hadoop-core-site.xml配置文件详解

hadoop.tmp.dir

A base for other temporary directories. 只可以设置一个值；建议设置到一个足够空间的地方，而不是默认的/tmp下 服务端参数，修改需重启

/tmp/hadoop-${user.name}

io.native.lib.available

Should native hadoop libraries, if present, be used. 是否启动Hadoop的本地库，默认启用。本地库可以加快基本操作，例如IO，压缩等。

true

hadoop.http.filter.initializers

A comma separated list of class names. Each class in the list must extend org.apache.hadoop.http.FilterInitializer. The corresponding Filter will be initialized. Then, the Filter will be applied to all user facing jsp and servlet web pages. The ordering of the list defines the ordering of the filters. Hadoop的Http服务中，用逗号分隔的一组过滤器类名，每个类必须扩展自org.apache.hadoop.http.FilterInitializer。 这些组件被初始化，应用于全部用户的JSP和Servlet页面。 列表中定义的顺序就是过滤器被调用的顺序。

org.apache.hadoop.http.lib.StaticUserWebFilter

hadoop.security.authorization

Is service-level authorization enabled? 是否开启安全服务验证 建议不开启。认证操作比较复杂，在公司内部网络下，重要性没那么高

false

hadoop.security.instrumentation.requires.admin

Indicates if administrator ACLs are required to access instrumentation servlets (JMX, METRICS, CONF, STACKS).

false

hadoop.security.authentication

Possible values are simple (no authentication), and kerberos 安全验证规则，可以是simple和kerberos。simple意味着不验证。

simple

hadoop.security.group.mapping

Class for user to group mapping (get groups for a given user) for ACL. The default implementation, org.apache.hadoop.security.JniBasedUnixGroupsMappingWithFallback, will determine if the Java Native Interface (JNI) is available. If JNI is available the implementation will use the API within hadoop to resolve a list of groups for a user. If JNI is not available then the shell implementation, ShellBasedUnixGroupsMapping, is used. This implementation shells out to the Linux/Unix environment with the bash -c groups command to resolve a list of groups for a user. user到group的映射类。ACL用它以给定user获取group。默认实现是 org.apache.hadoop.security.JniBasedUnixGroupsMappingWithFallback, 如果JNI有效，它将发挥作用，使用Hadoop的API去获取user的groups列表。如果JNI无效，会使用另一个基于shell的实现, ShellBasedUnixGroupsMapping。这个实现是基于Linux、Unix的shell的环境。

org.apache.hadoop.security.JniBasedUnixGroupsMappingWithFallback

hadoop.security.groups.cache.secs

This is the config controlling the validity of the entries in the cache containing the user->group mapping. When this duration has expired, then the implementation of the group mapping provider is invoked to get the groups of the user and then cached back. user到gourp映射缓存的有效时间。如果超时，会再次调用去获取新的映射关系然后缓存起来。

300

hadoop.security.groups.cache.warn.after.ms

If looking up a single user to group takes longer than this amount of milliseconds, we will log a warning message.

5000

hadoop.security.group.mapping.ldap.url

The URL of the LDAP server to use for resolving user groups when using the LdapGroupsMapping user to group mapping.

hadoop.security.group.mapping.ldap.ssl

Whether or not to use SSL when connecting to the LDAP server.

false

hadoop.security.group.mapping.ldap.ssl.keystore

File path to the SSL keystore that contains the SSL certificate required by the LDAP server.

hadoop.security.group.mapping.ldap.ssl.keystore.password.file

The path to a file containing the password of the LDAP SSL keystore. IMPORTANT: This file should be readable only by the Unix user running the daemons.

hadoop.security.group.mapping.ldap.bind.user

The distinguished name of the user to bind as when connecting to the LDAP server. This may be left blank if the LDAP server supports anonymous binds.

hadoop.security.group.mapping.ldap.bind.password.file

The path to a file containing the password of the bind user. IMPORTANT: This file should be readable only by the Unix user running the daemons.

hadoop.security.group.mapping.ldap.base

The search base for the LDAP connection. This is a distinguished name, and will typically be the root of the LDAP directory.

hadoop.security.group.mapping.ldap.search.filter.user

An additional filter to use when searching for LDAP users. The default will usually be appropriate for Active Directory installations. If connecting to an LDAP server with a non-AD schema, this should be replaced with (&(objectClass=inetOrgPerson)(uid={0}). {0} is a special string used to denote where the username fits into the filter.

