大數(shù)據(jù)開發(fā)系列四：hadoop& flink 配置kerberos認(rèn)證

IT那活兒發(fā)布于2023-01-11 13:19 / 4022人閱讀

點擊上方“IT那活兒”公眾號，關(guān)注后了解更多內(nèi)容，不管IT什么活兒，干就完了！！！

一

前言

我們之前的大數(shù)據(jù)開發(fā)系列文章介紹了kerberos是如何安裝與使用，本次在已安裝的kerberos服務(wù)基礎(chǔ)上對hadoop 與flink組件進(jìn)行kerberos認(rèn)證配置。

《大數(shù)據(jù)開發(fā)系列一：基于ip2region實現(xiàn)在流式數(shù)據(jù)ip歸屬轉(zhuǎn)換》
《大數(shù)據(jù)開發(fā)系列二：自定義flink Metric kafka Reporter》
《大數(shù)據(jù)開發(fā)系列三：kerberos認(rèn)證實踐》

二

環(huán)境依賴

類型	主機	主機hostname	安裝組件
kerberos服務(wù)端	192.168.199.102	bigdata-03	krb5-server krb5-workstation krb5-libs krb5-devel
kerberos客戶端	192.168.199.104	bigdata-05	krb5-workstation krb5-devel
hadoop環(huán)境	192.168.199.104	bigdata-05	hadoop-3.3.3

三

hadoop認(rèn)證配置

3.1 創(chuàng)建principle添加用戶

hadoop的kerberos認(rèn)證，一般需要配置三種principle，分別是 hadoop, host, HTTP。

格式為：用戶名/主機hostname@HADOOP.COM。

如果現(xiàn)有的HDFS和YARN守護(hù)程序用的是同一個用戶身份運行，可以配置成一個hadoop principle。

kadmin.local -q "addprinc -randkeyhadoop/bigdata-03@HADOOP.COM"
kadmin.local -q "addprinc -randkeyhadoop/bigdata-05@HADOOP.COM"

listprincs/list_principals 查詢所有用戶。

3.2 創(chuàng)建keytab密碼文件

kadmin.local -q "xst -k /root/keytabs/kerberos/hadoop.keytabhadoop/bigdata-03@HADOOP.COM"
kadmin.local -q "xst -k /root/keytabs/kerberos/hadoop.keytab hadoop/bigdata-05@HADOOP.COM"

查看：

klist -kt /root/keytabs/kerberos/hadoop.keytab
klist -kt /home/gpadmin/hadoop.keytab

3.3 hadoop配置修改

1）core-site.xml 配置新增

<property>
<name>hadoop.security.authenticationname>
<value>kerberosvalue>
property>
<property>
<name>hadoop.security.authorizationname>
<value>truevalue>
property>

2）hdfs-site.xml 配置新增


    <property>
      <name>dfs.block.access.token.enablename>
      <value>truevalue>
    property>
                <property>
                  <name>dfs.permissions.enabledname>
                  <value>falsevalue>
                property> 
    
    <property>
      <name>dfs.namenode.kerberos.principalname>
      <value>hadoop/_HOST@EXAMPLE.COMvalue>
    property>
    
    <property>
      <name>dfs.namenode.keytab.filename>
      <value>/home/gpadmin/hadoop.keytabvalue>
    property>

      
    <property>
      <name>dfs.secondary.namenode.kerberos.principalname>
      <value>hadoop/_HOST@EXAMPLE.COMvalue>
    property>
    
    <property>
      <name>dfs.secondary.namenode.keytab.filename>
      <value>/home/gpadmin/hadoop.keytabvalue>
    property>

    
    <property>
      <name>dfs.web.authentication.kerberos.principalname>
      <value>hadoop/_HOST@EXAMPLE.COMvalue>
    property>
        
    <property>
      <name>dfs.web.authentication.kerberos.keytabname>
      <value>/home/gpadmin/hadoop.keytabvalue>
    property>


    
    <property>
      <name>dfs.datanode.kerberos.principalname>
      <value>hadoop/_HOST@EXAMPLE.COMvalue>
    property>

    
    <property>
      <name>dfs.datanode.keytab.filename>
      <value>/home/gpadmin/hadoop.keytabvalue>
    property>


    
    <property>
      <name>dfs.data.transfer.protectionname>
      <value>authenticationvalue>
    property>

    
    <property>
      <name>dfs.http.policyname>
      <value>HTTPS_ONLYvalue>
      <description>所有開啟的web頁面均使用https, 細(xì)節(jié)在ssl server 和client那個配置文件內(nèi)配置description>
    property>

3）yarn-site.xml配置新增


<property>
  <name>yarn.resourcemanager.principalname>
  <value>hadoop/_HOST@EXAMPLE.COMvalue>
property>


<property>
  <name>yarn.resourcemanager.keytabname>
  <value>/home/gpadmin/hadoop.keytabvalue>
property>


<property>
  <name>yarn.nodemanager.principalname>
  <value>hadoop/_HOST@EXAMPLE.COMvalue>
property>


