diff options
Diffstat (limited to 'inventory')
| -rw-r--r-- | inventory/aws/hosts/ec2.ini | 54 | ||||
| -rwxr-xr-x | inventory/aws/hosts/ec2.py | 285 | ||||
| -rwxr-xr-x | inventory/gce/hosts/gce.py | 252 |
3 files changed, 494 insertions, 97 deletions
diff --git a/inventory/aws/hosts/ec2.ini b/inventory/aws/hosts/ec2.ini index 5ee51c84f..64c097d47 100644 --- a/inventory/aws/hosts/ec2.ini +++ b/inventory/aws/hosts/ec2.ini @@ -29,17 +29,32 @@ regions_exclude = us-gov-west-1,cn-north-1 # in the event of a collision. destination_variable = public_dns_name +# This allows you to override the inventory_name with an ec2 variable, instead +# of using the destination_variable above. Addressing (aka ansible_ssh_host) +# will still use destination_variable. Tags should be written as 'tag_TAGNAME'. +hostname_variable = tag_Name + # For server inside a VPC, using DNS names may not make sense. When an instance # has 'subnet_id' set, this variable is used. If the subnet is public, setting # this to 'ip_address' will return the public IP address. For instances in a # private subnet, this should be set to 'private_ip_address', and Ansible must # be run from within EC2. The key of an EC2 tag may optionally be used; however # the boto instance variables hold precedence in the event of a collision. -# WARNING: - instances that are in the private vpc, _without_ public ip address +# WARNING: - instances that are in the private vpc, _without_ public ip address # will not be listed in the inventory until You set: -# vpc_destination_variable = 'private_ip_address' +# vpc_destination_variable = private_ip_address vpc_destination_variable = ip_address +# The following two settings allow flexible ansible host naming based on a +# python format string and a comma-separated list of ec2 tags. Note that: +# +# 1) If the tags referenced are not present for some instances, empty strings +# will be substituted in the format string. +# 2) This overrides both destination_variable and vpc_destination_variable. +# +#destination_format = {0}.{1}.example.com +#destination_format_tags = Name,environment + # To tag instances on EC2 with the resource records that point to them from # Route53, uncomment and set 'route53' to True. route53 = False @@ -67,6 +82,9 @@ all_instances = False # 'all_rds_instances' to True return all RDS instances regardless of state. all_rds_instances = False +# Include RDS cluster information (Aurora etc.) +include_rds_clusters = False + # By default, only ElastiCache clusters and nodes in the 'available' state # are returned. Set 'all_elasticache_clusters' and/or 'all_elastic_nodes' # to True return all ElastiCache clusters and nodes, regardless of state. @@ -91,19 +109,16 @@ cache_path = ~/.ansible/tmp # To disable the cache, set this value to 0 cache_max_age = 300 -# These two settings allow flexible ansible host naming based on a format -# string and a comma-separated list of ec2 tags. The tags used must be -# present for all instances, or the code will fail. This overrides both -# destination_variable and vpc_destination_variable. -# destination_format = {0}.{1}.rhcloud.com -# destination_format_tags = Name,environment - # Organize groups into a nested/hierarchy instead of a flat namespace. nested_groups = False # Replace - tags when creating groups to avoid issues with ansible replace_dash_in_groups = False +# If set to true, any tag of the form "a,b,c" is expanded into a list +# and the results are used to create additional tag_* inventory groups. +expand_csv_tags = False + # The EC2 inventory output can become very large. To manage its size, # configure which groups should be created. group_by_instance_id = True @@ -147,9 +162,28 @@ group_by_elasticache_replication_group = True # You can use wildcards in filter values also. Below will list instances which # tag Name value matches webservers1* -# (ex. webservers15, webservers1a, webservers123 etc) +# (ex. webservers15, webservers1a, webservers123 etc) # instance_filters = tag:Name=webservers1* # A boto configuration profile may be used to separate out credentials # see http://boto.readthedocs.org/en/latest/boto_config_tut.html # boto_profile = some-boto-profile-name + + +[credentials] + +# The AWS credentials can optionally be specified here. Credentials specified +# here are ignored if the environment variable AWS_ACCESS_KEY_ID or +# AWS_PROFILE is set, or if the boto_profile property above is set. +# +# Supplying AWS credentials here is not recommended, as it introduces +# non-trivial security concerns. When going down this route, please make sure +# to set access permissions for this file correctly, e.g. handle it the same +# way as you would a private SSH key. +# +# Unlike the boto and AWS configure files, this section does not support +# profiles. +# +# aws_access_key_id = AXXXXXXXXXXXXXX +# aws_secret_access_key = XXXXXXXXXXXXXXXXXXX +# aws_security_token = XXXXXXXXXXXXXXXXXXXXXXXXXXXX