Kluster E-MapReduce (EMR) menggunakan tata letak direktori yang konsisten untuk binari layanan, log, dan file konfigurasi. Login ke node master kluster Anda untuk mengakses jalur-jalur ini secara langsung.
Jalur tersebut berbeda berdasarkan jenis kluster: DataLake cluster dan Hadoop cluster.
Kluster DataLake
Direktori layanan
Layanan diinstal di bawah /opt/apps/<SERVICE>/<service>-current.
| Layanan | Direktori |
|---|---|
| HDFS | /opt/apps/HDFS/hdfs-current |
| Hive | /opt/apps/HIVE/hive-current |
| Hudi | /opt/apps/HUDI/hudi-current |
| YARN | /opt/apps/YARN/yarn-current |
| Presto | /opt/apps/PRESTO/presto-current |
| Ranger | /opt/apps/RANGER/ranger-current |
Untuk mencari direktori instalasi layanan apa pun, jalankan:
env |grep <service-name>
Misalnya, jalankan env | grep hive untuk menemukan lokasi instalasi Hive. Output-nya mirip dengan:
JINDOTABLE_EXTRA_CLASSPATH=/opt/apps/METASTORE/metastore-current/hive2
HIVE_HOME=/opt/apps/HIVE/hive-current
HIVE_LOG_DIR=/var/log/taihao-apps/hive
HIVE_CONF_DIR=/etc/taihao-apps/hive-conf
PATH=/opt/apps/JINDOSDK/jindosdk-current/bin:/opt/apps/HADOOP-COMMON/hadoop-common-current/bin:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/opt/apps/HIVE/hive-current/bin:/opt/apps/JINDODATA/jindodata-current/bin:/opt/apps/JINDODATA/jindodata-current/sbin:/opt/apps/SPARK-EXTENSION/spark-extension-current/bin:/opt/apps/SPARK3/spark-current/bin:/root/bin
OLDPWD=/var/log/emr/hive
Direktori log
Log disimpan di bawah /var/log/emr/<service>/.
| Layanan | Direktori log |
|---|---|
| Spark | /var/log/emr/spark/ |
| Hive | /var/log/emr/hive/ |
| YARN | /var/log/emr/yarn/ |
| JindoSDK | /var/log/emr/jindosdk/ |
Direktori file konfigurasi
File konfigurasi disimpan di bawah /etc/emr/<service>-conf/.
| Layanan | Direktori konfigurasi |
|---|---|
| HDFS | /etc/emr/hdfs-conf/ |
| Spark | /etc/emr/spark-conf/ |
| Hive | /etc/emr/hive-conf/ |
| Hudi | /etc/emr/hudi-conf/ |
| Knox | /etc/emr/knox-conf/ |
| YARN | /etc/emr/hadoop-conf/ |
| ZooKeeper | /etc/emr/zookeeper-conf/ |
Kluster Hadoop
Direktori layanan
Layanan diinstal di bawah /usr/lib/<service>-current.
| Layanan | Direktori |
|---|---|
| Hadoop | /usr/lib/hadoop-current |
| Spark | /usr/lib/spark-current |
| Hive | /usr/lib/hive-current |
| Flink | /usr/lib/flink-current |
| Flume | /usr/lib/flume-current |
Untuk mencari direktori instalasi layanan apa pun, jalankan:
env |grep <service-name>
Misalnya, jalankan env | grep spark untuk menemukan lokasi instalasi Spark. Output-nya mirip dengan:
SPARK_HOME=/usr/lib/spark-current
SPARK_CONF_DIR=/etc/ecm/spark-conf
SPARK_LOG_DIR=/mnt/disk1/log/spark
PATH=/usr/lib/sqoop-current/bin:/usr/lib/jindosdk-current/bin:/usr/lib/hudi-current/bin:/usr/lib/hive-current/hcatalog/bin:/usr/lib/hive-current/bin:/usr/lib/datafactory-current/bin:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/usr/lib/flow-agent-current/bin:/usr/lib/hadoop-current/bin:/usr/lib/hadoop-current/sbin:/usr/lib/jindodata-current//bin:/usr/lib/jindodata-current//sbin:/usr/lib/spark-current/bin:/usr/lib/hadoop-current/bin:/usr/lib/hadoop-current/sbin:/root/bin
HADOOP_CLASSPATH=/opt/apps/extra-jars/*:/usr/lib/spark-current/yarn/spark-3.2.1-yarn-shuffle.jar
SPARK_PID_DIR=/usr/lib/spark-current/pids
Direktori log
Log disimpan di bawah /mnt/disk1/log/<service>/. Beberapa layanan menulis ke direktori berbeda tergantung pada jenis node.
| Layanan | Direktori log | Node |
|---|---|---|
| YARN ResourceManager | /mnt/disk1/log/hadoop-yarn |
Node master |
| YARN NodeManager | /mnt/disk1/log/hadoop-yarn |
Node core atau node task |
| HDFS NameNode | /mnt/disk1/log/hadoop-hdfs |
Node master |
| HDFS DataNode | /mnt/disk1/log/hadoop-hdfs |
Node core atau node task |
| Hive | /mnt/disk1/log/hive |
Master node |
| ESS | /mnt/disk1/log/ess/ |
Node master, node core, atau node task |
Direktori file konfigurasi
File konfigurasi disimpan di bawah /etc/ecm/<service>-conf/.
| Layanan | Direktori konfigurasi |
|---|---|
| Hadoop | /etc/ecm/hadoop-conf/ |
| Spark | /etc/ecm/spark-conf/ |
| Hive | /etc/ecm/hive-conf/ |
| Flink | /etc/ecm/flink-conf/ |
| Flume | /etc/ecm/flume-conf/ |
Akses SSH bersifat read-only untuk file konfigurasi. Untuk mengubah parameter konfigurasi, login ke Konsol EMR.
Direktori data
| Jenis data | Direktori |
|---|---|
| Data cache JindoFS | /mnt/disk1/jindodata/ |