def main(self):
options = ['--stdout']
parser = Transport(self.argv, options=options, check_version=False)
parser.catch_help = self._help()
parser.parse_args()
if len(self.argv) == 1:
raise SystemExit(
"A monitor hostname or a ceph.conf file is required as an argument"
)
node = self.argv[-1]
inventory = {}
with get_connection(node) as conn:
report = get_mon_report(conn)
try:
mons = report['monmap']['mons']
except KeyError:
raise SystemExit(report)
inventory['mons'] = [i['name'] for i in mons]
osds = report['osd_metadata']
inventory['osds'] = [i['hostname'] for i in osds]
if not inventory:
raise SystemExit(
'no hosts where found from remote monitor node: %s' % node)
generate_inventory(inventory, to_stdout=parser.get('--stdout'))
conn.exit()
return
def collect():
"""
The main collecting entrypoint. This function will call all the pieces
needed to build the complete metadata set of a remote system so that checks
can consume and verify that data.
After collection is done, the full contents of the metadata are available
at ``ceph_medic.metadata``
"""
cluster_nodes = metadata['nodes']
loader.write('collecting remote node information')
total_nodes = 0
failed_nodes = 0
for node_type, nodes in cluster_nodes.items():
for node in nodes:
# check if a node type exists for this node before doing any work:
try:
metadata[node_type]
except KeyError:
msg = "Skipping node {} from unknown host group: {}".format(
node, node_type)
logger.warning(msg)
continue
total_nodes += 1
hostname = node['host']
loader.write('Host: %-20s connection: [%-20s]' %
(hostname, terminal.yellow('connecting')))
# TODO: make sure that the hostname is resolvable, trying to
# debug SSH issues with execnet is pretty hard/impossible, use
# util.net.host_is_resolvable
try:
logger.debug('attempting connection to host: %s', node['host'])
conn = get_connection(node['host'])
loader.write('Host: %-20s connection: [%-20s]' %
(hostname, terminal.green('connected')))
loader.write('\n')
except HostNotFound:
logger.exception('connection failed')
loader.write('Host: %-20s connection: [%-20s]' %
(hostname, terminal.red('failed')))
loader.write('\n')
failed_nodes += 1
continue
# send the full node metadata for global scope so that the checks
# can consume this
metadata[node_type][node['host']] = get_node_metadata(
conn, hostname, cluster_nodes)
conn.exit()
if failed_nodes == total_nodes:
loader.write(terminal.red('Collection failed!') + ' ' * 70 + '\n')
raise RuntimeError(
'All nodes failed to connect. Cannot run any checks')
else:
loader.write('Collection completed!' + ' ' * 70 + '\n')
def collect():
"""
The main collecting entrypoint. This function will call all the pieces
needed to build the complete metadata set of a remote system so that checks
can consume and verify that data.
After collection is done, the full contents of the metadata are available
at ``ceph_medic.metadata``
"""
cluster_nodes = metadata['nodes']
loader.write('collecting remote node information')
total_nodes = 0
failed_nodes = 0
has_cluster_data = False
for node_type, nodes in cluster_nodes.items():
for node in nodes:
# check if a node type exists for this node before doing any work:
try:
metadata[node_type]
except KeyError:
msg = "Skipping node {} from unknown host group: {}".format(
node, node_type)
logger.warning(msg)
continue
total_nodes += 1
hostname = node['host']
loader.write('Host: %-40s connection: [%-20s]' %
(hostname, terminal.yellow('connecting')))
# TODO: make sure that the hostname is resolvable, trying to
# debug SSH issues with execnet is pretty hard/impossible, use
# util.net.host_is_resolvable
try:
logger.debug('attempting connection to host: %s', node['host'])
conn = get_connection(node['host'],
container=node.get('container'))
loader.write('Host: %-40s connection: [%-20s]' %
(hostname, terminal.green('connected')))
loader.write('\n')
except HostNotFound as err:
logger.exception('connection failed')
loader.write('Host: %-40s connection: [%-20s]' %
(hostname, terminal.red('failed')))
loader.write('\n')
failed_nodes += 1
if metadata[node_type].get(hostname):
metadata[node_type].pop(hostname)
metadata['nodes'][node_type] = [
i for i in metadata['nodes'][node_type]
if i['host'] != hostname
]
metadata['failed_nodes'].update({hostname: str(err)})
continue
# send the full node metadata for global scope so that the checks
# can consume this
metadata[node_type][hostname] = get_node_metadata(
conn, hostname, cluster_nodes)
if node_type == 'mons': # if node type is monitor, admin privileges are most likely authorized
if not has_cluster_data:
cluster_data = collect_cluster(conn)
if cluster_data:
metadata['cluster'] = cluster_data
has_cluster_data = True
conn.exit()
if failed_nodes == total_nodes:
loader.write(terminal.red('Collection failed!') + ' ' * 70)
# TODO: this helps clear out the 'loader' line so that the error looks
# clean, but this manual clearing should be done automatically
terminal.write.raw('')
raise RuntimeError(
'All nodes failed to connect. Cannot run any checks')
if failed_nodes:
loader.write(
terminal.yellow(
'Collection completed with some failed connections' +
' ' * 70 + '\n'))
else:
loader.write('Collection completed!' + ' ' * 70 + '\n')