diff --git a/inventory/aws/hosts/ec2.py b/inventory/aws/hosts/ec2.py index 7dfcd7889..b71458a29 100755 --- a/inventory/aws/hosts/ec2.py +++ b/inventory/aws/hosts/ec2.py @@ -38,6 +38,7 @@ When run against a specific host, this script returns the following variables: - ec2_attachTime - ec2_attachment - ec2_attachmentId + - ec2_block_devices - ec2_client_token - ec2_deleteOnTermination - ec2_description @@ -132,6 +133,15 @@ from boto import elasticache from boto import route53 import six +from ansible.module_utils import ec2 as ec2_utils + +HAS_BOTO3 = False +try: + import boto3 + HAS_BOTO3 = True +except ImportError: + pass + from six.moves import configparser from collections import defaultdict @@ -142,6 +152,7 @@ except ImportError: class Ec2Inventory(object): + def _empty_inventory(self): return {"_meta" : {"hostvars" : {}}} @@ -158,6 +169,9 @@ class Ec2Inventory(object): # Boto profile to use (if any) self.boto_profile = None + # AWS credentials. + self.credentials = {} + # Read settings and parse CLI arguments self.parse_cli_args() self.read_settings() @@ -225,7 +239,7 @@ class Ec2Inventory(object): configRegions_exclude = config.get('ec2', 'regions_exclude') if (configRegions == 'all'): if self.eucalyptus_host: - self.regions.append(boto.connect_euca(host=self.eucalyptus_host).region.name) + self.regions.append(boto.connect_euca(host=self.eucalyptus_host).region.name, **self.credentials) else: for regionInfo in ec2.regions(): if regionInfo.name not in configRegions_exclude: @@ -237,6 +251,11 @@ class Ec2Inventory(object): self.destination_variable = config.get('ec2', 'destination_variable') self.vpc_destination_variable = config.get('ec2', 'vpc_destination_variable') + if config.has_option('ec2', 'hostname_variable'): + self.hostname_variable = config.get('ec2', 'hostname_variable') + else: + self.hostname_variable = None + if config.has_option('ec2', 'destination_format') and \ config.has_option('ec2', 'destination_format_tags'): self.destination_format = config.get('ec2', 'destination_format') @@ -257,6 +276,12 @@ class Ec2Inventory(object): if config.has_option('ec2', 'rds'): self.rds_enabled = config.getboolean('ec2', 'rds') + # Include RDS cluster instances? + if config.has_option('ec2', 'include_rds_clusters'): + self.include_rds_clusters = config.getboolean('ec2', 'include_rds_clusters') + else: + self.include_rds_clusters = False + # Include ElastiCache instances? self.elasticache_enabled = True if config.has_option('ec2', 'elasticache'): @@ -319,6 +344,29 @@ class Ec2Inventory(object): if config.has_option('ec2', 'boto_profile') and not self.boto_profile: self.boto_profile = config.get('ec2', 'boto_profile') + # AWS credentials (prefer environment variables) + if not (self.boto_profile or os.environ.get('AWS_ACCESS_KEY_ID') or + os.environ.get('AWS_PROFILE')): + if config.has_option('credentials', 'aws_access_key_id'): + aws_access_key_id = config.get('credentials', 'aws_access_key_id') + else: + aws_access_key_id = None + if config.has_option('credentials', 'aws_secret_access_key'): + aws_secret_access_key = config.get('credentials', 'aws_secret_access_key') + else: + aws_secret_access_key = None + if config.has_option('credentials', 'aws_security_token'): + aws_security_token = config.get('credentials', 'aws_security_token') + else: + aws_security_token = None + if aws_access_key_id: + self.credentials = { + 'aws_access_key_id': aws_access_key_id, + 'aws_secret_access_key': aws_secret_access_key + } + if aws_security_token: + self.credentials['security_token'] = aws_security_token + # Cache related cache_dir = os.path.expanduser(config.get('ec2', 'cache_path')) if self.boto_profile: @@ -326,10 +374,22 @@ class Ec2Inventory(object): if not os.path.exists(cache_dir): os.makedirs(cache_dir) - self.cache_path_cache = cache_dir + "/ansible-ec2.cache" - self.cache_path_index = cache_dir + "/ansible-ec2.index" + cache_name = 'ansible-ec2' + aws_profile = lambda: (self.boto_profile or + os.environ.get('AWS_PROFILE') or + os.environ.get('AWS_ACCESS_KEY_ID') or + self.credentials.get('aws_access_key_id', None)) + if aws_profile(): + cache_name = '%s-%s' % (cache_name, aws_profile()) + self.cache_path_cache = cache_dir + "/%s.cache" % cache_name + self.cache_path_index = cache_dir + "/%s.index" % cache_name self.cache_max_age = config.getint('ec2', 'cache_max_age') + if config.has_option('ec2', 'expand_csv_tags'): + self.expand_csv_tags = config.getboolean('ec2', 'expand_csv_tags') + else: + self.expand_csv_tags = False + # Configure nested groups instead of flat namespace. if config.has_option('ec2', 'nested_groups'): self.nested_groups = config.getboolean('ec2', 'nested_groups') @@ -391,7 +451,10 @@ class Ec2Inventory(object): # Instance filters (see boto and EC2 API docs). Ignore invalid filters. self.ec2_instance_filters = defaultdict(list) if config.has_option('ec2', 'instance_filters'): - for instance_filter in