network.asciidoc 11 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256
  1. [[modules-network]]
  2. === Networking
  3. Each {es} node has two different network interfaces. Clients send requests to
  4. {es}'s REST APIs using its <<http-settings,HTTP interface>>, but nodes
  5. communicate with other nodes using the <<transport-settings,transport
  6. interface>>. The transport interface is also used for communication with
  7. <<remote-clusters,remote clusters>>.
  8. You can configure both of these interfaces at the same time using the
  9. `network.*` settings. If you have a more complicated network, you might need to
  10. configure the interfaces independently using the `http.*` and `transport.*`
  11. settings. Where possible, use the `network.*` settings that apply to both
  12. interfaces to simplify your configuration and reduce duplication.
  13. By default {es} binds only to `localhost` which means it cannot be accessed
  14. remotely. This configuration is sufficient for a local development cluster made
  15. of one or more nodes all running on the same host. To form a cluster across
  16. multiple hosts, or which is accessible to remote clients, you must adjust some
  17. <<common-network-settings,network settings>> such as `network.host`.
  18. [WARNING]
  19. .Be careful with the network configuration!
  20. =============================
  21. Never expose an unprotected node to the public internet. If you do, you are
  22. permitting anyone in the world to download, modify, or delete any of the data
  23. in your cluster.
  24. =============================
  25. Configuring {es} to bind to a non-local address will <<dev-vs-prod,convert some
  26. warnings into fatal exceptions>>. If a node refuses to start after configuring
  27. its network settings then you must address the logged exceptions before
  28. proceeding.
  29. [[common-network-settings]]
  30. ==== Commonly used network settings
  31. Most users will need to configure only the following network settings.
  32. `network.host`::
  33. (<<static-cluster-setting,Static>>, string)
  34. Sets the address of this node for both HTTP and transport traffic. The node
  35. will bind to this address and will also use it as its publish address. Accepts
  36. an IP address, a hostname, or a <<network-interface-values,special value>>.
  37. +
  38. Defaults to `_local_`.
  39. `http.port`::
  40. (<<static-cluster-setting,Static>>, integer)
  41. The port to bind for HTTP client communication. Accepts a single value or a
  42. range. If a range is specified, the node will bind to the first available port
  43. in the range.
  44. +
  45. Defaults to `9200-9300`.
  46. `transport.port`::
  47. (<<static-cluster-setting,Static>>, integer)
  48. The port to bind for communication between nodes. Accepts a single value or a
  49. range. If a range is specified, the node will bind to the first available port
  50. in the range. Set this setting to a single port, not a range, on every
  51. master-eligible node.
  52. +
  53. Defaults to `9300-9400`.
  54. [[network-interface-values]]
  55. ==== Special values for network addresses
  56. You can configure {es} to automatically determine its addresses by using the
  57. following special values. Use these values when configuring
  58. `network.host`, `network.bind_host`, `network.publish_host`, and the
  59. corresponding settings for the HTTP and transport interfaces.
  60. `_local_`::
  61. Any loopback addresses on the system, for example `127.0.0.1`.
  62. `_site_`::
  63. Any site-local addresses on the system, for example `192.168.0.1`.
  64. `_global_`::
  65. Any globally-scoped addresses on the system, for example `8.8.8.8`.
  66. `_[networkInterface]_`::
  67. Use the addresses of the network interface called `[networkInterface]`. For
  68. example if you wish to use the addresses of an interface called `en0` then
  69. set `network.host: _en0_`.
  70. `0.0.0.0`::
  71. The addresses of all available network interfaces.
  72. NOTE: In some systems these special values resolve to multiple addresses. If
  73. so, {es} will select one of them as its publish address and may change its
  74. selection on each node restart. Ensure your node is accessible at every possible
  75. address.
  76. NOTE: Any values containing a `:` (e.g. an IPv6 address or some of the
  77. <<network-interface-values,special values>>) must be quoted because `:` is a
  78. special character in YAML.
  79. [[network-interface-values-ipv4-vs-ipv6]]
  80. ===== IPv4 vs IPv6
  81. These special values yield both IPv4 and IPv6 addresses by default, but you can
  82. also add an `:ipv4` or `:ipv6` suffix to limit them to just IPv4 or IPv6
  83. addresses respectively. For example, `network.host: "_en0:ipv4_"` would set this
  84. node's addresses to the IPv4 addresses of interface `en0`.
  85. [TIP]
  86. .Discovery in the Cloud
  87. ================================
  88. More special settings are available when running in the Cloud with either the
  89. {plugins}/discovery-ec2.html[EC2 discovery plugin] or the
  90. {plugins}/discovery-gce-network-host.html#discovery-gce-network-host[Google Compute Engine discovery plugin]
  91. installed.
  92. ================================
  93. [[modules-network-binding-publishing]]
  94. ==== Binding and publishing
  95. {es} uses network addresses for two distinct purposes known as binding and
  96. publishing. Most nodes will use the same address for everything, but more
  97. complicated setups may need to configure different addresses for different
  98. purposes.
  99. When an application such as {es} wishes to receive network communications, it
  100. must indicate to the operating system the address or addresses whose traffic it
  101. should receive. This is known as _binding_ to those addresses. {es} can bind to
  102. more than one address if needed, but most nodes only bind to a single address.
  103. {es} can only bind to an address if it is running on a host that has a network
  104. interface with that address. If necessary, you can configure the transport and
  105. HTTP interfaces to bind to different addresses.
  106. Each {es} node has an address at which clients and other nodes can contact it,
  107. known as its _publish address_. Each node has one publish address for its HTTP
  108. interface and one for its transport interface. These two addresses can be
  109. anything, and don't need to be addresses of the network interfaces on the host.
