Bug #25487 deleting ndb_cluster_connection object takes long time

  
  aim is to:
  a) if set_connect_timeout called, timeout connect attempt (for retry on
  next call) after timeout period
  b) preserve existing blocking behaviour otherwise (for, e.g. mgmapi)
  
  Related to customer issue with long time deleting ndb_cluster_connection
  object. believe we're hanging on the connect(2) call until timeout (when
  we then realise we should exit the thread).


ndb/include/mgmapi/mgmapi.h:
  add ndb_mgm_set_connect_timeout
ndb/include/util/SocketClient.hpp:
  add timeout (seconds) for max time to wait for connection
ndb/src/common/transporter/Transporter.cpp:
  set limit on amount of time we'll wait for tcp connect
ndb/src/common/util/SocketClient.cpp:
  only try to connect for a maximum of timeout time
ndb/src/mgmapi/mgmapi.cpp:
  add ndb_mgm_set_connect_timeout

ndb/include/mgmapi/mgmapi.h

@@ -540,6 +540,16 @@ extern "C" {
    */
   const char *ndb_mgm_get_connectstring(NdbMgmHandle handle, char *buf, int buf_sz);
 
+  /**
+   * Sets the number of seconds to wait for connect(2) during ndb_mgm_connect
+   * Default is no timeout
+   *
+   * @param handle  NdbMgmHandle
+   * @param seconds number of seconds
+   * @return non-zero on success
+   */
+  int ndb_mgm_set_connect_timeout(NdbMgmHandle handle, unsigned int seconds);
+
   /**
    * Connects to a management server. Connectstring is set by
    * ndb_mgm_set_connectstring().

ndb/include/util/SocketClient.hpp

@@ -23,6 +23,7 @@ class SocketClient
 {
   NDB_SOCKET_TYPE m_sockfd;
   struct sockaddr_in m_servaddr;
+  unsigned int m_connect_timeout_sec;
   unsigned short m_port;
   char *m_server_name;
   SocketAuthenticator *m_auth;
@@ -34,6 +35,9 @@ public:
     m_port = port;
     m_servaddr.sin_port = htons(m_port);
   };
+  void set_connect_timeout(unsigned int s) {
+    m_connect_timeout_sec= s;
+  }
   unsigned short get_port() { return m_port; };
   char *get_server_name() { return m_server_name; };
   int bind(const char* toaddress, unsigned short toport);

ndb/src/common/transporter/Transporter.cpp

@@ -79,9 +79,13 @@ Transporter::Transporter(TransporterRegistry &t_reg,
   if (isServer)
     m_socket_client= 0;
   else
+  {
     m_socket_client= new SocketClient(remoteHostName, s_port,
 				      new SocketAuthSimple("ndbd",
 							   "ndbd passwd"));
+
+    m_socket_client->set_connect_timeout(m_timeOutMillis/1000);
+  }
   DBUG_VOID_RETURN;
 }
 
@@ -140,9 +144,9 @@ Transporter::connect_client() {
     }
     sockfd= m_socket_client->connect();
   }
-  
+
   return connect_client(sockfd);
-}  
+}
 
 bool
 Transporter::connect_client(NDB_SOCKET_TYPE sockfd) {

ndb/src/common/util/SocketClient.cpp

@@ -26,6 +26,7 @@ SocketClient::SocketClient(const char *server_name, unsigned short port, SocketA
   m_port= port;
   m_server_name= server_name ? strdup(server_name) : 0;
   m_sockfd= NDB_INVALID_SOCKET;
+  m_connect_timeout_sec= 0;
 }
 