config.get('ec2', 'instance_filters', '').split(','): + + filters = [f for f in config.get('ec2', 'instance_filters').split(',') if f] + + for instance_filter in filters: instance_filter = instance_filter.strip() if not instance_filter or '=' not in instance_filter: continue @@ -410,7 +473,7 @@ class Ec2Inventory(object): help='Get all the variables about a specific instance') parser.add_argument('--refresh-cache', action='store_true', default=False, help='Force refresh of cache by making API requests to EC2 (default: False - use cache files)') - parser.add_argument('--boto-profile', action='store', + parser.add_argument('--profile', '--boto-profile', action='store', dest='boto_profile', help='Use boto profile for connections to EC2') self.args = parser.parse_args() @@ -428,6 +491,8 @@ class Ec2Inventory(object): if self.elasticache_enabled: self.get_elasticache_clusters_by_region(region) self.get_elasticache_replication_groups_by_region(region) + if self.include_rds_clusters: + self.include_rds_clusters_by_region(region) self.write_to_cache(self.inventory, self.cache_path_cache) self.write_to_cache(self.index, self.cache_path_index) @@ -435,7 +500,7 @@ class Ec2Inventory(object): def connect(self, region): ''' create connection to api server''' if self.eucalyptus: - conn = boto.connect_euca(host=self.eucalyptus_host) + conn = boto.connect_euca(host=self.eucalyptus_host, **self.credentials) conn.APIVersion = '2010-08-31' else: conn = self.connect_to_aws(ec2, region) @@ -449,7 +514,7 @@ class Ec2Inventory(object): return connect_args def connect_to_aws(self, module, region): - connect_args = {} + connect_args = self.credentials # only pass the profile name if it's set (as it is not supported by older boto versions) if self.boto_profile: @@ -475,8 +540,25 @@ class Ec2Inventory(object): else: reservations = conn.get_all_instances() + # Pull the tags back in a second step + # AWS are on record as saying that the tags fetched in the first `get_all_instances` request are not + # reliable and may be missing, and the only way to guarantee they are there is by calling `get_all_tags` + instance_ids = [] + for reservation in reservations: + instance_ids.extend([instance.id for instance in reservation.instances]) + + max_filter_value = 199 + tags = [] + for i in range(0, len(instance_ids), max_filter_value): + tags.extend(conn.get_all_tags(filters={'resource-type': 'instance', 'resource-id': instance_ids[i:i+max_filter_value]})) + + tags_by_instance_id = defaultdict(dict) + for tag in tags: + tags_by_instance_id[tag.res_id][tag.name] = tag.value + for reservation in reservations: for instance in reservation.instances: + instance.tags = tags_by_instance_id[instance.id] self.add_instance(instance, region) except boto.exception.BotoServerError as e: @@ -494,9 +576,14 @@ class Ec2Inventory(object): try: conn = self.connect_to_aws(rds, region) if conn: - instances = conn.get_all_dbinstances() - for instance in instances: - self.add_rds_instance(instance, region) + marker = None + while True: + instances = conn.get_all_dbinstances(marker=marker) + marker = instances.marker + for instance in instances: + self.add_rds_instance(instance, region) + if not marker: + break except boto.exception.BotoServerError as e: error = e.reason @@ -506,6 +593,65 @@ class Ec2Inventory(object): error = "Looks like AWS RDS is down:\n%s" % e.message self.fail_with_error(error, 'getting RDS instances') + def include_rds_clusters_by_region(self, region): + if not HAS_BOTO3: + self.fail_with_error("Working with RDS clusters requires boto3 - please install boto3 and try again", + "getting RDS clusters") + + client = ec2_utils.boto3_inventory_conn('client', 'rds', region, **self.credentials) + + marker, clusters = '', [] + while marker is not None: + resp = client.describe_db_clusters(Marker=marker) + clusters.extend(resp["DBClusters"]) + marker = resp.get('Marker', None) + + account_id = boto.connect_iam().get_user().arn.split(':')[4] + c_dict = {} + for c in clusters: + # remove these datetime objects as there is no serialisation to json + # currently in place and we don't need the data yet + if 'EarliestRestorableTime' in c: + del c['EarliestRestorableTime'] + if 'LatestRestorableTime' in c: + del c['LatestRestorableTime'] + + if self.ec2_instance_filters == {}: + matches_filter = True + else: + matches_filter = False + + try: + # arn:aws:rds:<region>:<account number>:<resourcetype>:<name> + tags = client.list_tags_for_resource( + ResourceName='arn:aws:rds:' + region + ':' + account_id + ':cluster:' + c['DBClusterIdentifier']) + c['Tags'] = tags['TagList'] + + if self.ec2_instance_filters: + for filter_key, filter_values in self.ec2_instance_filters.items(): + # get AWS tag key e.g. tag:env will be 'env' + tag_name = filter_key.split(":", 1)[1] + # Filter values is a list (if you put multiple values for the same tag name) + matches_filter = any(d['Key'] == tag_name and