HDFS Support for Multihomed Networks ==================================== This document is targetted to cluster administrators deploying `HDFS` in multihomed networks. Similar support for `YARN`/`MapReduce` is work in progress and will be documented when available. * [HDFS Support for Multihomed Networks](#HDFS_Support_for_Multihomed_Networks) * [Multihoming Background](#Multihoming_Background) * [Fixing Hadoop Issues In Multihomed Environments](#Fixing_Hadoop_Issues_In_Multihomed_Environments) * [Ensuring HDFS Daemons Bind All Interfaces](#Ensuring_HDFS_Daemons_Bind_All_Interfaces) * [Clients use Hostnames when connecting to DataNodes](#Clients_use_Hostnames_when_connecting_to_DataNodes) * [DataNodes use HostNames when connecting to other DataNodes](#DataNodes_use_HostNames_when_connecting_to_other_DataNodes) Multihoming Background ---------------------- In multihomed networks the cluster nodes are connected to more than one network interface. There could be multiple reasons for doing so. 1. **Security**: Security requirements may dictate that intra-cluster traffic be confined to a different network than the network used to transfer data in and out of the cluster. 2. **Performance**: Intra-cluster traffic may use one or more high bandwidth interconnects like Fiber Channel, Infiniband or 10GbE. 3. **Failover/Redundancy**: The nodes may have multiple network adapters connected to a single network to handle network adapter failure. Note that NIC Bonding (also known as NIC Teaming or Link Aggregation) is a related but separate topic. The following settings are usually not applicable to a NIC bonding configuration which handles multiplexing and failover transparently while presenting a single 'logical network' to applications. Fixing Hadoop Issues In Multihomed Environments ----------------------------------------------- ### Ensuring HDFS Daemons Bind All Interfaces By default `HDFS` endpoints are specified as either hostnames or IP addresses. In either case `HDFS` daemons will bind to a single IP address making the daemons unreachable from other networks. The solution is to have separate setting for server endpoints to force binding the wildcard IP address `INADDR_ANY` i.e. `0.0.0.0`. Do NOT supply a port number with any of these settings. **NOTE:** Prefer using hostnames over IP addresses in master/slave configuration files. dfs.namenode.rpc-bind-host 0.0.0.0 The actual address the RPC server will bind to. If this optional address is set, it overrides only the hostname portion of dfs.namenode.rpc-address. It can also be specified per name node or name service for HA/Federation. This is useful for making the name node listen on all interfaces by setting it to 0.0.0.0. dfs.namenode.servicerpc-bind-host 0.0.0.0 The actual address the service RPC server will bind to. If this optional address is set, it overrides only the hostname portion of dfs.namenode.servicerpc-address. It can also be specified per name node or name service for HA/Federation. This is useful for making the name node listen on all interfaces by setting it to 0.0.0.0. dfs.namenode.http-bind-host 0.0.0.0 The actual adress the HTTP server will bind to. If this optional address is set, it overrides only the hostname portion of dfs.namenode.http-address. It can also be specified per name node or name service for HA/Federation. This is useful for making the name node HTTP server listen on all interfaces by setting it to 0.0.0.0. dfs.namenode.https-bind-host 0.0.0.0 The actual adress the HTTPS server will bind to. If this optional address is set, it overrides only the hostname portion of dfs.namenode.https-address. It can also be specified per name node or name service for HA/Federation. This is useful for making the name node HTTPS server listen on all interfaces by setting it to 0.0.0.0. ### Clients use Hostnames when connecting to DataNodes By default `HDFS` clients connect to DataNodes using the IP address provided by the NameNode. Depending on the network configuration this IP address may be unreachable by the clients. The fix is letting clients perform their own DNS resolution of the DataNode hostname. The following setting enables this behavior. dfs.client.use.datanode.hostname true Whether clients should use datanode hostnames when connecting to datanodes. ### DataNodes use HostNames when connecting to other DataNodes Rarely, the NameNode-resolved IP address for a DataNode may be unreachable from other DataNodes. The fix is to force DataNodes to perform their own DNS resolution for inter-DataNode connections. The following setting enables this behavior. dfs.datanode.use.datanode.hostname true Whether datanodes should use datanode hostnames when connecting to other datanodes for data transfer.