zabbix使用zabbix_java_gateway 监控java应用进程

Posted 2020-08-21

JAVA-GATEWAY Zabbix本身不支持直接监控Java，在zabbix 1.8以前，只能使用Zapcat来做代理监控，而且要修改源代码，非常麻烦。所有后来为了解决这个监控问题，Zabbix和Java双方应运生成了各自的代理监控程序：zabbix 2.0以后添加了服务进程zabbix-java-gateway；Java有了JMX,全称是Java Management Extensions,即Java管理扩展 比如：当Zabbix-Server需要知道java应用程序的某项性能的时候，会启动自身的一个Zabbix-JavaPollers进程去连接Zabbix-JavaGateway请求数据，而ZabbixJavagateway收到请求后使用“JMXmanagementAPI”去查询特定的应用程序，而前提是应用程序这端在开启时需要“-Dcom.sun.management.jmxremote”参数来开启JMX远程查询就行。Java程序会启动自身的一个简单的小程序端口12345向Zabbix-JavaGateway提供请求数据。 开始监控部署   从上面的原理图中我们可以看出，配置Zabbix监控Java应用程序的关键点在于：配置Zabbix-JavaGateway、让Zabbix-Server能够连接Zabbix-JavaGateway、Tomcat开启JVM远程监控功能等 网上的大多数文章都配置的是被动模式（即zabbix server 去想java进程去取数据），这样会造成zabbix server的压力过大，所以应该采用主动模式（写一个zabbix类，然后主动去上报数据） 代码如下： zabbix.py cat zabbix.py #!/usr/bin/python import os import time import socket import struct import cPickle import logging try:     from hashlib import sha1 except ImportError:     from sha import sha as sha1 try:     import json except ImportError:     import simplejson as json DUMP = ‘dump‘ if not os.path.isdir(DUMP):     os.makedirs(DUMP, mode=0755) class Zabbix(object):     logger = logging.getLogger(‘zabbix‘)     def __init__(self, values=None):         if values is not None:             self.__dict__[‘values‘] = values     def __getattr__(self, name):         if name in (‘values‘):             return self.__dict__[name]         return None     def __setattr__(self, name, value):         if name == ‘values‘:             self.__dict__[name] = value     def gen_request(self, jsons):         if isinstance(jsons, basestring):             data = ‘%s\n‘ % jsons         else:             data = json.dumps(jsons)         header = ‘ZBXD\x01‘         datalen = struct.pack(‘Q‘, len(data))         return header + datalen + data     def dump(self, host, port, jsons):         data = {‘host‘: host, ‘port‘: port, ‘jsons‘: jsons}         hash = sha1(json.dumps(data)).hexdigest()         path = ‘%s.%s.%d.%s.error‘ % (host, port, int(time.time()), hash)         try:             write = open(os.path.join(DUMP, path), ‘wb‘)             cPickle.dump(data, write, -1)             write.close()         except:             self.logger.exception(‘cannot dump to file %s‘, path)     def get_zbx_result(self, host, port, jsons):         retry = 3         while retry:             try:                 data = self._get_zbx_result(host, port, jsons)                 if data is None:                     break                 return data             except socket.error:                 self.logger.exception(‘cannot communit with zabbix‘)             time.sleep(5)             retry -= 1         self.logger.error(‘cannot send data to zabbix server‘)         self.dump(host, port, jsons)     def _get_zbx_result(self, host, port, jsons):         sock = socket.socket(socket.AF_INET, socket.SOCK_STREAM)         print host,port,jsons         sock.connect((host, port))         sock.send(self.gen_request(jsons))         print "send %s" % self.gen_request(jsons)         self.logger.debug(‘sent %s‘, jsons)         recv = sock.recv(5)         if recv != ‘ZBXD\x01‘:             self.logger.error(‘Invalid Response‘)             self.dump(host, port, jsons)             return None         recv = sock.recv(8)         (datalen,) = struct.unpack(‘Q‘, recv)         data = sock.recv(datalen)         sock.close()         self.logger.debug(‘received %s‘, data)         return data     def getvalue(self):         # shoulde be {host: {key: value}}         return self.values     def run(self):         hostvalues = self.getvalue()         if not isinstance(hostvalues, dict):             self.logger.error(‘invalid hostvalues: %s‘, str(hostvalues))             return False         clock = int(time.time())         jsons = {             ‘request‘: ‘agent data‘,             ‘data‘: [],             ‘clock‘: clock,         }         data = jsons[‘data‘]         for host, values in hostvalues.iteritems():             for key, value in values.iteritems():                 data.append({                     ‘host‘: host,                     ‘key‘: key,                     ‘value‘: value,                     ‘clock‘: clock,                 })         return self.get_zbx_result(ZBX_HOST, ZBX_PORT, jsons)     def getjmx(self, host, port, keys):         jsons = {             ‘request‘: ‘java gateway jmx‘,             ‘conn‘: host,             ‘port‘: port,             ‘keys‘: keys,         }         return self.get_zbx_result(JMX_HOST, JMX_PORT, jsons) ZBX_HOST = ‘zabbix.server.com‘ ZBX_PORT = 10051 JMX_HOST = ‘localhost‘ JMX_PORT = 10052 handler = logging.FileHandler(filename=‘/tmp/zabbix.%s.log‘ % time.strftime(‘%Y%m%d‘), mode=‘a‘) formatter = logging.Formatter(‘%(asctime)s %(name)s %(lineno)s %(levelname)s %(message)s‘) handler.setFormatter(formatter) logger = logging.getLogger() logger.addHandler(handler) logger.setLevel(logging.DEBUG) if __name__ == ‘__main__‘:     zbx = Zabbix()     print zbx.getjmx(‘localhost‘, 8081, [‘jmx["Standalone:type=Manager,path=/,host=localhost",activeSessions]‘, ‘jmx["java.lang:type=Runtime",Uptime]‘, ‘jmx["Catalina:type=GlobalRequestProcessor,name=\\"http-bio-8080\\"",bytesSent]‘])     pass cat zabbix_wiki_node1_java.py    #!/usr/bin/env python #fileencoding: utf-8 ITEMS = [ ‘jmx["java.lang:type=ClassLoading",LoadedClassCount]‘, ‘jmx["java.lang:type=ClassLoading",TotalLoadedClassCount]‘, ‘jmx["java.lang:type=ClassLoading",UnloadedClassCount]‘, ‘jmx["java.lang:type=Memory",HeapMemoryUsage.committed]‘, ‘jmx["java.lang:type=Memory",HeapMemoryUsage.max]‘, ‘jmx["java.lang:type=Memory",HeapMemoryUsage.used]‘, ‘jmx["java.lang:type=Memory",NonHeapMemoryUsage.committed]‘, ‘jmx["java.lang:type=Memory",NonHeapMemoryUsage.max]‘, ‘jmx["java.lang:type=Memory",NonHeapMemoryUsage.used]‘, ‘jmx["java.lang:type=MemoryPool,name=Code Cache",Usage.committed]‘, ‘jmx["java.lang:type=MemoryPool,name=Code Cache",Usage.max]‘, ‘jmx["java.lang:type=MemoryPool,name=Code Cache",Usage.used]‘, ‘jmx["java.lang:type=MemoryPool,name=PS Eden Space",Usage.committed]‘, ‘jmx["java.lang:type=MemoryPool,name=PS Eden Space",Usage.max]‘, ‘jmx["java.lang:type=MemoryPool,name=PS Eden Space",Usage.used]‘, ‘jmx["java.lang:type=MemoryPool,name=PS Old