def get_cluster_topology(self, seed_nodes):
cluster_names = list()
self.schema_versions = list()
self.status_topology_hash = list()
topologies = list()
count = 0
tried_hosts = []
for host in seed_nodes:
tried_hosts.append(host)
conn = self._connection(host)
describe_res = self.run_nodetool(conn, "describecluster")
status_res = self.run_nodetool(conn, "status")
if (describe_res.status == 0) and (status_res.status == 0):
(cluster_name, schema_version) = cstar.nodetoolparser.parse_describe_cluster(describe_res.out)
if cluster_name not in cluster_names:
cluster_names.append(cluster_name)
topologies.append(cstar.nodetoolparser.parse_nodetool_status(status_res.out, cluster_name, self.reverse_dns_preheat, self.resolve_hostnames))
self.schema_versions.append(schema_version)
self.status_topology_hash.append(topologies[len(topologies) - 1].get_hash())
count += 1
if count >= MAX_ATTEMPTS:
break
if len(topologies) > 0:
final_topology = set()
for i in range(len(topologies)):
final_topology.update(topologies[i].hosts)
return Topology(final_topology)
raise HostIsDown("Could not find any working host while fetching topology. Is Cassandra actually running? Tried the following hosts:",
", ".join([x.ip for x in tried_hosts]))
def get_endpoint_mapping(self, topology):
count = 0
tried_hosts = []
for host in topology.get_up():
tried_hosts.append(host)
conn = self._connection(host)
mappings = []
if self.key_space:
keyspaces = [self.key_space]
else:
keyspaces = self.get_keyspaces(conn)
for keyspace in keyspaces:
if keyspace != "system":
debug("Fetching endpoint mapping for keyspace", keyspace)
res = conn.run(("nodetool", "describering", keyspace))
has_error = False
if res.status != 0:
has_error = True
break
describering = cstar.nodetoolparser.parse_nodetool_describering(res.out)
range_mapping = cstar.nodetoolparser.convert_describering_to_range_mapping(describering)
mappings.append(cstar.endpoint_mapping.parse(range_mapping, topology, lookup=ip_lookup))
if not has_error:
return cstar.endpoint_mapping.merge(mappings)
count += 1
if count >= MAX_ATTEMPTS:
break
raise HostIsDown("Could not find any working host while fetching endpoint mapping. Tried the following hosts:",
", ".join(host.fqdn for host in tried_hosts))
def find_next_host(strategy, topology, endpoint_mapping, progress, cluster_parallel, dc_parallel, max_concurrency,
stop_after, ignore_down_nodes):
"""Entry point for figuring out which host to do things on next."""
if stop_after and ((len(progress.running) + len(progress.done) + len(progress.failed)) >= stop_after):
return None
remaining = topology.without_hosts(progress.done).without_hosts(progress.running).without_hosts(progress.failed)
if progress.running and not cluster_parallel:
remaining = remaining.with_cluster(next(iter(progress.running)).cluster)
if progress.running and not dc_parallel:
remaining = remaining.with_dc(next(iter(progress.running)).dc)
if not remaining:
return None
if max_concurrency and (len(progress.running) >= max_concurrency):
return None
if not ignore_down_nodes:
for host in remaining:
if not host.is_up:
raise HostIsDown(host)
return _strategy_mapping[strategy](remaining, endpoint_mapping, progress.running)
def get_endpoint_mapping(self, topology):
clusters = []
failed_hosts = []
mappings = []
count = 0
endpoint_mappings = self.maybe_get_data_from_cache("endpoint_mapping")
if endpoint_mappings is not None:
return endpoint_mappings
for host in topology.get_up():
if host.cluster in clusters:
# We need to fetch keyspaces on one node per cluster, no more.
continue
count = 0
conn = self._connection(host)
if self.key_space:
keyspaces = [self.key_space]
else:
keyspaces = self.get_keyspaces(conn)
has_error = True
for keyspace in keyspaces:
if not keyspace.startswith("system"):
debug("Fetching endpoint mapping for keyspace", keyspace)
res = self.run_nodetool(conn, *("describering", keyspace))
has_error = False
if res.status != 0 and not keyspace.startswith("system"):
has_error = True
break
describering = cstar.nodetoolparser.parse_nodetool_describering(
res.out)
range_mapping = cstar.nodetoolparser.convert_describering_to_range_mapping(
describering)
mappings.append(
cstar.endpoint_mapping.parse(range_mapping,
topology,
lookup=ip_lookup))
if has_error:
if count >= MAX_ATTEMPTS:
failed_hosts += host
break
else:
clusters.append(host.cluster)
count += 1
if failed_hosts:
raise HostIsDown("Following hosts couldn't be reached: {}".format(
', '.join(host.fqdn for host in failed_hosts)))
endpoint_mappings = cstar.endpoint_mapping.merge(mappings)
pickle.dump(dict(endpoint_mappings),
open(self.get_cache_file_path("endpoint_mapping"), 'wb'))
return endpoint_mappings
def get_cluster_topology(self, seed_nodes):
count = 0
tried_hosts = []
for host in seed_nodes:
tried_hosts.append(host)
conn = self._connection(host)
describe_res = conn.run(("nodetool", "describecluster"))
topology_res = conn.run(("nodetool", "ring"))
if (describe_res.status == 0) and (topology_res.status == 0):
cluster_name = cstar.nodetoolparser.parse_describe_cluster(describe_res.out)
topology = cstar.nodetoolparser.parse_nodetool_ring(topology_res.out, cluster_name, self.reverse_dns_preheat)
return topology
count += 1
if count >= MAX_ATTEMPTS:
break
raise HostIsDown("Could not find any working host while fetching topology. Is Cassandra actually running? Tried the following hosts:",
", ".join(tried_hosts))
def run(self):
self.do_loop = True
cstar.jobwriter.write(self)
if not self.state.is_healthy():
raise HostIsDown(
"Can't run job because hosts are down: " + ", ".join(
host.fqdn for host in self.state.current_topology.get_down()))
while self.do_loop:
self.schedule_all_runnable_jobs()
if self.state.is_done():
self.do_loop = False
self.wait_for_any_job()
self.wait_for_all_jobs()
cstar.jobprinter.print_progress(self.state.original_topology,
self.state.progress,
self.state.current_topology.get_down())
self.print_outcome()