d['Value'] in filter_values for d in c['Tags']) + + if matches_filter: + # it matches a filter, so stop looking for further matches + break + + except Exception as e: + if e.message.find('DBInstanceNotFound') >= 0: + # AWS RDS bug (2016-01-06) means deletion does not fully complete and leave an 'empty' cluster. + # Ignore errors when trying to find tags for these + pass + + # ignore empty clusters caused by AWS bug + if len(c['DBClusterMembers']) == 0: + continue + elif matches_filter: + c_dict[c['DBClusterIdentifier']] = c + + self.inventory['db_clusters'] = c_dict + def get_elasticache_clusters_by_region(self, region): ''' Makes an AWS API call to the list of ElastiCache clusters (with nodes' info) in a particular region.''' @@ -514,7 +660,7 @@ class Ec2Inventory(object): # that's why we need to call describe directly (it would be called by # the shorthand method anyway...) try: - conn = elasticache.connect_to_region(region) + conn = self.connect_to_aws(elasticache, region) if conn: # show_cache_node_info = True # because we also want nodes' information @@ -531,7 +677,7 @@ class Ec2Inventory(object): try: # Boto also doesn't provide wrapper classes to CacheClusters or - # CacheNodes. Because of that wo can't make use of the get_list + # CacheNodes. Because of that we can't make use of the get_list # method in the AWSQueryConnection. Let's do the work manually clusters = response['DescribeCacheClustersResponse']['DescribeCacheClustersResult']['CacheClusters'] @@ -550,7 +696,7 @@ class Ec2Inventory(object): # that's why we need to call describe directly (it would be called by # the shorthand method anyway...) try: - conn = elasticache.connect_to_region(region) + conn = self.connect_to_aws(elasticache, region) if conn: response = conn.describe_replication_groups() @@ -565,7 +711,7 @@ class Ec2Inventory(object): try: # Boto also doesn't provide wrapper classes to ReplicationGroups - # Because of that wo can't make use of the get_list method in the + # Because of that we can't make use of the get_list method in the # AWSQueryConnection. Let's do the work manually replication_groups = response['DescribeReplicationGroupsResponse']['DescribeReplicationGroupsResult']['ReplicationGroups'] @@ -619,7 +765,7 @@ class Ec2Inventory(object): # Select the best destination address if self.destination_format and self.destination_format_tags: - dest = self.destination_format.format(*[ getattr(instance, 'tags').get(tag, 'nil') for tag in self.destination_format_tags ]) + dest = self.destination_format.format(*[ getattr(instance, 'tags').get(tag, '') for tag in self.destination_format_tags ]) elif instance.subnet_id: dest = getattr(instance, self.vpc_destination_variable, None) if dest is None: @@ -633,32 +779,46 @@ class Ec2Inventory(object): # Skip instances we cannot address (e.g. private VPC subnet) return + # Set the inventory name + hostname = None + if self.hostname_variable: + if self.hostname_variable.startswith('tag_'): + hostname = instance.tags.get(self.hostname_variable[4:], None) + else: + hostname = getattr(instance, self.hostname_variable) + + # If we can't get a nice hostname, use the destination address + if not hostname: + hostname = dest + else: + hostname = self.to_safe(hostname).lower() + # if we only want to include hosts that match a pattern, skip those that don't - if self.pattern_include and not self.pattern_include.match(dest): + if self.pattern_include and not self.pattern_include.match(hostname): return # if we need to exclude hosts that match a pattern, skip those - if self.pattern_exclude and self.pattern_exclude.match(dest): + if self.pattern_exclude and self.pattern_exclude.match(hostname): return # Add to index - self.index[dest] = [region, instance.id] + self.index[hostname] = [region, instance.id] # Inventory: Group by instance ID (always a group of 1) if self.group_by_instance_id: - self.inventory[instance.id] = [dest] + self.inventory[instance.id] = [hostname] if self.nested_groups: self.push_group(self.inventory, 'instances', instance.id) # Inventory: Group by region if self.group_by_region: - self.push(self.inventory, region, dest) + self.push(self.inventory, region, hostname) if self.nested_groups: self.push_group(self.inventory, 'regions', region) # Inventory: Group by availability zone if self.group_by_availability_zone: - self.push(self.inventory, instance.placement, dest) + self.push(self.inventory, instance.placement, hostname) if self.nested_groups: if self.group_by_region: self.push_group(self.inventory, region, instance.placement) @@ -667,28 +827,28 @@ class Ec2Inventory(object): # Inventory: Group by Amazon Machine Image (AMI) ID if self.group_by_ami_id: ami_id = self.to_safe(instance.image_id) - self.push(self.inventory, ami_id, dest) + self.push(self.inventory, ami_id, hostname) if self.nested_groups: self.push_group(self.inventory, 'images', ami_id) # Inventory: Group by instance type if