(&(objectClass=user)(sAMAccountName={0}))

hadoop.security.group.mapping.ldap.search.filter.group

An additional filter to use when searching for LDAP groups. This should be changed when resolving groups against a non-Active Directory installation. posixGroups are currently not a supported group class.

(objectClass=group)

hadoop.security.group.mapping.ldap.search.attr.member

The attribute of the group object that identifies the users that are members of the group. The default will usually be appropriate for any LDAP installation.

member

hadoop.security.group.mapping.ldap.search.attr.group.name

The attribute of the group object that identifies the group name. The default will usually be appropriate for all LDAP systems.

cn

hadoop.security.group.mapping.ldap.directory.search.timeout

The attribute applied to the LDAP SearchControl properties to set a maximum time limit when searching and awaiting a result. Set to 0 if infinite wait period is desired. Default is 10 seconds. Units in milliseconds.

10000

hadoop.security.service.user.name.key

For those cases where the same RPC protocol is implemented by multiple servers, this configuration is required for specifying the principal name to use for the service when the client wishes to make an RPC call. 如果相同的RPC协议被多个Server实现，这个配置是用来指定在客户端进行RPC调用时，使用哪个principal name去联系服务器。不建议使用

null

hadoop.security.uid.cache.secs

This is the config controlling the validity of the entries in the cache containing the userId to userName and groupId to groupName used by NativeIO getFstat(). 安全选项。不建议使用

14400

hadoop.rpc.protection

A comma-separated list of protection values for secured sasl connections. Possible values are authentication, integrity and privacy. authentication means authentication only and no integrity or privacy; integrity implies authentication and integrity are enabled; and privacy implies all of authentication, integrity and privacy are enabled. hadoop.security.saslproperties.resolver.class can be used to override the hadoop.rpc.protection for a connection at the server side. rpc连接保护。可取的值有authentication（认证）, integrity（完整） and privacy（隐私）。 不建议使用

authentication

hadoop.security.saslproperties.resolver.class

SaslPropertiesResolver used to resolve the QOP used for a connection. If not specified, the full set of values specified in hadoop.rpc.protection is used while determining the QOP used for the connection. If a class is specified, then the QOP values returned by the class will be used while determining the QOP used for the connection.

hadoop.work.around.non.threadsafe.getpwuid

Some operating systems or authentication modules are known to have broken implementations of getpwuid_r and getpwgid_r, such that these calls are not thread-safe. Symptoms of this problem include JVM crashes with a stack trace inside these functions. If your system exhibits this issue, enable this configuration parameter to include a lock around the calls as a workaround. An incomplete list of some systems known to have this issue is available at http://wiki.apache.org/hadoop/KnownBrokenPwuidImplementations 一些系统已知在调用getpwuid_r和getpwgid_r有问题，这些调用是非线程安全的。这个问题的主要表现特征是JVM崩溃。如果你的系统有这些问题，开启这个选项。默认是关闭的。

false

hadoop.kerberos.kinit.command

Used to periodically renew Kerberos credentials when provided to Hadoop. The default setting assumes that kinit is in the PATH of users running the Hadoop client. Change this to the absolute path to kinit if this is not the case. 用来定期的向Hadoop提供新的Kerberos证书。所提供命令需要能够在运行Hadoop客户端的用户路径中查找到，否则，请指定绝对路径。不建议使用

kinit

hadoop.security.auth_to_local

Maps kerberos principals to local user names 映射kerberos principals（代理人）到本地用户名

io.file.buffer.size

The size of buffer for use in sequence files. The size of this buffer should probably be a multiple of hardware page size (4096 on Intel x86), and it determines how much data is buffered during read and write operations. 在读写文件时使用的缓存大小。这个大小应该是内存Page的倍数。建议1M

4096

io.bytes.per.checksum

The number of bytes per checksum. Must not be larger than io.file.buffer.size. 每次进行校验和检查的字节数。一定不能大于io.file.buffer.size.