  110. The only requirements are that each node must be:
  111. * Accessible at its transport publish address by all other
  112. nodes in its cluster, and by any remote clusters that will discover it using
  113. <<sniff-mode>>.
  114. * Accessible at its HTTP publish address by all clients
  115. that will discover it using sniffing.
  116. ===== Using a single address
  117. The most common configuration is for {es} to bind to a single address at which
  118. it is accessible to clients and other nodes. In this configuration you should
  119. just set `network.host` to that address. You should not separately set any bind
  120. or publish addresses, nor should you separately configure the addresses for the
  121. HTTP or transport interfaces.
  122. ===== Using multiple addresses
  123. Use the <<advanced-network-settings, advanced network settings>> if you wish to
  124. bind {es} to multiple addresses, or to publish a different address from the
  125. addresses to which you are binding. Set `network.bind_host` to the bind
  126. addresses, and `network.publish_host` to the address at which this node is
  127. exposed. In complex configurations, you can configure these addresses
  128. differently for the HTTP and transport interfaces.
  129. [[advanced-network-settings]]
  130. ==== Advanced network settings
  131. These advanced settings let you bind to multiple addresses, or to use different
  132. addresses for binding and publishing. They are not required in most cases and
  133. you should not use them if you can use the <<common-network-settings,commonly
  134. used settings>> instead.
  135. `network.bind_host`::
  136. (<<static-cluster-setting,Static>>, string)
  137. The network address(es) to which the node should bind in order to listen for
  138. incoming connections. Accepts a list of IP addresses, hostnames, and
  139. <<network-interface-values,special values>>. Defaults to the address given by
  140. `network.host`. Use this setting only if binding to multiple addresses or using
  141. different addresses for publishing and binding.
  142. `network.publish_host`::
  143. (<<static-cluster-setting,Static>>, string)
  144. The network address that clients and other nodes can use to contact this node.
  145. Accepts an IP address, a hostname, or a <<network-interface-values,special
  146. value>>. Defaults to the address given by `network.host`. Use this setting only
  147. if binding to multiple addresses or using different addresses for publishing
  148. and binding.
  149. NOTE: You can specify a list of addresses for `network.host` and
  150. `network.publish_host`. You can also specify one or more hostnames or
  151. <<network-interface-values,special values>> that resolve to multiple addresses.
  152. If you do this then {es} chooses one of the addresses for its publish address.
  153. This choice uses heuristics based on IPv4/IPv6 stack preference and
  154. reachability and may change when the node restarts. Ensure
  155. each node is accessible at all possible publish addresses.
  156. [[tcp-settings]]
  157. ===== Advanced TCP settings
  158. Use the following settings to control the low-level parameters of the TCP
  159. connections used by the HTTP and transport interfaces.
  160. `network.tcp.keep_alive`::
  161. (<<static-cluster-setting,Static>>, boolean)
  162. Configures the `SO_KEEPALIVE` option for network sockets, which determines
  163. whether each connection sends TCP keepalive probes. Defaults to `true`.
  164. `network.tcp.keep_idle`::
  165. (<<static-cluster-setting,Static>>, integer)
  166. Configures the `TCP_KEEPIDLE` option for network sockets, which determines the
  167. time in seconds that a connection must be idle before starting to send TCP
  168. keepalive probes. Defaults to `-1`, which means to use the system default. This
  169. value cannot exceed `300` seconds. Only applicable on Linux and macOS.
  170. `network.tcp.keep_interval`::
  171. (<<static-cluster-setting,Static>>, integer)
  172. Configures the `TCP_KEEPINTVL` option for network sockets, which determines the
  173. time in seconds between sending TCP keepalive probes. Defaults to `-1`, which
  174. means to use the system default. This value cannot exceed `300` seconds. Only
  175. applicable on Linux and macOS.
  176. `network.tcp.keep_count`::
  177. (<<static-cluster-setting,Static>>, integer)
  178. Configures the `TCP_KEEPCNT` option for network sockets, which determines the
  179. number of unacknowledged TCP keepalive probes that may be sent on a connection
  180. before it is dropped. Defaults to `-1`, which means to use the system default.
  181. Only applicable on Linux and macOS.
  182. `network.tcp.no_delay`::
  183. (<<static-cluster-setting,Static>>, boolean)
  184. Configures the `TCP_NODELAY` option on network sockets, which determines
  185. whether {wikipedia}/Nagle%27s_algorithm[TCP no delay] is enabled. Defaults to
  186. `true`.
  187. `network.tcp.reuse_address`::
  188. (<<static-cluster-setting,Static>>, boolean)
  189. Configures the `SO_REUSEADDR` option for network sockets, which determines
  190. whether the address can be reused or not. Defaults to `false` on Windows and
  191. `true` otherwise.
  192. `network.tcp.send_buffer_size`::
  193. (<<static-cluster-setting,Static>>, <<byte-units,byte value>>)
  194. Configures the size of the TCP send buffer for network sockets. Defaults to
  195. `-1` which means to use the system default.
  196. `network.tcp.receive_buffer_size`::
  197. (<<static-cluster-setting,Static>>, <<byte-units,byte value>>)
  198. Configures the size of the TCP receive buffer. Defaults to `-1` which means to
  199. use the system default.
  200. include::http.asciidoc[]
  201. include::transport.asciidoc[]
  202. include::network/tracers.asciidoc[]
  203. include::network/threading.asciidoc[]