grpc 设置超时（c++）

这里用的测试代码是grpc源码中的HelloWord代码。

超时一般都是在客户端调用接口的时候要设置超时。分两种方式同步和异步：

同步设置超时如下：

 gpr_timespec timespec;
  timespec.tv_sec = 2;//设置阻塞时间为2秒
  timespec.tv_nsec = 0;
  timespec.clock_type = GPR_TIMESPAN;

ClientContext context;
  context.set_deadline(timespec); 

HelloRequest request;
 request.set_heads(user);
 HelloReply reply;
  
  // The actual RPC.
  Status status = stub_->SayHello(&context, request, &reply);

  // Act upon its status.
  if (status.ok()) {
   std::string message = reply.message();
   std::cout << message << std::endl;
  }
  else {
   std::cout << status.error_code() << ": " << status.error_message()
    << std::endl;
  }

如果status.ok()返回false,可以根据status.error_code()返回的错误码判断失败的原因。

错误码为  StatusCode 类型。

enum StatusCode {
  /// Not an error; returned on success.
  OK = 0,

  /// The operation was cancelled (typically by the caller).
  CANCELLED = 1,

  /// Unknown error. An example of where this error may be returned is if a
  /// Status value received from another address space belongs to an error-space
  /// that is not known in this address space. Also errors raised by APIs that
  /// do not return enough error information may be converted to this error.
  UNKNOWN = 2,

  /// Client specified an invalid argument. Note that this differs from
  /// FAILED_PRECONDITION. INVALID_ARGUMENT indicates arguments that are
  /// problematic regardless of the state of the system (e.g., a malformed file
  /// name).
  INVALID_ARGUMENT = 3,

  /// Deadline expired before operation could complete. For operations that
  /// change the state of the system, this error may be returned even if the
  /// operation has completed successfully. For example, a successful response
  /// from a server could have been delayed long enough for the deadline to
  /// expire.
  DEADLINE_EXCEEDED = 4,

  /// Some requested entity (e.g., file or directory) was not found.
  NOT_FOUND = 5,

  /// Some entity that we attempted to create (e.g., file or directory) already
  /// exists.
  ALREADY_EXISTS = 6,

  /// The caller does not have permission to execute the specified operation.
  /// PERMISSION_DENIED must not be used for rejections caused by exhausting
  /// some resource (use RESOURCE_EXHAUSTED instead for those errors).
  /// PERMISSION_DENIED must not be used if the caller can not be identified
  /// (use UNAUTHENTICATED instead for those errors).
  PERMISSION_DENIED = 7,

  /// The request does not have valid authentication credentials for the
  /// operation.
  UNAUTHENTICATED = 16,

  /// Some resource has been exhausted, perhaps a per-user quota, or perhaps the
  /// entire file system is out of space.
  RESOURCE_EXHAUSTED = 8,

  /// Operation was rejected because the system is not in a state required for
  /// the operation's execution. For example, directory to be deleted may be
  /// non-empty, an rmdir operation is applied to a non-directory, etc.
  ///
  /// A litmus test that may help a service implementor in deciding
  /// between FAILED_PRECONDITION, ABORTED, and UNAVAILABLE:
  ///  (a) Use UNAVAILABLE if the client can retry just the failing call.
  ///  (b) Use ABORTED if the client should retry at a higher-level
  ///      (e.g., restarting a read-modify-write sequence).
  ///  (c) Use FAILED_PRECONDITION if the client should not retry until
  ///      the system state has been explicitly fixed. E.g., if an "rmdir"
  ///      fails because the directory is non-empty, FAILED_PRECONDITION
  ///      should be returned since the client should not retry unless
  ///      they have first fixed up the directory by deleting files from it.
  ///  (d) Use FAILED_PRECONDITION if the client performs conditional
  ///      REST Get/Update/Delete on a resource and the resource on the
  ///      server does not match the condition. E.g., conflicting
  ///      read-modify-write on the same resource.
  FAILED_PRECONDITION = 9,

  /// The operation was aborted, typically due to a concurrency issue like
  /// sequencer check failures, transaction aborts, etc.
  ///
  /// See litmus test above for deciding between FAILED_PRECONDITION, ABORTED,
  /// and UNAVAILABLE.
  ABORTED = 10,

  /// Operation was attempted past the valid range. E.g., seeking or reading
  /// past end of file.
  ///
  /// Unlike INVALID_ARGUMENT, this error indicates a problem that may be fixed
  /// if the system state changes. For example, a 32-bit file system will
  /// generate INVALID_ARGUMENT if asked to read at an offset that is not in the
  /// range [0,2^32-1], but it will generate OUT_OF_RANGE if asked to read from
  /// an offset past the current file size.
  ///
  /// There is a fair bit of overlap between FAILED_PRECONDITION and
  /// OUT_OF_RANGE. We recommend using OUT_OF_RANGE (the more specific error)
  /// when it applies so that callers who are iterating through a space can
  /// easily look for an OUT_OF_RANGE error to detect when they are done.
  OUT_OF_RANGE = 11,

  /// Operation is not implemented or not supported/enabled in this service.
  UNIMPLEMENTED = 12,

  /// Internal errors. Means some invariants expected by underlying System has
  /// been broken. If you see one of these errors, Something is very broken.
  INTERNAL = 13,

  /// The service is currently unavailable. This is a most likely a transient
  /// condition and may be corrected by retrying with a backoff.
  ///
  /// \warning Although data MIGHT not have been transmitted when this
  /// status occurs, there is NOT A GUARANTEE that the server has not seen
  /// anything. So in general it is unsafe to retry on this status code
  /// if the call is non-idempotent.
  ///
  /// See litmus test above for deciding between FAILED_PRECONDITION, ABORTED,
  /// and UNAVAILABLE.
  UNAVAILABLE = 14,

  /// Unrecoverable data loss or corruption.
  DATA_LOSS = 15,

  /// Force users to include a default branch:
  DO_NOT_USE = -1
};

异步设置超时如下：

 HelloRequest request;
 request.set_heads(user);

 HelloReply reply;

 ClientContext context;

 CompletionQueue cq;

Status status;

std::unique_ptr > rpc(
        stub_->AsyncSayHello(&context, request, &cq));

 rpc->Finish(&reply, &status, (void*)1);
  void* got_tag;
  bool ok = false;

gpr_timespec time;
 time.tv_sec = 2;//设置2秒超时
 time.tv_nsec = 0;
 time.clock_type = GPR_TIMESPAN;
 CompletionQueue::NextStatus nextStatus =cq.AsyncNext(&got_tag, &ok, time);

if(nextStatus==GOT_EVENT)

{

GPR_ASSERT(got_tag == (void*)1);

 GPR_ASSERT(ok);

}

else if(nextStatus==TIMEOUT)

{

//超时

}

可以根据NextStatus返回值判断AsyncNext调用是否成功。

  enum NextStatus {
    SHUTDOWN,   ///< The completion queue has been shutdown.
    GOT_EVENT,  ///< Got a new event; \a tag will be filled in with its
                ///< associated value; \a ok indicating its success.
    TIMEOUT     ///< deadline was reached.
  };