<property>
  <name>yarn.nodemanager.keytabname>
  <value>/home/gpadmin/hadoop.keytabvalue>
property>

4）mapred.xml配置新增


<property>
<name>mapreduce.jobhistory.principalname>
<value>hadoop/_HOST@EXAMPLE.COMvalue>
property>


<property>
<name>mapreduce.jobhistory.keytabname>
<value>/home/gpadmin/hadoop.keytabvalue>
property>

5）ssl-server配置新增

<property>
<name>ssl.server.truststore.locationname>
<value>/home/gpadmin/kerberos_https/keystorevalue>
<description>Truststore to be used by NN and DN. Must be specified.
description>
property>

<property>
<name>ssl.server.truststore.passwordname>
<value>passwordvalue>
<description>Optional. Default value is "".
description>
property>

<property>
<name>ssl.server.truststore.typename>
<value>jksvalue>
<description>Optional. The keystore file format, default value is "jks".
description>
property>

<property>
<name>ssl.server.truststore.reload.intervalname>
<value>10000value>
<description>Truststore reload check interval, in milliseconds.
Default value is 10000 (10 seconds).
description>
property>

<property>
<name>ssl.server.keystore.locationname>
<value>/home/gpadmin/kerberos_https/keystorevalue>
<description>Keystore to be used by NN and DN. Must be specified.
description>
property>

<property>
<name>ssl.server.keystore.passwordname>
<value>passwordvalue>
<description>Must be specified.
description>
property>

<property>
<name>ssl.server.keystore.keypasswordname>
<value>passwordvalue>
<description>Must be specified.
description>
property>

<property>
<name>ssl.server.keystore.typename>
<value>jksvalue>
<description>Optional. The keystore file format, default value is "jks".
description>
property>

<property>
<name>ssl.server.exclude.cipher.listname>
<value>TLS_ECDHE_RSA_WITH_RC4_128_SHA,SSL_DHE_RSA_EXPORT_WITH_DES40_CBC_SHA,
SSL_RSA_WITH_DES_CBC_SHA,SSL_DHE_RSA_WITH_DES_CBC_SHA,
SSL_RSA_EXPORT_WITH_RC4_40_MD5,SSL_RSA_EXPORT_WITH_DES40_CBC_SHA,
SSL_RSA_WITH_RC4_128_MD5value>
<description>Optional. The weak security cipher suites that you want excluded
from SSL communication.description>
property>

3.4 https證書配置

keytool -keystore keystore -alias hadoop -validity 365000 -
keystore/home/gpadmin/kerberos_https/keystore/keystore -
genkey -keyalg RSA -keysize 2048 -dname "CN=hadoop, 
OU=shsnc, O=snc, L=hunan, ST=changsha, C=CN"

生成keystore 證書文件。

3.5 認(rèn)證測試

查看 hdfs 目錄：hdfs dfs -ls /

報錯信息：2022-11-22 10:22:15,444 WARN ipc.Client: Exception encountered while connecting to the server

org.apache.hadoop.security.AccessControlException: Client cannot authenticate via:[TOKEN, KERBEROS]

說明已加了認(rèn)證后不能直接訪問，客戶端先進(jìn)行認(rèn)證才能正常訪問目錄結(jié)構(gòu)。

kinit  -kt /home/gpadmin/hadoop.keytabhadoop/bigdata-05@HADOOP.COM

四

flink認(rèn)證配置

4.1 認(rèn)證用戶配置

如果hdfs-site.xml 屬性項配置：

<property>
<name>dfs.permissions.enabledname>
<value>truevalue>
property>

為true時，新建憑證為hadoop 安裝用戶，如以gpadmin用戶安裝了hadoop。
kadmin.local -q "xst -k /root/keytabs/kerberos/hadoop.keytab gpadmin@HADOOP.COM"
為false時，新建憑證可以不是hadoop 安裝用戶。
kadmin.local -q "xst -k /root/keytabs/kerberos/hadoop.keytab xx@HADOOP.COM"

驗證：

klist -kt /root/keytabs/kerberos/hadoop.keytab

4.2 flink-conf.yaml 新增配置

security.kerberos.login.use-ticket-cache: true
security.kerberos.login.keytab: /home/gpadmin/hadoop.keytab
security.kerberos.login.principal: gpadmin@HADOOP.COM
security.kerberos.login.contexts: Client

4.3 認(rèn)證測試

flink run -m yarn-cluster
-p 1
-yjm 1024
-ytm 1024
-ynm amp_zabbix
-c com.shsnc.fk.task.tokafka.ExtratMessage2KafkaTask
-yt /home/gpadmin/jar_repo/config/krb5.conf
-yD env.java.opts.jobmanager=-Djava.security.krb5.conf=krb5.conf
-yD env.java.opts.taskmanager=-Djava.security.krb5.conf=krb5.conf
-yD security.kerberos.login.keytab=/home/gpadmin/hadoop.keytab
-yD security.kerberos.login.principal=gpadmin@HADOOP.COM
$jarname