balancer.go 11 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271
  1. /*
  2. *
  3. * Copyright 2017 gRPC authors.
  4. *
  5. * Licensed under the Apache License, Version 2.0 (the "License");
  6. * you may not use this file except in compliance with the License.
  7. * You may obtain a copy of the License at
  8. *
  9. * http://www.apache.org/licenses/LICENSE-2.0
  10. *
  11. * Unless required by applicable law or agreed to in writing, software
  12. * distributed under the License is distributed on an "AS IS" BASIS,
  13. * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
  14. * See the License for the specific language governing permissions and
  15. * limitations under the License.
  16. *
  17. */
  18. // Package balancer defines APIs for load balancing in gRPC.
  19. // All APIs in this package are experimental.
  20. package balancer
  21. import (
  22. "errors"
  23. "net"
  24. "strings"
  25. "golang.org/x/net/context"
  26. "google.golang.org/grpc/connectivity"
  27. "google.golang.org/grpc/credentials"
  28. "google.golang.org/grpc/resolver"
  29. )
  30. var (
  31. // m is a map from name to balancer builder.
  32. m = make(map[string]Builder)
  33. )
  34. // Register registers the balancer builder to the balancer map. b.Name
  35. // (lowercased) will be used as the name registered with this builder.
  36. //
  37. // NOTE: this function must only be called during initialization time (i.e. in
  38. // an init() function), and is not thread-safe. If multiple Balancers are
  39. // registered with the same name, the one registered last will take effect.
  40. func Register(b Builder) {
  41. m[strings.ToLower(b.Name())] = b
  42. }
  43. // Get returns the resolver builder registered with the given name.
  44. // Note that the compare is done in a case-insenstive fashion.
  45. // If no builder is register with the name, nil will be returned.
  46. func Get(name string) Builder {
  47. if b, ok := m[strings.ToLower(name)]; ok {
  48. return b
  49. }
  50. return nil
  51. }
  52. // SubConn represents a gRPC sub connection.
  53. // Each sub connection contains a list of addresses. gRPC will
  54. // try to connect to them (in sequence), and stop trying the
  55. // remainder once one connection is successful.
  56. //
  57. // The reconnect backoff will be applied on the list, not a single address.
  58. // For example, try_on_all_addresses -> backoff -> try_on_all_addresses.
  59. //
  60. // All SubConns start in IDLE, and will not try to connect. To trigger
  61. // the connecting, Balancers must call Connect.
  62. // When the connection encounters an error, it will reconnect immediately.
  63. // When the connection becomes IDLE, it will not reconnect unless Connect is
  64. // called.
  65. //
  66. // This interface is to be implemented by gRPC. Users should not need a
  67. // brand new implementation of this interface. For the situations like
  68. // testing, the new implementation should embed this interface. This allows
  69. // gRPC to add new methods to this interface.
  70. type SubConn interface {
  71. // UpdateAddresses updates the addresses used in this SubConn.
  72. // gRPC checks if currently-connected address is still in the new list.
  73. // If it's in the list, the connection will be kept.
  74. // If it's not in the list, the connection will gracefully closed, and
  75. // a new connection will be created.
  76. //
  77. // This will trigger a state transition for the SubConn.
  78. UpdateAddresses([]resolver.Address)
  79. // Connect starts the connecting for this SubConn.
  80. Connect()
  81. }
  82. // NewSubConnOptions contains options to create new SubConn.
  83. type NewSubConnOptions struct{}
  84. // ClientConn represents a gRPC ClientConn.
  85. //
  86. // This interface is to be implemented by gRPC. Users should not need a
  87. // brand new implementation of this interface. For the situations like
  88. // testing, the new implementation should embed this interface. This allows
  89. // gRPC to add new methods to this interface.
  90. type ClientConn interface {
  91. // NewSubConn is called by balancer to create a new SubConn.
  92. // It doesn't block and wait for the connections to be established.