Gen",Usage.committed]‘, ‘jmx["java.lang:type=MemoryPool,name=PS Old Gen",Usage.max]‘, ‘jmx["java.lang:type=MemoryPool,name=PS Old Gen",Usage.used]‘, ‘jmx["java.lang:type=MemoryPool,name=PS Perm Gen",Usage.committed]‘, ‘jmx["java.lang:type=MemoryPool,name=PS Perm Gen",Usage.max]‘, ‘jmx["java.lang:type=MemoryPool,name=PS Perm Gen",Usage.used]‘, ‘jmx["java.lang:type=MemoryPool,name=PS Survivor Space",Usage.committed]‘, ‘jmx["java.lang:type=MemoryPool,name=PS Survivor Space",Usage.max]‘, ‘jmx["java.lang:type=MemoryPool,name=PS Survivor Space",Usage.used]‘, ‘jmx["java.lang:type=OperatingSystem",MaxFileDescriptorCount]‘, ‘jmx["java.lang:type=OperatingSystem",OpenFileDescriptorCount]‘, ‘jmx["java.lang:type=Runtime",Uptime]‘, ‘jmx["java.lang:type=Threading",DaemonThreadCount]‘, ‘jmx["java.lang:type=Threading",PeakThreadCount]‘, ‘jmx["java.lang:type=Threading",ThreadCount]‘, ‘jmx["java.lang:type=Threading",TotalStartedThreadCount]‘, ‘jmx["Standalone:type=GlobalRequestProcessor,name=http-8090",bytesReceived]‘, ‘jmx["Standalone:type=GlobalRequestProcessor,name=http-8090",bytesSent]‘, ‘jmx["Standalone:type=GlobalRequestProcessor,name=http-8090",errorCount]‘, ‘jmx["Standalone:type=GlobalRequestProcessor,name=http-8090",processingTime]‘, ‘jmx["Standalone:type=GlobalRequestProcessor,name=http-8090",requestCount]‘, ‘jmx["Standalone:type=Manager,path=/,host=localhost",activeSessions]‘, ‘jmx["Standalone:type=Manager,path=/,host=localhost",maxActiveSessions]‘, ‘jmx["Standalone:type=Manager,path=/,host=localhost",maxActive]‘, ‘jmx["Standalone:type=Manager,path=/,host=localhost",rejectedSessions]‘, ‘jmx["Standalone:type=Manager,path=/,host=localhost",sessionCounter]‘, ‘jmx["Standalone:type=ProtocolHandler,port=8090",compression]‘, ‘jmx["Standalone:type=ThreadPool,name=http-8090",currentThreadCount]‘, ‘jmx["Standalone:type=ThreadPool,name=http-8090",currentThreadsBusy]‘, ‘jmx["Standalone:type=ThreadPool,name=http-8090",maxThreads]‘, ] from zabbix import Zabbix from zabbix import json def getjmxkey(key):     # key = key.replace(‘http-8080‘, ‘\\"http-bio-8080\\"‘)     # key = key.replace(‘path=/‘, ‘context=/‘)     return key def getjmxvalue(value):     if isinstance(value, dict):         return value.get(‘value‘, u‘‘).encode(‘utf8‘)     return ‘‘ if __name__ == ‘__main__‘:     host = ‘it-tw01‘     zbx = Zabbix()     jmxkeys = [getjmxkey(x) for x in ITEMS]     data = zbx.getjmx(‘it-tw01‘, ‘8410‘, jmxkeys)     try:         results = json.loads(data)     except:         results = {}     if isinstance(results, dict) and results[‘response‘] == ‘success‘:         jmxvalues = [getjmxvalue(x) for x in results.get(‘data‘, [])]         hostvalues = {host: dict(zip(ITEMS, jmxvalues))}         Zabbix(hostvalues).run() # vim: set sta sw=4 et: crontab * * * * * /home/sankuai/monitor/zabbix_wiki_node1_java.py zabbix /etc/zabbix# ls zabbix_agentd.conf  zabbix_agentd.confn-place  zabbix_agentd.d  zabbix_java_gateway.conf it-tw01需要关联的模板见附件 特别注意： 编译安装zabbix server需要加上--enable-java以支持jmx监控，如果之前的zabbix server没加，那么请重新编译安装，参考编译参数 安装软件 yum install -y java java-devel zabbix-java-gateway 更多的请参考这个文章：http://www.iyunv.com/thread-269939-1-1.html