512

io.skip.checksum.errors

If true, when a checksum error is encountered while reading a sequence file, entries are skipped, instead of throwing an exception. 是否跳过校验和错误，默认是否，校验和异常时会抛出错误。

false

io.compression.codecs

A comma-separated list of the compression codec classes that can be used for compression/decompression. In addition to any classes specified with this property (which take precedence), codec classes on the classpath are discovered using a Java ServiceLoader. 压缩和解压缩编码类列表，用逗号分隔。这些类是使用Java ServiceLoader加载。

io.compression.codec.bzip2.library

The native-code library to be used for compression and decompression by the bzip2 codec. This library could be specified either by by name or the full pathname. In the former case, the library is located by the dynamic linker, usually searching the directories specified in the environment variable LD_LIBRARY_PATH. The value of “system-native” indicates that the default system library should be used. To indicate that the algorithm should operate entirely in Java, specify “java-builtin”.

system-native

io.serializations

A list of serialization classes that can be used for obtaining serializers and deserializers. 序列化类列表，可以被用来获取序列化器和反序列化器（serializers and deserializers）。

org.apache.hadoop.io.serializer.WritableSerialization,org.apache.hadoop.io.serializer.avro.AvroSpecificSerialization,org.apache.hadoop.io.serializer.avro.AvroReflectSerialization

io.seqfile.local.dir

The local directory where sequence file stores intermediate data files during merge. May be a comma-separated list of directories on different devices in order to spread disk i/o. Directories that do not exist are ignored. 本地文件目录。sequence file在merge过程中存储内部数据的地方。可以是逗号分隔的一组目录。最好在不同磁盘以分散IO。实际不存在的目录会被忽略。

${hadoop.tmp.dir}/io/local

io.map.index.skip

Number of index entries to skip between each entry. Zero by default. Setting this to values larger than zero can facilitate opening large MapFiles using less memory. 跳过的索引实体数量在entry之间。默认是0。设置大于0的值可以用更少的内存打开大MapFiles。注意：MpaFile是一组Sequence文件，是排序后的，带内部索引的文件

0

io.map.index.interval

MapFile consist of two files - data file (tuples) and index file (keys). For every io.map.index.interval records written in the data file, an entry (record-key, data-file-position) is written in the index file. This is to allow for doing binary search later within the index file to look up records by their keys and get their closest positions in the data file. MapFile包含两个文件，数据文件和索引文件。每io.map.index.interval个记录写入数据文件，一条记录(行key，数据文件位置)写入索引文件。

128

fs.defaultFS

The name of the default file system. A URI whose scheme and authority determine the FileSystem implementation. The uri’s scheme determines the config property (fs.SCHEME.impl) naming the FileSystem implementation class. The uri’s authority is used to determine the host, port, etc. for a filesystem. 默认文件系统的名称。URI形式。uri’s的scheme需要由(fs.SCHEME.impl)指定文件系统实现类。 uri’s的authority部分用来指定host, port等。默认是本地文件系统。 HA方式，这里设置服务名，例如：hdfs://mycluster1 HDFS的客户端访问HDFS需要此参数。

file:///

fs.default.name

Deprecated. Use (fs.defaultFS) property instead 过时。使用(fs.defaultFS)代替

file:///

fs.trash.interval

Number of minutes after which the checkpoint gets deleted. If zero, the trash feature is disabled. This option may be configured both on the server and the client. If trash is disabled server side then the client side configuration is checked. If trash is enabled on the server side then the value configured on the server is used and the client configuration value is ignored. 以分钟为单位的垃圾回收时间，垃圾站中数据超过此时间，会被删除。如果是0，垃圾回收机制关闭。可以配置在服务器端和客户端。如果在服务器端配置trash无效，会检查客户端配置。如果服务器端配置有效，客户端配置会忽略。 建议开启，建议4320（3天） 垃圾回收站，如有同名文件被删除，会给文件顺序编号，例如：a.txt,a.txt(1)

0

fs.trash.checkpoint.interval

Number of minutes between trash checkpoints. Should be smaller or equal to fs.trash.interval. If zero, the value is set to the value of fs.trash.interval. Every time the checkpointer runs it creates a new checkpoint out of current and removes checkpoints created more than fs.trash.interval minutes ago. 以分钟为单位的垃圾回收检查间隔。应该小于或等于fs.trash.interval。如果是0，值等同于fs.trash.interval。每次检查器运行，会创建新的检查点。 建议设置为60（1小时）

0

fs.AbstractFileSystem.file.impl

The AbstractFileSystem for file: uris.

org.apache.hadoop.fs.local.LocalFs

fs.AbstractFileSystem.hdfs.impl

The FileSystem for hdfs: uris. 文件系统实现类：hdfs

org.apache.hadoop.fs.Hdfs

fs.AbstractFileSystem.viewfs.impl

The AbstractFileSystem for view file system for viewfs: uris (ie client side mount table:).

org.apache.hadoop.fs.viewfs.ViewFs

fs.ftp.host

FTP filesystem connects to this server

0.0.0.0

fs.ftp.host.port

FTP filesystem connects to fs.ftp.host on this port

21

fs.df.interval

Disk usage statistics refresh interval in msec.

