Hadoop Best Practices
Last updated: 2019-07-26 17:44:47PDF
In Hadoop, distributed file system HDFS, resource scheduling framework YARN, and iterative computing framework MR. Tencent Cloud's Hadoop version that have integrated with COS, allowing you to access to COS using hadoop fs command lines so as to separate compute and storage apart. Below are some best practices:
For both high-availability (HA) cluster and non-HA cluster, do not format the namenode; otherwise, your data will be lost permanently. Tencent Cloud shall not be responsible under any circumstance for any loss of data caused by formatting the namenode.
The fair scheduler is enabled by default, and you can change the scheduler based on your actual needs.