ctdb/server/ctdb_keepalive.c

   1 /*
   2    monitoring links to all other nodes to detect dead nodes
   3
   4
   5    Copyright (C) Ronnie Sahlberg 2007
   6
   7    This program is free software; you can redistribute it and/or modify
   8    it under the terms of the GNU General Public License as published by
   9    the Free Software Foundation; either version 3 of the License, or
  10    (at your option) any later version.
  11
  12    This program is distributed in the hope that it will be useful,
  13    but WITHOUT ANY WARRANTY; without even the implied warranty of
  14    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  15    GNU General Public License for more details.
  16
  17    You should have received a copy of the GNU General Public License
  18    along with this program; if not, see <http://www.gnu.org/licenses/>.
  19 */
  20
  21 #include "includes.h"
  22 #include "system/filesys.h"
  23 #include "system/wait.h"
  24 #include "../include/ctdb_private.h"
  25
  26
  27 /*
  28   see if any nodes are dead
  29  */
  30 static void ctdb_check_for_dead_nodes(struct event_context *ev, struct timed_event *te,
  31                                       struct timeval t, void *private_data)
  32 {
  33         struct ctdb_context *ctdb = talloc_get_type(private_data, struct ctdb_context);
  34         int i;
  35
  36         /* send a keepalive to all other nodes, unless */
  37         for (i=0;i<ctdb->num_nodes;i++) {
  38                 struct ctdb_node *node = ctdb->nodes[i];
  39
  40                 if (node->flags & NODE_FLAGS_DELETED) {
  41                         continue;
  42                 }
  43
  44                 if (node->pnn == ctdb->pnn) {
  45                         continue;
  46                 }
  47
  48                 if (node->flags & NODE_FLAGS_DISCONNECTED) {
  49                         /* it might have come alive again */
  50                         if (node->rx_cnt != 0) {
  51                                 ctdb_node_connected(node);
  52                         }
  53                         continue;
  54                 }
  55
  56
  57                 if (node->rx_cnt == 0) {
  58                         node->dead_count++;
  59                 } else {
  60                         node->dead_count = 0;
  61                 }
  62
  63                 node->rx_cnt = 0;
  64
  65                 if (node->dead_count >= ctdb->tunable.keepalive_limit) {
  66                         DEBUG(DEBUG_NOTICE,("dead count reached for node %u\n", node->pnn));
  67                         ctdb_node_dead(node);
  68                         ctdb_send_keepalive(ctdb, node->pnn);
  69                         /* maybe tell the transport layer to kill the
  70                            sockets as well?
  71                         */
  72                         continue;
  73                 }
  74
  75                 DEBUG(DEBUG_DEBUG,("sending keepalive to %u\n", node->pnn));
  76                 ctdb_send_keepalive(ctdb, node->pnn);
  77
  78                 node->tx_cnt = 0;
  79         }
  80
  81         event_add_timed(ctdb->ev, ctdb->keepalive_ctx,
  82                         timeval_current_ofs(ctdb->tunable.keepalive_interval, 0),
  83                         ctdb_check_for_dead_nodes, ctdb);
  84 }
  85
  86
  87 void ctdb_start_keepalive(struct ctdb_context *ctdb)
  88 {
  89         struct timed_event *te;
  90
  91         ctdb->keepalive_ctx = talloc_new(ctdb);
  92         CTDB_NO_MEMORY_FATAL(ctdb, ctdb->keepalive_ctx);
  93
  94         te = event_add_timed(ctdb->ev, ctdb->keepalive_ctx,
  95                              timeval_current_ofs(ctdb->tunable.keepalive_interval, 0),
  96                              ctdb_check_for_dead_nodes, ctdb);
  97         CTDB_NO_MEMORY_FATAL(ctdb, te);
  98
  99         DEBUG(DEBUG_NOTICE,("Keepalive monitoring has been started\n"));
 100 }
 101
 102 void ctdb_stop_keepalive(struct ctdb_context *ctdb)
 103 {
 104         talloc_free(ctdb->keepalive_ctx);
 105         ctdb->keepalive_ctx = NULL;
 106 }
 107