配置文件

[server]
hostname=localhost
url_port=8440
secured_url_port=8441
connect_retry_delay=10
max_reconnect_retry_delay=30


[agent]
logdir=/var/log/ambari-agent
piddir=/var/run/ambari-agent
prefix=/var/lib/ambari-agent/data
;loglevel=(DEBUG/INFO)
loglevel=INFO
data_cleanup_interval=86400
data_cleanup_max_age=2592000
data_cleanup_max_size_MB = 100
ping_port=8670
cache_dir=/var/lib/ambari-agent/cache
tolerate_download_failures=true
run_as_user=root
parallel_execution=0
alert_grace_period=5
status_command_timeout=5
alert_kinit_timeout=14400000
system_resource_overrides=/etc/resource_overrides
; memory_threshold_soft_mb=400
; memory_threshold_hard_mb=1000
; ignore_mount_points=/mnt/custom1,/mnt/custom2

[security]
keysdir=/var/lib/ambari-agent/keys
server_crt=ca.crt
passphrase_env_var_name=AMBARI_PASSPHRASE
ssl_verify_cert=0
credential_lib_dir=/var/lib/ambari-agent/cred/lib
credential_conf_dir=/var/lib/ambari-agent/cred/conf
credential_shell_cmd=org.apache.hadoop.security.alias.CredentialShell
force_https_protocol=PROTOCOL_TLSv1_2

[network]
; this option apply only for Agent communication
use_system_proxy_settings=true

[services]
pidLookupPath=/var/run/

[heartbeat]
state_interval_seconds=60
dirs=/etc/hadoop,/etc/hadoop/conf,/etc/hbase,/etc/hcatalog,/etc/hive,/etc/oozie,
  /etc/sqoop,
  /var/run/hadoop,/var/run/zookeeper,/var/run/hbase,/var/run/templeton,/var/run/oozie,
  /var/log/hadoop,/var/log/zookeeper,/var/log/hbase,/var/run/templeton,/var/log/hive
; 0 - unlimited
log_lines_count=300
idle_interval_min=1
idle_interval_max=10


[logging]
syslog_enabled=0

心跳通信(Heartbeat)

发送给ambari-server的数据

{
    "alerts": [],
    "nodeStatus": {
        "status": "HEALTHY",
        "cause": "NONE"
    },
    "timestamp": 1536306543920,
    "hostname": "xxxxx.com",
    "responseId": 1139,
    "reports": [],
    "mounts": [{
        "available": "97151016",
        "used": "112461784",
        "percent": "54%",
        "device": "/dev/vdb",
        "mountpoint": "/data01",
        "type": "xfs",
        "size": "209612800"
    }],
    "recoveryTimestamp": 1536305501690,
    "agentEnv": {
        "transparentHugePage": "",
        "hostHealth": {
            "agentTimeStampAtReporting": 1536306543990,
            "activeJavaProcs": [ {
                "command": "../java/jre/bin/java -Xmx256m -Dflume.monitoring.monitorInterval=120 -XX:+HeapDumpOnOutOfMemoryError -Dflume.monitoring.type=com.suning.flume.monitor.AgentMonitorService -cp /opt/sunflower/conf:/opt/sunflower/lib/*:/opt/sunflower/plugins.d/flume-agent/lib/* -Djava.library.path= org.apache.flume.node.Application --conf-file ../conf/custem.conf --name agent",
                "pid": 12252,
                "hadoop": false,
                "user": "root"
            }],
            "liveServices": [{
                "status": "Healthy",
                "name": "ntpd or chronyd",
                "desc": ""
            }]
        },
        "reverseLookup": true,
        "alternatives": [],
        "hasUnlimitedJcePolicy": false,
        "umask": "18",
        "firewallName": "iptables",
        "stackFoldersAndFiles": [],
        "existingUsers": [],
        "firewallRunning": false
    },
    "recoveryReport": {
        "componentReports": [],
        "summary": "RECOVERABLE"
    },
    "componentStatus": []
}