60000

fs.du.interval

File space usage statistics refresh interval in msec.

600000

fs.s3.block.size

Block size to use when writing files to S3.

67108864

fs.s3.buffer.dir

Determines where on the local filesystem the S3 filesystem should store files before sending them to S3 (or after retrieving them from S3).

${hadoop.tmp.dir}/s3

fs.s3.maxRetries

The maximum number of retries for reading or writing files to S3, before we signal failure to the application.

4

fs.s3.sleepTimeSeconds

The number of seconds to sleep between each S3 retry.

10

fs.swift.impl

The implementation class of the OpenStack Swift Filesystem

org.apache.hadoop.fs.swift.snative.SwiftNativeFileSystem

fs.automatic.close

By default, FileSystem instances are automatically closed at program exit using a JVM shutdown hook. Setting this property to false disables this behavior. This is an advanced option that should only be used by server applications requiring a more carefully orchestrated shutdown sequence. 默认的，文件系统实例在程序退出时自动关闭，通过JVM shutdown hook方式。可以把此属性设置为false取消这种操作。这是一个高级选项，需要使用者特别关注关闭顺序。不要关闭

true

fs.s3n.block.size

Block size to use when reading files using the native S3 filesystem (s3n: URIs).

67108864

fs.s3n.multipart.uploads.enabled

Setting this property to true enables multiple uploads to native S3 filesystem. When uploading a file, it is split into blocks if the size is larger than fs.s3n.multipart.uploads.block.size.

false

fs.s3n.multipart.uploads.block.size

The block size for multipart uploads to native S3 filesystem. Default size is 64MB.

67108864

fs.s3n.multipart.copy.block.size

The block size for multipart copy in native S3 filesystem. Default size is 5GB.

5368709120

io.seqfile.compress.blocksize

The minimum block size for compression in block compressed SequenceFiles. SequenceFiles以块压缩方式压缩时，块大小大于此值时才启动压缩。

1000000

io.seqfile.lazydecompress

Should values of block-compressed SequenceFiles be decompressed only when necessary. 懒惰解压，仅在必要时解压，仅对块压缩的SequenceFiles有效。

true

io.seqfile.sorter.recordlimit

The limit on number of records to be kept in memory in a spill in SequenceFiles.Sorter 在SequenceFiles.Sorter spill过程中，保存在内存中的记录数

1000000

io.mapfile.bloom.size

The size of BloomFilter-s used in BloomMapFile. Each time this many keys is appended the next BloomFilter will be created (inside a DynamicBloomFilter). Larger values minimize the number of filters, which slightly increases the performance, but may waste too much space if the total number of keys is usually much smaller than this number.

1048576

io.mapfile.bloom.error.rate

The rate of false positives in BloomFilter-s used in BloomMapFile. As this value decreases, the size of BloomFilter-s increases exponentially. This value is the probability of encountering false positives (default is 0.5%).

0.005

hadoop.util.hash.type

The default implementation of Hash. Currently this can take one of the two values: ‘murmur’ to select MurmurHash and ‘jenkins’ to select JenkinsHash. 默认Hash算法实现. ‘murmur’:MurmurHash, ‘jenkins’:JenkinsHash.

murmur

ipc.client.idlethreshold

Defines the threshold number of connections after which connections will be inspected for idleness. 连接数阀值，超过此值，需要进行空闲连接检查

4000

ipc.client.kill.max

Defines the maximum number of clients to disconnect in one go. 定义客户端最大数量，超过会被断开连接

10

ipc.client.connection.maxidletime

The maximum time in msec after which a client will bring down the connection to the server. 毫秒，最大时间，超过后客户端会断开和服务器的连接。

10000

ipc.client.connect.max.retries

Indicates the number of retries a client will make to establish a server connection. 客户端连接重试次数。

10

ipc.client.connect.retry.interval

Indicates the number of milliseconds a client will wait for before retrying to establish a server connection.