self.group_by_instance_type: type_name = self.to_safe('type_' + instance.instance_type) - self.push(self.inventory, type_name, dest) + self.push(self.inventory, type_name, hostname) if self.nested_groups: self.push_group(self.inventory, 'types', type_name) # Inventory: Group by key pair if self.group_by_key_pair and instance.key_name: key_name = self.to_safe('key_' + instance.key_name) - self.push(self.inventory, key_name, dest) + self.push(self.inventory, key_name, hostname) if self.nested_groups: self.push_group(self.inventory, 'keys', key_name) # Inventory: Group by VPC if self.group_by_vpc_id and instance.vpc_id: vpc_id_name = self.to_safe('vpc_id_' + instance.vpc_id) - self.push(self.inventory, vpc_id_name, dest) + self.push(self.inventory, vpc_id_name, hostname) if self.nested_groups: self.push_group(self.inventory, 'vpcs', vpc_id_name) @@ -697,7 +857,7 @@ class Ec2Inventory(object): try: for group in instance.groups: key = self.to_safe("security_group_" + group.name) - self.push(self.inventory, key, dest) + self.push(self.inventory, key, hostname) if self.nested_groups: self.push_group(self.inventory, 'security_groups', key) except AttributeError: @@ -707,34 +867,41 @@ class Ec2Inventory(object): # Inventory: Group by tag keys if self.group_by_tag_keys: for k, v in instance.tags.items(): - if v: - key = self.to_safe("tag_" + k + "=" + v) + if self.expand_csv_tags and v and ',' in v: + values = map(lambda x: x.strip(), v.split(',')) else: - key = self.to_safe("tag_" + k) - self.push(self.inventory, key, dest) - if self.nested_groups: - self.push_group(self.inventory, 'tags', self.to_safe("tag_" + k)) + values = [v] + + for v in values: if v: - self.push_group(self.inventory, self.to_safe("tag_" + k), key) + key = self.to_safe("tag_" + k + "=" + v) + else: + key = self.to_safe("tag_" + k) + self.push(self.inventory, key, hostname) + if self.nested_groups: + self.push_group(self.inventory, 'tags', self.to_safe("tag_" + k)) + if v: + self.push_group(self.inventory, self.to_safe("tag_" + k), key) # Inventory: Group by Route53 domain names if enabled if self.route53_enabled and self.group_by_route53_names: route53_names = self.get_instance_route53_names(instance) for name in route53_names: - self.push(self.inventory, name, dest) + self.push(self.inventory, name, hostname) if self.nested_groups: self.push_group(self.inventory, 'route53', name) # Global Tag: instances without tags if self.group_by_tag_none and len(instance.tags) == 0: - self.push(self.inventory, 'tag_none', dest) + self.push(self.inventory, 'tag_none', hostname) if self.nested_groups: self.push_group(self.inventory, 'tags', 'tag_none') # Global Tag: tag all EC2 instances - self.push(self.inventory, 'ec2', dest) + self.push(self.inventory, 'ec2', hostname) - self.inventory["_meta"]["hostvars"][dest] = self.get_host_info_dict_from_instance(instance) + self.inventory["_meta"]["hostvars"][hostname] = self.get_host_info_dict_from_instance(instance) + self.inventory["_meta"]["hostvars"][hostname]['ansible_ssh_host'] = dest def add_rds_instance(self, instance, region): @@ -752,24 +919,38 @@ class Ec2Inventory(object): # Skip instances we cannot address (e.g. private VPC subnet) return + # Set the inventory name + hostname = None + if self.hostname_variable: + if self.hostname_variable.startswith('tag_'): + hostname = instance.tags.get(self.hostname_variable[4:], None) + else: + hostname = getattr(instance, self.hostname_variable) + + # If we can't get a nice hostname, use the destination address + if not hostname: + hostname = dest + + hostname = self.to_safe(hostname).lower() + # Add to index - self.index[dest] = [region, instance.id] + self.index[hostname] = [region, instance.id] # Inventory: Group by instance ID (always a group of 1) if self.group_by_instance_id: - self.inventory[instance.id] = [dest] + self.inventory[instance.id] = [hostname] if self.nested_groups: self.push_group(self.inventory, 'instances', instance.id) # Inventory: Group by region if self.group_by_region: - self.push(self.inventory, region, dest) + self.push(self.inventory, region, hostname) if self.nested_groups: self.push_group(self.inventory, 'regions', region) # Inventory: Group by availability zone if self.group_by_availability_zone: - self.push(self.inventory, instance.availability_zone, dest) + self.push(self.inventory, instance.availability_zone, hostname) if self.nested_groups: if self.group_by_region: self.push_group(self.inventory, region, instance.availability_zone) @@ -778,14 +959,14 @@ class Ec2Inventory(object): # Inventory: Group by instance type if self.group_by_instance_type: type_name = self.to_safe('type_' + instance.instance_class) - self.push(self.inventory, type_name, dest) + self.push(self.inventory, type_name, hostname) if self.nested_groups: self.push_group(self.inventory, 'types', type_name) # Inventory: Group by VPC if self.group_by_vpc_id and instance.subnet_group and