 0:
sys.stderr.write('RUN [{0}]\n'.format(command))
sys.stderr.flush()
try:
subprocess.check_call(command, shell=True)
except subprocess.CalledProcessError as exc:
raise ActionFailed("ACTION [{0}] failed: {1}"
.format(args.action, exc))
class ActionFailed(Exception):
"""
Raised when action failed.
"""
pass
if __name__ == '__main__':
"""
User command to invoke ansible playbooks in a "known" configuration
Reads ~/.openshift-ansible for default configuration items
[DEFAULT]
validate_cluster_ids = False
cluster_ids = marketing,sales
providers = gce,aws,libvirt,openstack
"""
warning = ("================================================================================\n"
"ATTENTION: You are running a community supported utility that has not been\n"
"tested by Red Hat. Visit https://docs.openshift.com for supported installation\n"
"instructions.\n"
"================================================================================\n\n")
sys.stderr.write(warning)
cluster_config = ConfigParser.SafeConfigParser({
'cluster_ids': 'marketing,sales',
'validate_cluster_ids': 'False',
'providers': 'gce,aws,libvirt,openstack',
})
path = os.path.expanduser("~/.openshift-ansible")
if os.path.isfile(path):
cluster_config.read(path)
cluster = Cluster()
parser = argparse.ArgumentParser(
formatter_class=argparse.RawDescriptionHelpFormatter,
description='Python wrapper to ensure proper configuration for OpenShift ansible playbooks',
epilog='''\
This wrapper is overriding the following ansible variables:
* ANSIBLE_SSH_ARGS:
If not set in the environment, this wrapper will use the following value:
`-o ForwardAgent=yes -o StrictHostKeyChecking=no -o UserKnownHostsFile=/dev/null -o ControlMaster=auto -o ControlPersist=600s`
If set in the environment, the environment variable value is left untouched and used.
* ANSIBLE_SSH_PIPELINING:
If not set in the environment, this wrapper will set it to `True`.
If you experience issues with Ansible SSH pipelining, you can disable it by explicitly setting this environment variable to `False`.
'''
)
parser.add_argument('-v', '--verbose', action='count',
help='Multiple -v options increase the verbosity')
parser.add_argument('--version', action='version', version='%(prog)s 0.3')
meta_parser = argparse.ArgumentParser(add_help=False)
providers = cluster_config.get('DEFAULT', 'providers').split(',')
meta_parser.add_argument('provider', choices=providers, help='provider')
if cluster_config.get('DEFAULT', 'validate_cluster_ids').lower() in ("yes", "true", "1"):
meta_parser.add_argument('cluster_id', choices=cluster_config.get('DEFAULT', 'cluster_ids').split(','),
help='prefix for cluster VM names')
else:
meta_parser.add_argument('cluster_id', help='prefix for cluster VM names')
meta_parser.add_argument('-t', '--deployment-type',
choices=['origin', 'atomic-enterprise', 'openshift-enterprise'],
help='Deployment type. (default: origin)')
meta_parser.add_argument('-o', '--option', action='append',
help='options')
meta_parser.add_argument('--env', default='dev', type=str,
help='environment for the cluster. Defaults to \'dev\'.')
meta_parser.add_argument('-p', '--profile', action='store_true',
help='Enable playbook profiling')
action_parser = parser.add_subparsers(dest='action', title='actions',
description='Choose from valid actions')
create_parser = action_parser.add_parser('create', help='Create a cluster',
parents=[meta_parser])
create_parser.add_argument('-c', '--cloudprovider', action='store_true',
help='Enable the cloudprovider')
create_parser.add_argument('-m', '--masters', default=1, type=int,
help='number of masters to create in cluster')
create_parser.add_argument('-n', '--nodes', default=2, type=int,
help='number of nodes to create in cluster')
create_parser.add_argument('-i', '--infra', default=1, type=int,
help='number of infra nodes to create in cluster')
create_parser.add_argument('-e', '--etcd', default=0, type=int,
help='number of external etcd hosts to create in cluster')
create_parser.set_defaults(func=cluster.create)
create_parser = action_parser.add_parser('add-nodes', help='Add nodes to a cluster',
parents=[meta_parser])
create_parser.add_argument('-n', '--nodes', default=1, type=int,
help='number of nodes to add to the cluster')
create_parser.add_argument('-i', '--infra', default=1, type=int,
help='number of infra nodes to add to the cluster')
create_parser.set_defaults(func=cluster.add_nodes)
config_parser = action_parser.add_parser('config',
help='Configure or reconfigure a cluster',
parents=[meta_parser])
config_parser.set_defaults(func=cluster.config)
terminate_parser = action_parser.add_parser('terminate',
help='Destroy a cluster',
parents=[meta_parser])
terminate_parser.add_argument('-f', '--force', action='store_true',
help='Destroy cluster without confirmation')
terminate_parser.set_defaults(func=cluster.terminate)
update_parser = action_parser.add_parser('update',
help='Update OpenShift across cluster',
parents=[meta_parser])
update_parser.add_argument('-f', '--force', action='store_true',
help='Update cluster without confirmation')
update_parser.set_defaults(func=cluster.update)
list_parser = action_parser.add_parser('list', help='List VMs in cluster',
parents=[meta_parser])
list_parser.set_defaults(func=cluster.list)
service_parser = action_parser.add_parser('service', help='service for openshift across cluster',
parents=[meta_parser])
# choices are the only ones valid for the ansible service module: http://docs.ansible.com/service_module.html
service_parser.add_argument('state', choices=['started', 'stopped', 'restarted', 'reloaded'],
help='make service call across cluster')
service_parser.set_defaults(func=cluster.service)
args = parser.parse_args()
if 'terminate' == args.action and not args.force:
answer = raw_input("This will destroy the ENTIRE {0} cluster. Are you sure? [y/N] ".format(args.cluster_id))
if answer not in ['y', 'Y']:
sys.stderr.write('\nACTION [terminate] aborted by user!\n')
exit(1)
if 'update' == args.action and not args.force:
answer = raw_input(
"This is destructive and could corrupt {0} cluster. Continue? [y/N] ".format(args.cluster_id))
if answer not in ['y', 'Y']:
sys.stderr.write('\nACTION [update] aborted by user!\n')
exit(1)
try:
args.func(args)
except Exception as exc:
if args.verbose:
traceback.print_exc(file=sys.stderr)
else:
print >>sys.stderr, exc
exit(1)