1000

ipc.client.connect.timeout

Indicates the number of milliseconds a client will wait for the socket to establish a server connection.

20000

ipc.client.connect.max.retries.on.timeouts

Indicates the number of retries a client will make on socket timeout to establish a server connection. 在连接超时后，客户端连接重试次数

45

ipc.server.listen.queue.size

Indicates the length of the listen queue for servers accepting client connections. 定义服务器端接收客户端连接的监听队列长度

128

ipc.server.tcpnodelay

Turn on/off Nagle’s algorithm for the TCP socket connection on the server. Setting to true disables the algorithm and may decrease latency with a cost of more/smaller packets. 在服务器端开启/关闭Nagle’s算法，此算法可以延迟小数据包发送，从而达到网络流量更有效利用。但是这对小数据包是不利的。默认关闭。建议false，即开启Nagle算法

false

ipc.client.tcpnodelay

Turn on/off Nagle’s algorithm for the TCP socket connection on the client. Setting to true disables the algorithm and may decrease latency with a cost of more/smaller packets.

false

hadoop.rpc.socket.factory.class.default

Default SocketFactory to use. This parameter is expected to be formatted as “package.FactoryClassName”.

org.apache.hadoop.net.StandardSocketFactory

hadoop.rpc.socket.factory.class.ClientProtocol

SocketFactory to use to connect to a DFS. If null or empty, use hadoop.rpc.socket.class.default. This socket factory is also used by DFSClient to create sockets to DataNodes.

hadoop.socks.server

Address (host:port) of the SOCKS server to be used by the SocksSocketFactory.

net.topology.node.switch.mapping.impl

The default implementation of the DNSToSwitchMapping. It invokes a script specified in net.topology.script.file.name to resolve node names. If the value for net.topology.script.file.name is not set, the default value of DEFAULT_RACK is returned for all node names. 机架感知实现类

org.apache.hadoop.net.ScriptBasedMapping

net.topology.impl

The default implementation of NetworkTopology which is classic three layer one.

org.apache.hadoop.net.NetworkTopology

net.topology.script.file.name

The script name that should be invoked to resolve DNS names to NetworkTopology names. Example: the script would take host.foo.bar as an argument, and return /rack1 as the output. 配合ScriptBasedMapping使用。脚本文件。此脚本文件，输入是ip地址，输出是机架路径。

net.topology.script.number.args

The max number of args that the script configured with net.topology.script.file.name should be run with. Each arg is an IP address. 机架感知脚本文件的参数最大数量。脚本每次运行被传递的参数，每个参数是一个ip地址

100

net.topology.table.file.name

The file name for a topology file, which is used when the net.topology.node.switch.mapping.impl property is set to org.apache.hadoop.net.TableMapping. The file format is a two column text file, with columns separated by whitespace. The first column is a DNS or IP address and the second column specifies the rack where the address maps. If no entry corresponding to a host in the cluster is found, then /default-rack is assumed. 在net.topology.script.file.name被设置为 org.apache.hadoop.net.TableMapping时，可以使用此配置。文件格式是一个有两个列的文本文件，使用空白字符分隔。第一列是DNS或IP地址，第二列是机架路径。如无指定，使用默认机架（/default-rack）

file.stream-buffer-size

The size of buffer to stream files. The size of this buffer should probably be a multiple of hardware page size (4096 on Intel x86), and it determines how much data is buffered during read and write operations.

4096

file.bytes-per-checksum

The number of bytes per checksum. Must not be larger than file.stream-buffer-size

512

file.client-write-packet-size

Packet size for clients to write

65536

file.blocksize

Block size

67108864

file.replication

Replication factor

1

s3.stream-buffer-size

The size of buffer to stream files. The size of this buffer should probably be a multiple of hardware page size (4096 on Intel x86), and it determines how much data is buffered during read and write operations.

4096

s3.bytes-per-checksum

The number of bytes per checksum. Must not be larger than s3.stream-buffer-size

512

s3.client-write-packet-size

Packet size for clients to write

65536

s3.blocksize

Block size

67108864

s3.replication

Replication factor

3

s3native.stream-buffer-size

The size of buffer to stream files. The size of this buffer should probably be a multiple of hardware page size (4096 on Intel x86), and it determines how much data is buffered during read and write operations.