instance.subnet_group.vpc_id: vpc_id_name = self.to_safe('vpc_id_' + instance.subnet_group.vpc_id) - self.push(self.inventory, vpc_id_name, dest) + self.push(self.inventory, vpc_id_name, hostname) if self.nested_groups: self.push_group(self.inventory, 'vpcs', vpc_id_name) @@ -794,7 +975,7 @@ class Ec2Inventory(object): try: if instance.security_group: key = self.to_safe("security_group_" + instance.security_group.name) - self.push(self.inventory, key, dest) + self.push(self.inventory, key, hostname) if self.nested_groups: self.push_group(self.inventory, 'security_groups', key) @@ -805,20 +986,21 @@ class Ec2Inventory(object): # Inventory: Group by engine if self.group_by_rds_engine: - self.push(self.inventory, self.to_safe("rds_" + instance.engine), dest) + self.push(self.inventory, self.to_safe("rds_" + instance.engine), hostname) if self.nested_groups: self.push_group(self.inventory, 'rds_engines', self.to_safe("rds_" + instance.engine)) # Inventory: Group by parameter group if self.group_by_rds_parameter_group: - self.push(self.inventory, self.to_safe("rds_parameter_group_" + instance.parameter_group.name), dest) + self.push(self.inventory, self.to_safe("rds_parameter_group_" + instance.parameter_group.name), hostname) if self.nested_groups: self.push_group(self.inventory, 'rds_parameter_groups', self.to_safe("rds_parameter_group_" + instance.parameter_group.name)) # Global Tag: all RDS instances - self.push(self.inventory, 'rds', dest) + self.push(self.inventory, 'rds', hostname) - self.inventory["_meta"]["hostvars"][dest] = self.get_host_info_dict_from_instance(instance) + self.inventory["_meta"]["hostvars"][hostname] = self.get_host_info_dict_from_instance(instance) + self.inventory["_meta"]["hostvars"][hostname]['ansible_ssh_host'] = dest def add_elasticache_cluster(self, cluster, region): ''' Adds an ElastiCache cluster to the inventory and index, as long as @@ -1131,6 +1313,8 @@ class Ec2Inventory(object): instance_vars['ec2_placement'] = value.zone elif key == 'ec2_tags': for k, v in value.items(): + if self.expand_csv_tags and ',' in v: + v = list(map(lambda x: x.strip(), v.split(','))) key = self.to_safe('ec2_tag_' + k) instance_vars[key] = v elif key == 'ec2_groups': @@ -1141,6 +1325,10 @@ class Ec2Inventory(object): group_names.append(group.name) instance_vars["ec2_security_group_ids"] = ','.join([str(i) for i in group_ids]) instance_vars["ec2_security_group_names"] = ','.join([str(i) for i in group_names]) + elif key == 'ec2_block_device_mapping': + instance_vars["ec2_block_devices"] = {} + for k, v in value.items(): + instance_vars["ec2_block_devices"][ os.path.basename(k) ] = v.volume_id else: pass # TODO Product codes if someone finds them useful @@ -1321,4 +1509,3 @@ class Ec2Inventory(object): # Run the script Ec2Inventory() - diff --git a/inventory/gce/hosts/gce.py b/inventory/gce/hosts/gce.py index cce3c5f35..2be46a58c 100755 --- a/inventory/gce/hosts/gce.py +++ b/inventory/gce/hosts/gce.py @@ -70,7 +70,8 @@ Examples: $ contrib/inventory/gce.py --host my_instance Author: Eric Johnson <erjohnso@google.com> -Version: 0.0.1 +Contributors: Matt Hite <mhite@hotmail.com>, Tom Melendez <supertom@google.com> +Version: 0.0.3 ''' __requires__ = ['pycrypto>=2.6'] @@ -84,13 +85,19 @@ except ImportError: pass USER_AGENT_PRODUCT="Ansible-gce_inventory_plugin" -USER_AGENT_VERSION="v1" +USER_AGENT_VERSION="v2" import sys import os import argparse + +from time import time + import ConfigParser +import logging +logging.getLogger('libcloud.common.google').addHandler(logging.NullHandler()) + try: import json except ImportError: @@ -101,31 +108,103 @@ try: from libcloud.compute.providers import get_driver _ = Provider.GCE except: - print("GCE inventory script requires libcloud >= 0.13") - sys.exit(1) + sys.exit("GCE inventory script requires libcloud >= 0.13") + + +class CloudInventoryCache(object): + def __init__(self, cache_name='ansible-cloud-cache', cache_path='/tmp', + cache_max_age=300): + cache_dir = os.path.expanduser(cache_path) + if not os.path.exists(cache_dir): + os.makedirs(cache_dir) + self.cache_path_cache = os.path.join(cache_dir, cache_name) + + self.cache_max_age = cache_max_age + + def is_valid(self, max_age=None): + ''' Determines if the cache files have expired, or if it is still valid ''' + + if max_age is None: + max_age = self.cache_max_age + + if os.path.isfile(self.cache_path_cache): + mod_time = os.path.getmtime(self.cache_path_cache) + current_time = time() + if (mod_time + max_age) > current_time: + return True + + return False + + def get_all_data_from_cache(self, filename=''): + ''' Reads the JSON inventory from the cache file. Returns Python dictionary. ''' + + data = '' + if not filename: + filename = self.cache_path_cache + with open(filename, 'r') as cache: + data = cache.read() + return json.loads(data) + + def write_to_cache(self, data, filename=''): + ''' Writes data to file as JSON. Returns