  93. // Behaviors of the SubConn can be controlled by options.
  94. NewSubConn([]resolver.Address, NewSubConnOptions) (SubConn, error)
  95. // RemoveSubConn removes the SubConn from ClientConn.
  96. // The SubConn will be shutdown.
  97. RemoveSubConn(SubConn)
  98. // UpdateBalancerState is called by balancer to nofity gRPC that some internal
  99. // state in balancer has changed.
  100. //
  101. // gRPC will update the connectivity state of the ClientConn, and will call pick
  102. // on the new picker to pick new SubConn.
  103. UpdateBalancerState(s connectivity.State, p Picker)
  104. // ResolveNow is called by balancer to notify gRPC to do a name resolving.
  105. ResolveNow(resolver.ResolveNowOption)
  106. // Target returns the dial target for this ClientConn.
  107. Target() string
  108. }
  109. // BuildOptions contains additional information for Build.
  110. type BuildOptions struct {
  111. // DialCreds is the transport credential the Balancer implementation can
  112. // use to dial to a remote load balancer server. The Balancer implementations
  113. // can ignore this if it does not need to talk to another party securely.
  114. DialCreds credentials.TransportCredentials
  115. // Dialer is the custom dialer the Balancer implementation can use to dial
  116. // to a remote load balancer server. The Balancer implementations
  117. // can ignore this if it doesn't need to talk to remote balancer.
  118. Dialer func(context.Context, string) (net.Conn, error)
  119. // ChannelzParentID is the entity parent's channelz unique identification number.
  120. ChannelzParentID int64
  121. }
  122. // Builder creates a balancer.
  123. type Builder interface {
  124. // Build creates a new balancer with the ClientConn.
  125. Build(cc ClientConn, opts BuildOptions) Balancer
  126. // Name returns the name of balancers built by this builder.
  127. // It will be used to pick balancers (for example in service config).
  128. Name() string
  129. }
  130. // PickOptions contains addition information for the Pick operation.
  131. type PickOptions struct{}
  132. // DoneInfo contains additional information for done.
  133. type DoneInfo struct {
  134. // Err is the rpc error the RPC finished with. It could be nil.
  135. Err error
  136. // BytesSent indicates if any bytes have been sent to the server.
  137. BytesSent bool
  138. // BytesReceived indicates if any byte has been received from the server.
  139. BytesReceived bool
  140. }
  141. var (
  142. // ErrNoSubConnAvailable indicates no SubConn is available for pick().
  143. // gRPC will block the RPC until a new picker is available via UpdateBalancerState().
  144. ErrNoSubConnAvailable = errors.New("no SubConn is available")
  145. // ErrTransientFailure indicates all SubConns are in TransientFailure.
  146. // WaitForReady RPCs will block, non-WaitForReady RPCs will fail.
  147. ErrTransientFailure = errors.New("all SubConns are in TransientFailure")
  148. )
  149. // Picker is used by gRPC to pick a SubConn to send an RPC.
  150. // Balancer is expected to generate a new picker from its snapshot every time its
  151. // internal state has changed.
  152. //
  153. // The pickers used by gRPC can be updated by ClientConn.UpdateBalancerState().
  154. type Picker interface {
  155. // Pick returns the SubConn to be used to send the RPC.
  156. // The returned SubConn must be one returned by NewSubConn().
  157. //
  158. // This functions is expected to return:
  159. // - a SubConn that is known to be READY;
  160. // - ErrNoSubConnAvailable if no SubConn is available, but progress is being
  161. // made (for example, some SubConn is in CONNECTING mode);
  162. // - other errors if no active connecting is happening (for example, all SubConn
  163. // are in TRANSIENT_FAILURE mode).
  164. //
  165. // If a SubConn is returned:
  166. // - If it is READY, gRPC will send the RPC on it;
  167. // - If it is not ready, or becomes not ready after it's returned, gRPC will block
  168. // until UpdateBalancerState() is called and will call pick on the new picker.