4096

s3native.bytes-per-checksum

The number of bytes per checksum. Must not be larger than s3native.stream-buffer-size

512

s3native.client-write-packet-size

Packet size for clients to write

65536

s3native.blocksize

Block size

67108864

s3native.replication

Replication factor

3

ftp.stream-buffer-size

The size of buffer to stream files. The size of this buffer should probably be a multiple of hardware page size (4096 on Intel x86), and it determines how much data is buffered during read and write operations.

4096

ftp.bytes-per-checksum

The number of bytes per checksum. Must not be larger than ftp.stream-buffer-size

512

ftp.client-write-packet-size

Packet size for clients to write

65536

ftp.blocksize

Block size

67108864

ftp.replication

Replication factor

3

tfile.io.chunk.size

Value chunk size in bytes. Default to 1MB. Values of the length less than the chunk size is guaranteed to have known value length in read time (See also TFile.Reader.Scanner.Entry.isValueLengthKnown()).

1048576

tfile.fs.output.buffer.size

Buffer size used for FSDataOutputStream in bytes.

262144

tfile.fs.input.buffer.size

Buffer size used for FSDataInputStream in bytes.

262144

hadoop.http.authentication.type

Defines authentication used for Oozie HTTP endpoint. Supported values are: simple | kerberos | #AUTHENTICATION_HANDLER_CLASSNAME#

simple

hadoop.http.authentication.token.validity

Indicates how long (in seconds) an authentication token is valid before it has to be renewed.

36000

hadoop.http.authentication.signature.secret.file

The signature secret for signing the authentication tokens. The same secret should be used for JT/NN/DN/TT configurations.

${user.home}/hadoop-http-auth-signature-secret

hadoop.http.authentication.cookie.domain

The domain to use for the HTTP cookie that stores the authentication token. In order to authentiation to work correctly across all Hadoop nodes web-consoles the domain must be correctly set. IMPORTANT: when using IP addresses, browsers ignore cookies with domain settings. For this setting to work properly all nodes in the cluster must be configured to generate URLs with hostname.domain names on it.

hadoop.http.authentication.simple.anonymous.allowed

Indicates if anonymous requests are allowed when using ‘simple’ authentication.

true

hadoop.http.authentication.kerberos.principal

Indicates the Kerberos principal to be used for HTTP endpoint. The principal MUST start with ‘HTTP/’ as per Kerberos HTTP SPNEGO specification.

HTTP/_HOST@LOCALHOST

hadoop.http.authentication.kerberos.keytab

Location of the keytab file with the credentials for the principal. Referring to the same keytab file Oozie uses for its Kerberos credentials for Hadoop.

${user.home}/hadoop.keytab

dfs.ha.fencing.methods

List of fencing methods to use for service fencing. May contain builtin methods (eg shell and sshfence) or user-defined method. HDFS的HA功能的防脑裂方法。可以是内建的方法(例如shell和sshfence)或者用户定义的方法。建议使用sshfence(hadoop:9922)，括号内的是用户名和端口，注意，这需要NN的2台机器之间能够免密码登陆

dfs.ha.fencing.ssh.connect-timeout

SSH connection timeout, in milliseconds, to use with the builtin sshfence fencer.

30000

dfs.ha.fencing.ssh.private-key-files

The SSH private key files to use with the builtin sshfence fencer. 使用sshfence时，SSH的私钥文件。 使用了sshfence，这个必须指定

hadoop.http.staticuser.user

The user name to filter as, on static web filters while rendering content. An example use is the HDFS web UI (user to be used for browsing files).

dr.who

ha.zookeeper.quorum

A list of ZooKeeper server addresses, separated by commas, that are to be used by the ZKFailoverController in automatic failover. Ha功能，需要一组zk地址，用逗号分隔。被ZKFailoverController使用于自动失效备援failover。

ha.zookeeper.session-timeout.ms

The session timeout to use when the ZKFC connects to ZooKeeper. Setting this value to a lower value implies that server crashes will be detected more quickly, but risks triggering failover too aggressively in the case of a transient error or network blip. ZK连接超时。ZKFC连接ZK时用。设置一个小值可以更快的探测到服务器崩溃（crash),但也会更频繁的触发失效备援，在传输错误或者网络不畅时。建议****10s-30s