True. ''' + if not filename: + filename = self.cache_path_cache + json_data = json.dumps(data) + with open(filename, 'w') as cache: + cache.write(json_data) + return True class GceInventory(object): def __init__(self): + # Cache object + self.cache = None + # dictionary containing inventory read from disk + self.inventory = {} + # Read settings and parse CLI arguments self.parse_cli_args() + self.config = self.get_config() self.driver = self.get_gce_driver() + self.ip_type = self.get_inventory_options() + if self.ip_type: + self.ip_type = self.ip_type.lower() + + # Cache management + start_inventory_time = time() + cache_used = False + if self.args.refresh_cache or not self.cache.is_valid(): + self.do_api_calls_update_cache() + else: + self.load_inventory_from_cache() + cache_used = True + self.inventory['_meta']['stats'] = {'use_cache': True} + self.inventory['_meta']['stats'] = { + 'inventory_load_time': time() - start_inventory_time, + 'cache_used': cache_used + } # Just display data for specific host if self.args.host: - print(self.json_format_dict(self.node_to_dict( - self.get_instance(self.args.host)), - pretty=self.args.pretty)) - sys.exit(0) - - # Otherwise, assume user wants all instances grouped - print(self.json_format_dict(self.group_instances(), - pretty=self.args.pretty)) + print(self.json_format_dict( + self.inventory['_meta']['hostvars'][self.args.host], + pretty=self.args.pretty)) + else: + # Otherwise, assume user wants all instances grouped + zones = self.parse_env_zones() + print(self.json_format_dict(self.inventory, + pretty=self.args.pretty)) sys.exit(0) - def get_gce_driver(self): - """Determine the GCE authorization settings and return a - libcloud driver. + def get_config(self): + """ + Reads the settings from the gce.ini file. + + Populates a SafeConfigParser object with defaults and + attempts to read an .ini-style configuration from the filename + specified in GCE_INI_PATH. If the environment variable is + not present, the filename defaults to gce.ini in the current + working directory. """ gce_ini_default_path = os.path.join( os.path.dirname(os.path.realpath(__file__)), "gce.ini") @@ -140,14 +219,57 @@ class GceInventory(object): 'gce_service_account_pem_file_path': '', 'gce_project_id': '', 'libcloud_secrets': '', + 'inventory_ip_type': '', + 'cache_path': '~/.ansible/tmp', + 'cache_max_age': '300' }) if 'gce' not in config.sections(): config.add_section('gce') + if 'inventory' not in config.sections(): + config.add_section('inventory') + if 'cache' not in config.sections(): + config.add_section('cache') + config.read(gce_ini_path) + ######### + # Section added for processing ini settings + ######### + + # Set the instance_states filter based on config file options + self.instance_states = [] + if config.has_option('gce', 'instance_states'): + states = config.get('gce', 'instance_states') + # Ignore if instance_states is an empty string. + if states: + self.instance_states = states.split(',') + + # Caching + cache_path = config.get('cache', 'cache_path') + cache_max_age = config.getint('cache', 'cache_max_age') + # TOOD(supertom): support project-specific caches + cache_name = 'ansible-gce.cache' + self.cache = CloudInventoryCache(cache_path=cache_path, + cache_max_age=cache_max_age, + cache_name=cache_name) + return config + + def get_inventory_options(self): + """Determine inventory options. Environment variables always + take precedence over configuration files.""" + ip_type = self.config.get('inventory', 'inventory_ip_type') + # If the appropriate environment variables are set, they override + # other configuration + ip_type = os.environ.get('INVENTORY_IP_TYPE', ip_type) + return ip_type + + def get_gce_driver(self): + """Determine the GCE authorization settings and return a + libcloud driver. + """ # Attempt to get GCE params from a configuration file, if one # exists. - secrets_path = config.get('gce', 'libcloud_secrets') + secrets_path = self.config.get('gce', 'libcloud_secrets') secrets_found = False try: import secrets @@ -161,8 +283,7 @@ class GceInventory(object): if not secrets_path.endswith('secrets.py'): err = "Must specify libcloud secrets file as " err += "/absolute/path/to/secrets.py" - print(err) - sys.exit(1) + sys.exit(err) sys.path.append(os.path.dirname(secrets_path)) try: import secrets @@ -173,10 +294,10 @@ class GceInventory(object): pass if not secrets_found: args = [ - config.get('gce','gce_service_account_email_address'), - config.get('gce','gce_service_account_pem_file_path') + self.config.get('gce','gce_service_account_email_address'), + self.config.get('gce','gce_service_account_pem_file_path') ] - kwargs = {'project': config.get('gce', 'gce_project_id')} + kwargs = {'project': self.config.get('gce', 'gce_project_id')} # If the appropriate environment variables are set, they override # other configuration; process those into our args and kwargs. @@ -191,6 +312,14 @@ class