要了解kerberos除了它的[http://web.mit.edu/kerberos/krb5-1.10/ 官网][[BR]] 找了很多久,比较好的中文说明居然在[http://docs.oracle.com/cd/E26926_01/html/E25889/intro-1.html#scrolltoc oracle-solaris上面],大公司就是大公司 选择一套机器做KDC,测试环境选择了192.168.11.63 - hadooptest-11-63.pconline.ctc[[BR]] Client有四台机器: 192.168.11.64 - hadooptest-11-64.pconline.ctc[[BR]] 192.168.11.65 - hadooptest-11-65.pconline.ctc[[BR]] 192.168.11.66 - hadooptest-11-66.pconline.ctc[[BR]] 192.168.11.67 - hadooptest-11-67.pconline.ctc[[BR]] tar -zxvf krb5-1.10.3.tar.gz[[BR]] cd krb5-1.10.3/src[[BR]] ./configure[[BR]] make[[BR]] make install 1. 编辑 11.63 上/etc/krb5.conf 文件[[BR]] {{{ logging] default = FILE:/var/log/krb5libs.log kdc = FILE:/var/log/krb5kdc.log admin_server = FILE:/var/log/kadmind.log [libdefaults] default_realm = LOCALDOMAIN dns_lookup_realm = false dns_lookup_kdc = false ticket_lifetime = 24h forwardable = yes [realms] LOCALDOMAIN = { kdc = hadooptest-11-63.pconline.ctc:88 admin_server = hadooptest-11-63.pconline.ctc:749 default_domain = localdomain } [domain_realm] .localdomain = LOCALDOMAIN localdomain = LOCALDOMAIN [appdefaults] pam = { debug = false ticket_lifetime = 36000 renew_lifetime = 36000 forwardable = true krb4_convert = false } }}} 2.新建 /usr/local/var/krb5kdc/kdc.conf文件[[BR]] {{{ [kdcdefaults] v4_mode = nopreauth kdc_ports = 750,88 kdc_tcp_ports = 88 [realms] LOCALDOMAIN = { acl_file = /usr/local/var/krb5kdc/kadm5.acl dict_file = /usr/share/dict/words admin_keytab = /usr/local/var/krb5kdc/kadm5.keytab kdc_ports = 750,88 max_life = 1d 0h 0m 0s max_renewable_life = 7d 0h 0m 0s supported_enctypes = des3-hmac-sha1:normal des-cbc-crc:normal des:normal des:v4 des:norealm des:onlyrealm default_principal_flags = +preauth } }}} 3.新建Kerberos数据库[[BR]] {{{ # /usr/local/sbin/kdb5_util create -r LOCALDOMAIN -s[[BR]] }}} 4.在/usr/local/var/krb5kdc/目录下新建kadm5.acl文件,内容如下: */admin@LOCALDOMAIN * 5.开始为KDC设置初始用户信息,这里需要在KDC上执行kadmin.local命令(该命令仅能在KDC上运行,如果你需要在其他机器上管理kerberos的话,直接运行kadmin) # /usr/local/sbin/kadmin.local[[BR]] Enter password forprincipal "admin/admin@LOCALDOMAIN":[[BR]] kadmin.local: addprinc admin/admin@LOCALDOMAIN[[BR]] Re-enter password forprincipal "admin/admin@LOCALDOMAIN":[[BR]] Principal "admin/admin@LOCALDOMAIN"created.[[BR]] 生成admin keytab文件:[[BR]] kadmin.local: ktadd -k /usr/local/var/krb5kdc/kadm5.keytab kadmin/admin kadmin/changepw[[BR]] Entry forprincipal kadmin/admin with kvno 3, encryption type DES-CBC-CRC added to keytab WRFILE:/usr/local/var/krb5kdc/kadm5.keytab.[[BR]] Entry forprincipal kadmin/changepw with kvno 3, encryption type DES-CBC-CRC added to keytab WRFILE:/usr/local/var/krb5kdc/kadm5.keytab.[[BR]] 6.启动KDC和kadmind[[BR]] # /usr/local/sbin/krb5kdc[[BR]] # /usr/local/sbin/kadmind[[BR]] 7.为了使集群内所有机器都有Kerberos工具,你需要在集群中每个机器上安装Kerberos程序。并给出/etc/krb5.conf配置文件,并不需要做其他配置。 8.首先在KDC上为Kerberos添加一个新的管理员hadoop/admin:[[BR]] # /usr/local/sbin/kadmin.local[[BR]] Enter password forprincipal "hadoop/admin@LOCALDOMAIN":[[BR]] kadmin.local: addprinc hadoop/admin@LOCALDOMAIN[[BR]] Re-enter password forprincipal "hadoop/admin@LOCALDOMAIN":[[BR]] Principal "hadoop/admin@LOCALDOMAIN"created.[[BR]] 9.在各机器上增加hadoop用户 /usr/local/bin/kadmin[[BR]] addprinc -randkey host/hadooptest-11-64.pconline.ctc@LOCALDOMAIN[[BR]] addprinc -randkey hadoop/hadooptest-11-64.pconline.ctc@LOCALDOMAIN[[BR]] ''这里遇到一个问题,我们的hadoop core-site.xml使用的短名 hadooptest-11-64,要改为长名 hadooptest-11-64.pconline.ctc,不然kerberos里面会认为不同的实体[[BR]]'' ktadd -k /data/hadoop-1.0.3/conf/hadoop.keytab hadoop/hadooptest-11-64.pconline.ctc@LOCALDOMAIN host/hadooptest-11-64.pconline.ctc@LOCALDOMAIN ---- 查看keytab文件内容,这个不用执行[[BR]] /usr/local/bin/klist -e -k -t hadoop.keytab 10. 