5000

ha.zookeeper.parent-znode

The ZooKeeper znode under which the ZK failover controller stores its information. Note that the nameservice ID is automatically appended to this znode, so it is not normally necessary to configure this, even in a federated environment. ZK失效备援功能，需要在ZK上创建节点，这里是根节点的名称。ZKFC会在这下面工作。注意，NameService ID会 被写到此节点下，所以即便是开启federation功能，也仅需要指定一个值。

/hadoop-ha

ha.zookeeper.acl

A comma-separated list of ZooKeeper ACLs to apply to the znodes used by automatic failover. These ACLs are specified in the same format as used by the ZooKeeper CLI. If the ACL itself contains secrets, you may instead specify a path to a file, prefixed with the ‘@’ symbol, and the value of this configuration will be loaded from within.

world:anyone:rwcda

ha.zookeeper.auth

A comma-separated list of ZooKeeper authentications to add when connecting to ZooKeeper. These are specified in the same format as used by the “addauth” command in the ZK CLI. It is important that the authentications specified here are sufficient to access znodes with the ACL specified in ha.zookeeper.acl. If the auths contain secrets, you may instead specify a path to a file, prefixed with the ‘@’ symbol, and the value of this configuration will be loaded from within.

hadoop.ssl.keystores.factory.class

The keystores factory to use for retrieving certificates.

org.apache.hadoop.security.ssl.FileBasedKeyStoresFactory

hadoop.ssl.require.client.cert

Whether client certificates are required

false

hadoop.ssl.hostname.verifier

The hostname verifier to provide for HttpsURLConnections. Valid values are: DEFAULT, STRICT, STRICT_I6, DEFAULT_AND_LOCALHOST and ALLOW_ALL

DEFAULT

hadoop.ssl.server.conf

Resource file from which ssl server keystore information will be extracted. This file is looked up in the classpath, typically it should be in Hadoop conf/ directory.

ssl-server.xml

hadoop.ssl.client.conf

Resource file from which ssl client keystore information will be extracted This file is looked up in the classpath, typically it should be in Hadoop conf/ directory.

ssl-client.xml

hadoop.ssl.enabled

Deprecated. Use dfs.http.policy and yarn.http.policy instead.

false

hadoop.jetty.logs.serve.aliases

Enable/Disable aliases serving from jetty

true

fs.permissions.umask-mode

The umask used when creating files and directories. Can be in octal or in symbolic. Examples are: “022” (octal for u=rwx,g=r-x,o=r-x in symbolic), or “u=rwx,g=rwx,o=” (symbolic for 007 in octal).

022

ha.health-monitor.connect-retry-interval.ms

How often to retry connecting to the service.

1000

ha.health-monitor.check-interval.ms

How often to check the service.

1000

ha.health-monitor.sleep-after-disconnect.ms

How long to sleep after an unexpected RPC error.

1000

ha.health-monitor.rpc-timeout.ms

Timeout for the actual monitorHealth() calls.

45000

ha.failover-controller.new-active.rpc-timeout.ms

Timeout that the FC waits for the new active to become active FC等待新的NN变成active状态的超时时间。

60000

ha.failover-controller.graceful-fence.rpc-timeout.ms

Timeout that the FC waits for the old active to go to standby FC等待旧的active变成standby的超时时间。

5000

ha.failover-controller.graceful-fence.connection.retries

FC connection retries for graceful fencing

1

ha.failover-controller.cli-check.rpc-timeout.ms

Timeout that the CLI (manual) FC waits for monitorHealth, getServiceState

20000

ipc.client.fallback-to-simple-auth-allowed

When a client is configured to attempt a secure connection, but attempts to connect to an insecure server, that server may instruct the client to switch to SASL SIMPLE (unsecure) authentication. This setting controls whether or not the client will accept this instruction from the server. When false (the default), the client will not allow the fallback to SIMPLE authentication, and will abort the connection.

false

fs.client.resolve.remote.symlinks

Whether to resolve symlinks when accessing a remote Hadoop filesystem. Setting this to false causes an exception to be thrown upon encountering a symlink. This setting does not apply to local filesystems, which automatically resolve local symlinks.

true

nfs3.server.port

Specify the port number used by Hadoop NFS.

2049

nfs3.mountd.port

Specify the port number used by Hadoop mount daemon.

