I'm looking for information on how many simultaneous disk and node failures a cluster could withstand with the following set up:
1) 29 datanodes
2) 3 jbod data disks each
3) replication set to 3x
I'm guessing it could withstand up to two full node failures and up to 3 disks? How would one determine this?
@msheean253 - For disk failures, it depends on the property "DataNode failed disk tolerance". You can search for the same in the HDFS configs from Ambari. You need to configure based upon Disk failure.
If Replication Factor is set to 3, then you can have max 2 DN down without any impact on the data.