 SocketClient::~SocketClient()
@@ -58,7 +59,7 @@ SocketClient::init()
   if (m_sockfd == NDB_INVALID_SOCKET) {
     return false;
   }
-  
+
   DBUG_PRINT("info",("NDB_SOCKET: %d", m_sockfd));
 
   return true;
@@ -104,6 +105,13 @@ SocketClient::bind(const char* bindaddress, unsigned short localport)
 NDB_SOCKET_TYPE
 SocketClient::connect(const char *toaddress, unsigned short toport)
 {
+  fd_set rset, wset;
+  struct timeval tval;
+  int r;
+  bool use_timeout;
+  socklen_t len;
+  int flags;
+
   if (m_sockfd == NDB_INVALID_SOCKET)
   {
     if (!init()) {
@@ -127,14 +135,58 @@ SocketClient::connect(const char *toaddress, unsigned short toport)
     if (Ndb_getInAddr(&m_servaddr.sin_addr, m_server_name))
       return NDB_INVALID_SOCKET;
   }
-  
-  const int r = ::connect(m_sockfd, (struct sockaddr*) &m_servaddr, sizeof(m_servaddr));
-  if (r == -1) {
+
+  flags= fcntl(m_sockfd, F_GETFL, 0);
+  fcntl(m_sockfd, F_SETFL, flags | O_NONBLOCK);
+
+  r= ::connect(m_sockfd, (struct sockaddr*) &m_servaddr, sizeof(m_servaddr));
+
+  if (r == 0)
+    goto done; // connected immediately.
+
+  if (r < 0 && (errno != EINPROGRESS)) {
+    NDB_CLOSE_SOCKET(m_sockfd);
+    m_sockfd= NDB_INVALID_SOCKET;
+    return NDB_INVALID_SOCKET;
+  }
+
+  FD_ZERO(&rset);
+  FD_SET(m_sockfd, &rset);
+  wset= rset;
+  tval.tv_sec= m_connect_timeout_sec;
+  tval.tv_usec= 0;
+  use_timeout= m_connect_timeout_sec;
+
+  if ((r= select(m_sockfd+1, &rset, &wset, NULL,
+                 use_timeout? &tval : NULL)) == 0)
+  {
     NDB_CLOSE_SOCKET(m_sockfd);
     m_sockfd= NDB_INVALID_SOCKET;
     return NDB_INVALID_SOCKET;
   }
 
+  if (FD_ISSET(m_sockfd, &rset) || FD_ISSET(m_sockfd, &wset))
+  {
+    len= sizeof(r);
+    if (getsockopt(m_sockfd, SOL_SOCKET, SO_ERROR, &r, &len) < 0)
+    {
+      // Solaris got an error... different than others
+      NDB_CLOSE_SOCKET(m_sockfd);
+      m_sockfd= NDB_INVALID_SOCKET;
+      return NDB_INVALID_SOCKET;
+    }
+  }
+  else
+  {
+    // select error, probably m_sockfd not set.
+    NDB_CLOSE_SOCKET(m_sockfd);
+    m_sockfd= NDB_INVALID_SOCKET;
+    return NDB_INVALID_SOCKET;
+  }
+
+done:
+  fcntl(m_sockfd, F_SETFL, flags);
+
   if (m_auth) {
     if (!m_auth->client_authenticate(m_sockfd))
     {

ndb/src/mgmapi/mgmapi.cpp

@@ -93,6 +93,7 @@ struct ndb_mgm_handle {
   char last_error_desc[NDB_MGM_MAX_ERR_DESC_SIZE];
   int read_timeout;
   int write_timeout;
+  unsigned int connect_timeout;
 
   NDB_SOCKET_TYPE socket;
 
@@ -159,6 +160,7 @@ ndb_mgm_create_handle()
   h->socket          = NDB_INVALID_SOCKET;
   h->read_timeout    = 50000;
   h->write_timeout   = 100;
+  h->connect_timeout = 0;
   h->cfg_i           = -1;
   h->errstream       = stdout;
   h->m_name          = 0;
@@ -426,6 +428,16 @@ int ndb_mgm_is_connected(NdbMgmHandle handle)
   return handle->connected;
 }
 
+extern "C"
+int ndb_mgm_set_connect_timeout(NdbMgmHandle handle, unsigned int seconds)
+{
+  if(!handle)
+    return -1;
+
+  handle->connect_timeout= seconds;
+  return 0;
+}
+
 /**
  * Connect to a management server
  */
@@ -456,6 +468,7 @@ ndb_mgm_connect(NdbMgmHandle handle, int no_retries,
   Uint32 i;
   int binderror = 0;
   SocketClient s(0, 0);
+  s.set_connect_timeout(handle->connect_timeout);
   if (!s.init())
   {
     fprintf(handle->errstream,