4242

hadoop.user.group.static.mapping.overrides

Static mapping of user to groups. This will override the groups if available in the system for the specified user. In otherwords, groups look-up will not happen for these users, instead groups mapped in this configuration will be used. Mapping should be in this format. user1=group1,group2;user2=;user3=group2; Default, “dr.who=;” will consider “dr.who” as user without groups.

dr.who=;

rpc.metrics.quantile.enable

Setting this property to true and rpc.metrics.percentiles.intervals to a comma-separated list of the granularity in seconds, the 50/75/90/95/99th percentile latency for rpc queue/processing time in milliseconds are added to rpc metrics.

false

rpc.metrics.percentiles.intervals

fs.default.name

hdfs://hadoopmaster:9000

定义HadoopMaster的URI和端口

fs.checkpoint.dir

/opt/data/hadoop1/hdfs/namesecondary1

定义hadoop的name备份的路径，官方文档说是读取这个，写入dfs.name.dir

fs.checkpoint.period

1800

定义name备份的备份间隔时间，秒为单位，只对snn生效，默认一小时

fs.checkpoint.size

33554432

以日志大小间隔做备份间隔，只对snn生效，默认64M

io.compression.codecs

org.apache.hadoop.io.compress.DefaultCodec, com.hadoop.compression.lzo.LzoCodec, com.hadoop.compression.lzo.LzopCodec, org.apache.hadoop.io.compress.GzipCodec, org.apache.hadoop.io.compress.BZip2Codec (排版调整，实际配置不要回车)

Hadoop所使用的编解码器，gzip和bzip2为自带，lzo需安装hadoopgpl或者kevinweil，逗号分隔，snappy也需要单独安装

io.compression.codec.lzo.class

com.hadoop.compression.lzo.LzoCodec

LZO所使用的压缩编码器

topology.script.file.name

/hadoop/bin/RackAware.py

机架感知脚本位置

topology.script.number.args

1000

机架感知脚本管理的主机数，IP地址

fs.trash.interval

10800

HDFS垃圾箱设置，可以恢复误删除，分钟数，0为禁用，添加该项无需重启hadoop

hadoop.http.filter.initializers

org.apache.hadoop.security. AuthenticationFilterInitializer (排版调整，实际配置不要回车)

需要jobtracker,tasktracker namenode,datanode等http访问端口用户验证使用，需配置所有节点

hadoop.http.authentication.type

simple | kerberos | #AUTHENTICATION_HANDLER_CLASSNAME#

验证方式，默认为简单，也可自己定义class,需配置所有节点

hadoop.http.authentication. token.validity (排版调整，实际配置不要回车)

36000

验证令牌的有效时间,需配置所有节点

hadoop.http.authentication. signature.secret (排版调整，实际配置不要回车)

默认可不写参数

默认不写在hadoop启动时自动生成私密签名,需配置所有节点

hadoop.http.authentication.cookie.domain

domian.tld

http验证所使用的cookie的域名，IP地址访问则该项无效，必须给所有节点都配置域名才可以。

hadoop.http.authentication. simple.anonymous.allowed (排版调整，实际配置不要回车)

true | false

简单验证专用，默认允许匿名访问，true

hadoop.http.authentication. kerberos.principal (排版调整，实际配置不要回车)

HTTP/localhost@$LOCALHOST

Kerberos验证专用，参加认证的实体机必须使用HTTP作为K的Name

hadoop.http.authentication. kerberos.keytab (排版调整，实际配置不要回车)

/home/xianglei/hadoop.keytab

Kerberos验证专用，密钥文件存放位置

hadoop.security.authorization

true|false

Hadoop服务层级验证安全验证，需配合hadoop-policy.xml使用，配置好以后用dfsadmin,mradmin -refreshServiceAcl刷新生效

io.file.buffer.size

131072

用作序列化文件处理时读写buffer的大小

hadoop.security.authentication

simple | kerberos

hadoop本身的权限验证，非http访问，simple或者kerberos

hadoop.logfile.size

1000000000

设置日志文件大小，超过则滚动新日志

hadoop.logfile.count

20

最大日志数

io.bytes.per.checksum

1024

每校验码所校验的字节数，不要大于io.file.buffer.size

io.skip.checksum.errors

true | false

处理序列化文件时跳过校验码错误，不抛异常。默认false

io.serializations

org.apache.hadoop.io.serializer.WritableSerialization

序列化的编解码器

io.seqfile.compress.blocksize

1024000

块压缩的序列化文件的最小块大小，字节

webinterface.private.actions

true | false

设为true，则JT和NN的tracker网页会出现杀任务删文件等操作连接，默认是false