Dumping YARN Job Logs to COS

Last updated: 2021-07-01 15:34:12

    By default, Hadoop stores YARN job logs in HDFS. Tencent Cloud EMR also provides the ability to store YARN job logs in external storage (COS).


    The EMR cluster needs to support COS. For more information, please see Analyzing Data in HDFS/COS with API.


    1. Modify the configuration in yarn-site.xml and deliver the configuration to all nodes.
    2. Add a new configuration item to core-site.xml and deliver it to all nodes.
    3. Restart all the nodemanager/datanode services in the cluster.
    4. Run the hive/spark job to view the job logs stored in COS.
      hdfs dfs -ls cosn://[bucket_name]/[logs_dirs]