修改配置文件,这些是固定的,网上很多地方有说, _HOST类似宏定义,有些地方读配置,有些地方读本机的长名[[BR]] {{{ 配置 core-site.xml hadoop.security.authorization true Is service-level authorization enabled? hadoop.security.authentication kerberos Possible values are simple (no authentication), and kerberos 配置 hdfs-site.xml dfs.https.address hadooptest-11-63.pconline.ctc:50470 dfs.https.port 50470 dfs.block.access.token.enable true dfs.namenode.keytab.file /data/hadoop-1.0.3/conf/hadoop.keytab dfs.namenode.kerberos.principal hadoop/_HOST@LOCALDOMAIN dfs.namenode.kerberos.https.principal host/_HOST@LOCALDOMAIN dfs.secondary.http.address hadooptest-11-64.pconline.ctc:50090 dfs.secondary.https.address 0.0.0.0:50495 dfs.secondary.https.port 50495 dfs.secondary.namenode.keytab.file /data/hadoop-1.0.3/conf/hadoop.keytab dfs.secondary.namenode.kerberos.principal hadoop/_HOST@LOCALDOMAIN dfs.secondary.namenode.kerberos.https.principal host/_HOST@LOCALDOMAIN dfs.datanode.data.dir.perm 700 Permissions for the directories on on the local filesystem where the DFS data node store its blocks. The permissions can either be octal or symbolic. dfs.datanode.address 0.0.0.0:1004 dfs.datanode.http.address 0.0.0.0:1006 dfs.datanode.keytab.file /data/hadoop-1.0.3/conf/hadoop.keytab dfs.datanode.kerberos.principal hadoop/_HOST@LOCALDOMAIN dfs.datanode.kerberos.https.principal host/_HOST@LOCALDOMAIN }}} 11. 启动dfs, /data/hadoop-1.0.3/bin/hadoop-daemon.sh start namenode /data/hadoop-1.0.3/bin/hadoop-daemons.sh --config /data/hadoop-1.0.3/conf/ --hosts masters start secondarynamenode hadoop-env.sh 新增一行 export HADOOP_SECURE_DN_USER=hadoop 指定以hadoop身份启动datanode 启动datanode遇到一个bug[[BR]] https://issues.apache.org/jira/browse/HDFS-3402 要修改bin/hadoop文件,使用sudo 来启动datanode, datanode 启动完有两个进程,一个是root的SecureDataNodeStarter,一个是hadoop的SecureDataNodeStarter 12.修改mapred-site.xml 文件 {{{ mapreduce.jobtracker.kerberos.principal hadoop/_HOST@LOCALDOMAIN mapreduce.jobtracker.kerberos.https.principal host/_HOST@LOCALDOMAIN mapreduce.jobtracker.keytab.file /data/hadoop-1.0.3/conf/hadoop.keytab mapreduce.tasktracker.kerberos.principal hadoop/_HOST@LOCALDOMAIN mapreduce.tasktracker.kerberos.https.principal host/_HOST@LOCALDOMAIN mapreduce.tasktracker.keytab.file /data/hadoop-1.0.3/conf/hadoop.keytab mapred.task.tracker.task-controller org.apache.hadoop.mapred.DefaultTaskController mapreduce.tasktracker.group hadoop }}} 13.启动mapred 可能要修改 taskcontroller.cfg start-mapred.sh ---- 14.配置hbase {{{ hbase.regionserver.kerberos.principal hadoop/_HOST@LOCALDOMAIN hbase.regionserver.keytab.file /data/hadoop-1.0.3/conf/hadoop.keytab hbase.master.kerberos.principal hadoop/_HOST@LOCALDOMAIN hbase.master.keytab.file /data/hadoop-1.0.3/conf/hadoop.keytab hbase.security.authentication kerberos hbase.security.authorization true hbase.coprocessor.region.classes org.apache.hadoop.hbase.security.token.TokenProvider hbase.coprocessor.master.classes org.apache.hadoop.hbase.security.access.AccessController hbase.coprocessor.region.classes org.apache.hadoop.hbase.security.token.TokenProvider, org.apache.hadoop.hbase.security.access.AccessController }}} 不过,装到最后还是悲剧了,因为我们的kerberos版本太高 [https://ccp.cloudera.com/display/CDHDOC/Appendix+A+-+Troubleshooting 参考 cloudera网站] 里面机关重重,各种版本一不小心就悲剧了