GceInventory(object): ) return gce + def parse_env_zones(self): + '''returns a list of comma separated zones parsed from the GCE_ZONE environment variable. + If provided, this will be used to filter the results of the grouped_instances call''' + import csv + reader = csv.reader([os.environ.get('GCE_ZONE',"")], skipinitialspace=True) + zones = [r for r in reader] + return [z for z in zones[0]] + def parse_cli_args(self): ''' Command line argument processing ''' @@ -202,6 +331,9 @@ class GceInventory(object): help='Get all information about an instance') parser.add_argument('--pretty', action='store_true', default=False, help='Pretty format (default: False)') + parser.add_argument( + '--refresh-cache', action='store_true', default=False, + help='Force refresh of cache by making API requests (default: False - use cache files)') self.args = parser.parse_args() @@ -211,11 +343,17 @@ class GceInventory(object): if inst is None: return {} - if inst.extra['metadata'].has_key('items'): + if 'items' in inst.extra['metadata']: for entry in inst.extra['metadata']['items']: md[entry['key']] = entry['value'] net = inst.extra['networkInterfaces'][0]['network'].split('/')[-1] + # default to exernal IP unless user has specified they prefer internal + if self.ip_type == 'internal': + ssh_host = inst.private_ips[0] + else: + ssh_host = inst.public_ips[0] if len(inst.public_ips) >= 1 else inst.private_ips[0] + return { 'gce_uuid': inst.uuid, 'gce_id': inst.id, @@ -231,29 +369,67 @@ class GceInventory(object): 'gce_metadata': md, 'gce_network': net, # Hosts don't have a public name, so we add an IP - 'ansible_ssh_host': inst.public_ips[0] if len(inst.public_ips) >= 1 else inst.private_ips[0] + 'ansible_ssh_host': ssh_host } - def get_instance(self, instance_name): - '''Gets details about a specific instance ''' + def load_inventory_from_cache(self): + ''' Loads inventory from JSON on disk. ''' + try: - return self.driver.ex_get_node(instance_name) + self.inventory = self.cache.get_all_data_from_cache() + hosts = self.inventory['_meta']['hostvars'] except Exception as e: - return None - - def group_instances(self): + print( + "Invalid inventory file %s. Please rebuild with -refresh-cache option." + % (self.cache.cache_path_cache)) + raise + + def do_api_calls_update_cache(self): + ''' Do API calls and save data in cache. ''' + zones = self.parse_env_zones() + data = self.group_instances(zones) + self.cache.write_to_cache(data) + self.inventory = data + + def list_nodes(self): + all_nodes = [] + params, more_results = {'maxResults': 500}, True + while more_results: + self.driver.connection.gce_params=params + all_nodes.extend(self.driver.list_nodes()) + more_results = 'pageToken' in params + return all_nodes + + def group_instances(self, zones=None): '''Group all instances''' groups = {} meta = {} meta["hostvars"] = {} - for node in self.driver.list_nodes(): + for node in self.list_nodes(): + + # This check filters on the desired instance states defined in the + # config file with the instance_states config option. + # + # If the instance_states list is _empty_ then _ALL_ states are returned. + # + # If the instance_states list is _populated_ then check the current + # state against the instance_states list + if self.instance_states and not node.extra['status'] in self.instance_states: + continue + name = node.name meta["hostvars"][name] = self.node_to_dict(node) zone = node.extra['zone'].name - if groups.has_key(zone): groups[zone].append(name) + + # To avoid making multiple requests per zone + # we list all nodes and then filter the results + if zones and zone not in zones: + continue + + if zone in groups: groups[zone].append(name) else: groups[zone] = [name] tags = node.extra['tags'] @@ -262,25 +438,25 @@ class GceInventory(object): tag = t[6:] else: tag = 'tag_%s' % t - if groups.has_key(tag): groups[tag].append(name) + if tag in groups: groups[tag].append(name) else: groups[tag] = [name] net = node.extra['networkInterfaces'][0]['network'].split('/')[-1] net = 'network_%s' % net - if groups.has_key(net): groups[net].append(name) + if net in groups: groups[net].append(name) else: groups[net] = [name] machine_type = node.size - if groups.has_key(machine_type): groups[machine_type].append(name) + if machine_type in groups: groups[machine_type].append(name) else: groups[machine_type] = [name] image = node.image and node.image or 'persistent_disk' - if groups.has_key(image): groups[image].append(name) + if image in groups: groups[image].append(name) else: groups[image] = [name] status = node.extra['status'] stat = 'status_%s' % status.lower() - if groups.has_key(stat): groups[stat].append(name) + if stat in groups: groups[stat].append(name) else: groups[stat] = [name] groups["_meta"] = meta @@ -296,6 +472,6 @@ class GceInventory(object): else: return json.dumps(data) - # Run the script -GceInventory() +if __name__ == '__main__': + GceInventory() |