参数解析

nodeStatus：agent节点状态，进入心跳阶段，状态均为健康HEALTHY
timestamp：本次心跳阶段当前时间
hostname：agent所在节点的hostname，config.get('agent', 'hostname_script')获取不到则通过socket.getfqdn()来获取主机名
responseId：本地心跳阶段的唯一ID，如果是第一次心跳则为-1
recoveryTimestamp：暂时不分析recovery
recoveryReport：暂时不分析recovery
reports：对服务组件执行INSTALL/START/STOP等操作得到的结果
componentStatus：服务组件死活检查结果

reports&componentStatus JSON

{
    "componentStatus": [{
        "status": "HEALTHY",
        "componentName": "DATANODE"
    }],
    "reports": [{
            "status": "FAILED",
            "taskId": 3
        },
        {
            "status": "COMPLETE",
            "taskId": 4
        },
        {
            "status": "IN_PROGRESS",
            "stderr": "...",
            "stdout": "...",
            "clusterName": "cc",
            "structuredOut": "{}",
            "roleCommand": "INSTALL",
            "serviceName": "HDFS",
            "role": "DATANODE",
            "actionId": "1-1",
            "taskId": 5,
            "exitCode": 777
        },
        {
            "status": "IN_PROGRESS",
            "stderr": "...",
            "stdout": "...",
            "structuredOut": {
                "var1": "test1",
                "var2": "test2"
            },
            "clusterName": "cc",
            "roleCommand": "INSTALL",
            "serviceName": "HDFS",
            "role": "DATANODE",
            "actionId": "1-1",
            "taskId": 6,
            "exitCode": 777
        }
    ]
}

reports&componentStatus代码片段

      resultReports = []
      resultComponentStatus = []
      for key, item in self.current_state.items():
        command = item[0]
        report = item[1]
        if command ['commandType'] in [ActionQueue.EXECUTION_COMMAND, ActionQueue.BACKGROUND_EXECUTION_COMMAND]:
          if (report['status']) != ActionQueue.IN_PROGRESS_STATUS:
            resultReports.append(report)
            # Removing complete/failed command status from dict
            del self.current_state[key]
          else:
            in_progress_report = self.generate_in_progress_report(command, report)
            resultReports.append(in_progress_report)
        elif command ['commandType'] == ActionQueue.STATUS_COMMAND:
          resultComponentStatus.append(report)
          # Component status is useful once, removing it
          del self.current_state[key]
        elif command ['commandType'] in [ActionQueue.AUTO_EXECUTION_COMMAND]:
          logger.debug("AUTO_EXECUTION_COMMAND task deleted " + str(command['commandId']))
          del self.current_state[key]
          pass
      result = {
        'reports': resultReports,
        'componentStatus': resultComponentStatus
      }
      return result

mounts：运行df以查找主机上的磁盘。 仅适用于Linux平台。 请注意，此解析器忽略任何带空格的文件系统和任何带空格的安装

我们需要通过几个参数来过滤mount：

• 挂载设备不在忽略的列表中
• 当前进程正在运行的用户可以访问
• 它不是文件挂载（docker environment）
• 挂载路径或挂载路径的一部分不在黑名单中

available：可用大小
used：已经使用的大小
percent：已经使用的百分比
device：代表该文件系统是哪个分区，所以列出的是设备名称
mountpoint：挂载点
type：文件系统类型
size：分区的大小
agentEnv：返回有关主机的各种详细信息

"agentEnv": {
        "transparentHugePage": "",//透明大页信息
        "hostHealth": {
            "agentTimeStampAtReporting": 1536306543990,//记录该信息的当前时间戳
            "activeJavaProcs": [ {//通过/proc获取进程来进行分析
                "command": "",///proc/cmdline：系统启动时输入的内核命令行参数
                "pid": 12252,//进程号
                "hadoop": false,//为hadoop或者zookeeper一类的则为true,反之为false
                "user": "root"//执行用户
            }],
            "liveServices": [{//serviceName ntpd status
                "status": "Healthy",
                "name": "ntpd or chronyd",
                "desc": ""
            }]
        },
        "reverseLookup": true,//检查主机`socket.fqdn`是否解析为当前主机ip
        "alternatives": [],//获取项目的替代文件夹路径
        "hasUnlimitedJcePolicy": false,//测试JVM的JCE策略，以查看是否支持无限密钥长度
        "umask": "18",//权限值 777
        "firewallName": "iptables",//iptables
        "stackFoldersAndFiles": [],//获取linux用户目录列表信息
        "existingUsers": [],//获取已存在的以下DEFAULT_USERS 列表中的linux用户
        "firewallRunning": false//查看防火墙信息
    }

DEFAULT_USERS = [
    "hive", "ambari-qa", "oozie", "hbase", "hcat", "mapred",
    "hdfs", "zookeeper", "flume", "sqoop", "sqoop2",
    "hue", "yarn", "tez", "storm", "falcon", "kafka", "knox", "ams",
    "hadoop", "spark", "accumulo", "atlas", "mahout", "ranger", "kms", "zeppelin"
  ]

未完待续.....