def monitor_states(self, connection):
logger.debug("enter monitor_states_and_get_nodes")
object_name = "org.apache.cassandra.net:type=FailureDetector"
attr = "SimpleStates"
path = self.build_path(object_name)
metrics.define_store(path + "." + attr, self.metrics_namespace_prefix)
simple_states = connection.getAttribute(
javax.management.ObjectName(object_name), attr).toString()
simple_states = simple_states.replace("{",
"").replace("}", "").replace(
"/", "").split(", ")
attr_vlaue = 0
logger.info("simple_states={}".format(simple_states))
for item in simple_states:
ip, status = item.split("=")
self.all_nodes.add(ip)
tagv = ip.replace(".", "_").replace(":", "_")
cur_tagk = {"instance": tagv}
if status != "UP":
attr_vlaue = 1
logger.info("ip={},SimpleStates={}".format(ip, status))
metrics.emit_store(path + "." + attr,
attr_vlaue,
prefix=self.metrics_namespace_prefix,
tagkv=cur_tagk)
def define_tagkv(self):
try:
publish_address_path = "http.publish_address"
nodes_info = self.client.nodes.info()
if nodes_info is None:
return []
cluster_name = self.find(nodes_info, "cluster_name")
nodes_info = self.find(nodes_info, "nodes")
tagv_arr = [self.filename]
for node_name in nodes_info:
node_value = nodes_info.get(node_name)
publish_address = self.find(node_value, publish_address_path)
tagv = publish_address.replace(".", "_").replace(":", "_")
self.addrdict[node_name] = tagv
tagv_arr.append(tagv)
metrics.define_tagkv('instance', tagv_arr)
self.metrics_namespace_prefix = '' + cluster_name
metrics.define_store("cluster.health_status",
self.metrics_namespace_prefix)
for index in self.indices:
metrics.define_store(index, self.metrics_namespace_prefix)
except Exception as e:
logger.warning(e)
def define_tagkv(self):
tagv_arr = [self.filename]
for host in self.hosts:
temp = {}
ip = host.split(":")[0]
port = host.split(":")[1]
temp["host"] = ip
temp["port"] = port
tagv_arr.append(host.replace(".", "_").replace(":", "_"))
metrics.define_tagkv('instance', tagv_arr)
for index in self.indices:
metrics.define_store(index, self.metrics_namespace_prefix)
def monitor(self):
"""
question:https://stackoverflow.com/questions/7861299/jpype-doesnt-work-inside-thread
:return:
"""
jpype.attachThreadToJVM()
metrics.define_counter("indices.monitor.error",
self.metrics_namespace_prefix)
for node in self.all_nodes:
try:
cur_tagk = {
"instance": node.replace(".", "_").replace(":", "_")
}
connection_factory = self.get_connection_factory(node, "7199")
connection = connection_factory.getMBeanServerConnection()
logger.info("monitor,host={}".format(node))
self.monitor_heap_usesage(node, connection)
self.monitor_states(connection)
logger.info(
"objectname_attr_dict={}".format(objectname_attr_dict))
for object_name, attrs in objectname_attr_dict.items():
for attribute in attrs:
try:
attr_vlaue = connection.getAttribute(
javax.management.ObjectName(object_name),
attribute)
except Exception as e:
logger.info(
"monitor getAttribute ex,attribute={},exception={}"
.format(attribute, e))
continue
if isNumber(attr_vlaue) == False:
continue
object_name_path = self.build_path(object_name)
metrics.define_store(
object_name_path + "." + attribute,
self.metrics_namespace_prefix)
logger.info(
"monitor,object_name_path={},object_name={},attr_vlaue={}"
.format(object_name_path, object_name, attr_vlaue))
metrics.emit_store(
object_name_path + "." + attribute,
attr_vlaue,
prefix=self.metrics_namespace_prefix,
tagkv=cur_tagk)
except Exception as e:
logger.exception(e)
finally:
connection_factory.close()
def monitor_heap_usesage(self, ip, connection):
object_name = "java.lang:type=Memory" # 查询内存
path = self.build_path(object_name)
attribute = "HeapMemoryUsage"
metrics.define_store(path + "." + attribute,
self.metrics_namespace_prefix)
tagv = ip.replace(".", "_").replace(":", "_")
cur_tagk = {"instance": tagv}
attr = connection.getAttribute(
javax.management.ObjectName(object_name), attribute)
used = attr.contents.get("used")
logger.debug("used={}".format(used))
metrics.emit_store(path + "." + attribute,
used,
prefix=self.metrics_namespace_prefix,
tagkv=cur_tagk)