  169. //
  170. // If the returned error is not nil:
  171. // - If the error is ErrNoSubConnAvailable, gRPC will block until UpdateBalancerState()
  172. // - If the error is ErrTransientFailure:
  173. // - If the RPC is wait-for-ready, gRPC will block until UpdateBalancerState()
  174. // is called to pick again;
  175. // - Otherwise, RPC will fail with unavailable error.
  176. // - Else (error is other non-nil error):
  177. // - The RPC will fail with unavailable error.
  178. //
  179. // The returned done() function will be called once the rpc has finished, with the
  180. // final status of that RPC.
  181. // done may be nil if balancer doesn't care about the RPC status.
  182. Pick(ctx context.Context, opts PickOptions) (conn SubConn, done func(DoneInfo), err error)
  183. }
  184. // Balancer takes input from gRPC, manages SubConns, and collects and aggregates
  185. // the connectivity states.
  186. //
  187. // It also generates and updates the Picker used by gRPC to pick SubConns for RPCs.
  188. //
  189. // HandleSubConnectionStateChange, HandleResolvedAddrs and Close are guaranteed
  190. // to be called synchronously from the same goroutine.
  191. // There's no guarantee on picker.Pick, it may be called anytime.
  192. type Balancer interface {
  193. // HandleSubConnStateChange is called by gRPC when the connectivity state
  194. // of sc has changed.
  195. // Balancer is expected to aggregate all the state of SubConn and report
  196. // that back to gRPC.
  197. // Balancer should also generate and update Pickers when its internal state has
  198. // been changed by the new state.
  199. HandleSubConnStateChange(sc SubConn, state connectivity.State)
  200. // HandleResolvedAddrs is called by gRPC to send updated resolved addresses to
  201. // balancers.
  202. // Balancer can create new SubConn or remove SubConn with the addresses.
  203. // An empty address slice and a non-nil error will be passed if the resolver returns
  204. // non-nil error to gRPC.
  205. HandleResolvedAddrs([]resolver.Address, error)
  206. // Close closes the balancer. The balancer is not required to call
  207. // ClientConn.RemoveSubConn for its existing SubConns.
  208. Close()
  209. }
  210. // ConnectivityStateEvaluator takes the connectivity states of multiple SubConns
  211. // and returns one aggregated connectivity state.
  212. //
  213. // It's not thread safe.
  214. type ConnectivityStateEvaluator struct {
  215. numReady uint64 // Number of addrConns in ready state.
  216. numConnecting uint64 // Number of addrConns in connecting state.
  217. numTransientFailure uint64 // Number of addrConns in transientFailure.
  218. }
  219. // RecordTransition records state change happening in subConn and based on that
  220. // it evaluates what aggregated state should be.
  221. //
  222. // - If at least one SubConn in Ready, the aggregated state is Ready;
  223. // - Else if at least one SubConn in Connecting, the aggregated state is Connecting;
  224. // - Else the aggregated state is TransientFailure.
  225. //
  226. // Idle and Shutdown are not considered.
  227. func (cse *ConnectivityStateEvaluator) RecordTransition(oldState, newState connectivity.State) connectivity.State {
  228. // Update counters.
  229. for idx, state := range []connectivity.State{oldState, newState} {
  230. updateVal := 2*uint64(idx) - 1 // -1 for oldState and +1 for new.
  231. switch state {
  232. case connectivity.Ready:
  233. cse.numReady += updateVal
  234. case connectivity.Connecting:
  235. cse.numConnecting += updateVal
  236. case connectivity.TransientFailure:
  237. cse.numTransientFailure += updateVal
  238. }
  239. }
  240. // Evaluate.
  241. if cse.numReady > 0 {
  242. return connectivity.Ready
  243. }
  244. if cse.numConnecting > 0 {
  245. return connectivity.Connecting
  246. }
  247. return connectivity.TransientFailure
  248. }