Dagang Wei created YARN-8256:
--------------------------------
Summary: Pluggable provider for node membership management
Key: YARN-8256
URL: https://issues.apache.org/jira/browse/YARN-8256
Project: Hadoop YARN
Issue Type: New Feature
Components: resourcemanager
Affects Versions: 3.0.2, 2.8.3
Reporter: Dagang Wei
h1. Background
[HDFS-7541|https://issues.apache.org/jira/browse/HDFS-7541] introduced a
pluggable provider framework for node membership management, which gives HDFS
the flexibility to have different ways to manage node membership for different
needs.
[org.apache.hadoop.hdfs.server.blockmanagement.HostConfigManager|https://github.com/apache/hadoop/blob/trunk/hadoop-hdfs-project/hadoop-hdfs/src/main/java/org/apache/hadoop/hdfs/server/blockmanagement/HostConfigManager.java]
is the class which provides the abstraction. Currently, there are 2
implementations in the HDFS codebase:
1)
[org.apache.hadoop.hdfs.server.blockmanagement.HostFileManager|https://github.com/apache/hadoop/blob/trunk/hadoop-hdfs-project/hadoop-hdfs/src/main/java/org/apache/hadoop/hdfs/server/blockmanagement/HostFileManager.java]
which uses 2 config files which are defined by the properties dfs.hosts and
dfs.hosts.exclude.
2)
[org.apache.hadoop.hdfs.server.blockmanagement.CombinedHostFileManager|https://github.com/apache/hadoop/blob/trunk/hadoop-hdfs-project/hadoop-hdfs/src/main/java/org/apache/hadoop/hdfs/server/blockmanagement/CombinedHostFileManager.java]
which uses a single JSON file defined by the property dfs.hosts.
dfs.namenode.hosts.provider.classname is the property determining which
implementation is used
h1. Problem
YARN should be consistent with HDFS in terms of pluggable provider for node
membership management. The absence of it makes YARM impossible to have other
config sources, e.g., ZooKeeper, database, etc.
h1. Proposed solution
[org.apache.hadoop.yarn.server.resourcemanager.NodesListManager|https://github.com/apache/hadoop/blob/trunk/hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-resourcemanager/src/main/java/org/apache/hadoop/yarn/server/resourcemanager/NodesListManager.java]
is the class for managing YARN node membership today. It uses
[HostsFileReader|https://github.com/apache/hadoop/blob/trunk/hadoop-common-project/hadoop-common/src/main/java/org/apache/hadoop/util/HostsFileReader.java]
to read config files specified by the property
yarn.resourcemanager.nodes.include-path for nodes to include and
yarn.resourcemanager.nodes.nodes.exclude-path for nodes to exclude.
The proposed solution is to
1) introduce a new interface {color:green}HostsConfigManager{color} which
provides the abstraction for node membership management. Update
{color:green}NodeListManager{color} to depend on
{color:green}HostsConfigManager{color} instead of
{color:green}HostsFileReader{color}. Then create a wrapper class for
{color:green}HostsFileReader{color} which implements the interface.
2) introduce a new config property
{color:green}yarn.resourcemanager.hosts.provider.classname{color} for
specifying the implementation class. Set the default value to the wrapper class
of {color:green}HostsFileReader{color} for backward compatibility between new
code and old config.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
---------------------------------------------------------------------
To unsubscribe, e-mail: yarn-dev-unsubscr...@hadoop.apache.org
For additional commands, e-mail: yarn-dev-h...@hadoop.apache.org
