nodes into two separate files.
move the monitoring of keepalives for detecting connected/disconnected
remote nodes into ctdb_keepalive.c
server/ctdb_control.o server/ctdb_call.o server/ctdb_ltdb_server.o \
server/ctdb_traverse.o server/eventscript.o server/ctdb_takeover.o \
server/ctdb_serverids.o server/ctdb_persistent.o \
+ server/ctdb_keepalive.o \
$(CTDB_CLIENT_OBJ) $(CTDB_TCP_OBJ) @INFINIBAND_WRAPPER_OBJ@
TEST_BINS=bin/ctdb_bench bin/ctdb_fetch bin/ctdb_store bin/ctdb_randrec bin/ctdb_persistent bin/rb_test \
struct event_context *ev;
uint32_t recovery_mode;
TALLOC_CTX *tickle_update_context;
+ TALLOC_CTX *keepalive_ctx;
struct ctdb_tunable tunable;
enum ctdb_freeze_mode freeze_mode;
struct ctdb_freeze_handle *freeze_handle;
void ctdb_start_monitoring(struct ctdb_context *ctdb);
void ctdb_start_tcp_tickle_update(struct ctdb_context *ctdb);
void ctdb_send_keepalive(struct ctdb_context *ctdb, uint32_t destnode);
+void ctdb_start_keepalive(struct ctdb_context *ctdb);
+void ctdb_stop_keepalive(struct ctdb_context *ctdb);
void ctdb_daemon_cancel_controls(struct ctdb_context *ctdb, struct ctdb_node *node);
void ctdb_call_resend_all(struct ctdb_context *ctdb);
case CTDB_CONTROL_SHUTDOWN:
ctdb_stop_recoverd(ctdb);
+ ctdb_stop_keepalive(ctdb);
ctdb_stop_monitoring(ctdb);
ctdb_release_all_ips(ctdb);
ctdb->methods->shutdown(ctdb);
ctdb_register_message_handler(ctdb, ctdb, CTDB_SRVID_NODE_FLAGS_CHANGED,
flag_change_handler, NULL);
- /* start monitoring for dead nodes */
+ /* start monitoring for connected/disconnected nodes */
+ ctdb_start_keepalive(ctdb);
+
+ /* start monitoring for node health */
ctdb_start_monitoring(ctdb);
/* start periodic update of tcp tickle lists */
--- /dev/null
+/*
+ monitoring links to all other nodes to detect dead nodes
+
+
+ Copyright (C) Ronnie Sahlberg 2007
+
+ This program is free software; you can redistribute it and/or modify
+ it under the terms of the GNU General Public License as published by
+ the Free Software Foundation; either version 3 of the License, or
+ (at your option) any later version.
+
+ This program is distributed in the hope that it will be useful,
+ but WITHOUT ANY WARRANTY; without even the implied warranty of
+ MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ GNU General Public License for more details.
+
+ You should have received a copy of the GNU General Public License
+ along with this program; if not, see <http://www.gnu.org/licenses/>.
+*/
+
+#include "includes.h"
+#include "lib/events/events.h"
+#include "system/filesys.h"
+#include "system/wait.h"
+#include "../include/ctdb_private.h"
+
+
+/*
+ see if any nodes are dead
+ */
+static void ctdb_check_for_dead_nodes(struct event_context *ev, struct timed_event *te,
+ struct timeval t, void *private_data)
+{
+ struct ctdb_context *ctdb = talloc_get_type(private_data, struct ctdb_context);
+ int i;
+
+ /* send a keepalive to all other nodes, unless */
+ for (i=0;i<ctdb->num_nodes;i++) {
+ struct ctdb_node *node = ctdb->nodes[i];
+ if (node->pnn == ctdb->pnn) {
+ continue;
+ }
+
+ if (node->flags & NODE_FLAGS_DISCONNECTED) {
+ /* it might have come alive again */
+ if (node->rx_cnt != 0) {
+ ctdb_node_connected(node);
+ }
+ continue;
+ }
+
+
+ if (node->rx_cnt == 0) {
+ node->dead_count++;
+ } else {
+ node->dead_count = 0;
+ }
+
+ node->rx_cnt = 0;
+
+ if (node->dead_count >= ctdb->tunable.keepalive_limit) {
+ DEBUG(0,("dead count reached for node %u\n", node->pnn));
+ ctdb_node_dead(node);
+ ctdb_send_keepalive(ctdb, node->pnn);
+ /* maybe tell the transport layer to kill the
+ sockets as well?
+ */
+ continue;
+ }
+
+ if (node->tx_cnt == 0) {
+ DEBUG(5,("sending keepalive to %u\n", node->pnn));
+ ctdb_send_keepalive(ctdb, node->pnn);
+ }
+
+ node->tx_cnt = 0;
+ }
+
+ event_add_timed(ctdb->ev, ctdb->keepalive_ctx,
+ timeval_current_ofs(ctdb->tunable.keepalive_interval, 0),
+ ctdb_check_for_dead_nodes, ctdb);
+}
+
+
+void ctdb_start_keepalive(struct ctdb_context *ctdb)
+{
+ struct timed_event *te;
+
+ ctdb->keepalive_ctx = talloc_new(ctdb);
+ CTDB_NO_MEMORY_FATAL(ctdb, ctdb->keepalive_ctx);
+
+ te = event_add_timed(ctdb->ev, ctdb->keepalive_ctx,
+ timeval_current_ofs(ctdb->tunable.keepalive_interval, 0),
+ ctdb_check_for_dead_nodes, ctdb);
+ CTDB_NO_MEMORY_FATAL(ctdb, te);
+
+ DEBUG(0,("Keepalive monitoring has been started\n"));
+}
+
+void ctdb_stop_keepalive(struct ctdb_context *ctdb)
+{
+ talloc_free(ctdb->keepalive_ctx);
+ ctdb->keepalive_ctx = NULL;
+}
+
uint32_t next_interval;
};
-/*
- see if any nodes are dead
- */
-static void ctdb_check_for_dead_nodes(struct event_context *ev, struct timed_event *te,
- struct timeval t, void *private_data)
-{
- struct ctdb_context *ctdb = talloc_get_type(private_data, struct ctdb_context);
- int i;
-
- /* send a keepalive to all other nodes, unless */
- for (i=0;i<ctdb->num_nodes;i++) {
- struct ctdb_node *node = ctdb->nodes[i];
- if (node->pnn == ctdb->pnn) {
- continue;
- }
-
- if (node->flags & NODE_FLAGS_DISCONNECTED) {
- /* it might have come alive again */
- if (node->rx_cnt != 0) {
- ctdb_node_connected(node);
- }
- continue;
- }
-
-
- if (node->rx_cnt == 0) {
- node->dead_count++;
- } else {
- node->dead_count = 0;
- }
-
- node->rx_cnt = 0;
-
- if (node->dead_count >= ctdb->tunable.keepalive_limit) {
- DEBUG(0,("dead count reached for node %u\n", node->pnn));
- ctdb_node_dead(node);
- ctdb_send_keepalive(ctdb, node->pnn);
- /* maybe tell the transport layer to kill the
- sockets as well?
- */
- continue;
- }
-
- if (node->tx_cnt == 0) {
- DEBUG(5,("sending keepalive to %u\n", node->pnn));
- ctdb_send_keepalive(ctdb, node->pnn);
- }
-
- node->tx_cnt = 0;
- }
-
- event_add_timed(ctdb->ev, ctdb->monitor->monitor_context,
- timeval_current_ofs(ctdb->tunable.keepalive_interval, 0),
- ctdb_check_for_dead_nodes, ctdb);
-}
-
static void ctdb_check_health(struct event_context *ev, struct timed_event *te,
struct timeval t, void *private_data);
ctdb->monitor->monitor_context = talloc_new(ctdb->monitor);
CTDB_NO_MEMORY_FATAL(ctdb, ctdb->monitor->monitor_context);
- te = event_add_timed(ctdb->ev, ctdb->monitor->monitor_context,
- timeval_current_ofs(ctdb->tunable.keepalive_interval, 0),
- ctdb_check_for_dead_nodes, ctdb);
- CTDB_NO_MEMORY_FATAL(ctdb, te);
-
te = event_add_timed(ctdb->ev, ctdb->monitor->monitor_context,
timeval_current_ofs(1, 0),
ctdb_check_health, ctdb);