sin/samba-mirror
1
0
Fork 0
mirror of https://github.com/samba-team/samba.git synced 2025-01-10 01:18:15 +03:00
Code

first step in health monitoring of cluster nodes. When not healthy they will be marked disabled

Browse Source 
(This used to be ctdb commit d3dbd9fc4db21632075b56fc52cf95435c63374a)
This commit is contained in:
Andrew Tridgell 2007-06-05 17:43:19 +10:00
parent ee747b5bd6
commit ac55bc4166
9 changed files with 22 additions and 11 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

8
ctdb/common/ctdb.c
View File

@ -222,14 +222,16 @@ uint32_t ctdb_get_vnn(struct ctdb_context *ctdb)
}
/*
return the number of connected nodes
return the number of enabled nodes
*/
uint32_t ctdb_get_num_connected_nodes(struct ctdb_context *ctdb)
uint32_t ctdb_get_num_enabled_nodes(struct ctdb_context *ctdb)
{
int i;
uint32_t count=0;
for (i=0;i<ctdb->vnn_map->size;i++) {
if (ctdb->nodes[ctdb->vnn_map->map[i]]->flags & NODE_FLAGS_CONNECTED) {
struct ctdb_node *node = ctdb->nodes[ctdb->vnn_map->map[i]];
if ((node->flags & NODE_FLAGS_CONNECTED) &&
!(node->flags & NODE_FLAGS_DISABLED)) {
count++;
}
}

2
ctdb/common/ctdb_client.c
View File

@ -1364,7 +1364,7 @@ struct ctdb_db_context *ctdb_attach(struct ctdb_context *ctdb, const char *name)
ctdb_db->db_id = *(uint32_t *)data.dptr;
talloc_free(data.dptr);
ret = ctdb_ctrl_getdbpath(ctdb, timeval_current_ofs(1, 0), CTDB_CURRENT_NODE, ctdb_db->db_id, ctdb_db, &ctdb_db->db_path);
ret = ctdb_ctrl_getdbpath(ctdb, timeval_current_ofs(2, 0), CTDB_CURRENT_NODE, ctdb_db->db_id, ctdb_db, &ctdb_db->db_path);
if (ret != 0) {
DEBUG(0,("Failed to get dbpath for database '%s'\n", name));
talloc_free(ctdb_db);

2
ctdb/common/ctdb_recoverd.c
View File

@ -697,6 +697,8 @@ again:
"MonitorFrequency", &ctdb->tunable.monitor_frequency);
ctdb_ctrl_get_tunable(ctdb, CONTROL_TIMEOUT(), CTDB_CURRENT_NODE,
"ElectionTimeout", &ctdb->tunable.election_timeout);
ctdb_ctrl_get_tunable(ctdb, CONTROL_TIMEOUT(), CTDB_CURRENT_NODE,
"TakeoverTimeout", &ctdb->tunable.takeover_timeout);
vnn = ctdb_ctrl_getvnn(ctdb, CONTROL_TIMEOUT(), CTDB_CURRENT_NODE);
if (vnn == (uint32_t)-1) {

2
ctdb/common/ctdb_traverse.c
View File

@ -372,7 +372,7 @@ int32_t ctdb_control_traverse_data(struct ctdb_context *ctdb, TDB_DATA data, TDB
if (key.dsize == 0 && data.dsize == 0) {
state->null_count++;
if (state->null_count != ctdb_get_num_connected_nodes(ctdb)) {
if (state->null_count != ctdb_get_num_enabled_nodes(ctdb)) {
return 0;
}
}

1
ctdb/common/ctdb_tunables.c
View File

@ -35,6 +35,7 @@ static const struct {
{ "RecoverTimeout", 5, offsetof(struct ctdb_tunable, recover_timeout) },
{ "MonitorFrequency", 1, offsetof(struct ctdb_tunable, monitor_frequency) },
{ "ElectionTimeout", 3, offsetof(struct ctdb_tunable, election_timeout) },
{ "TakeoverTimeout", 5, offsetof(struct ctdb_tunable, takeover_timeout) },
};
/*

6
ctdb/config/events.d/50.samba
View File

@ -21,8 +21,10 @@ case $cmd in
service smb stop > /dev/null 2>&1
service winbind stop > /dev/null 2>&1
# start Samba service
service smb start
# start Samba service. Start it reniced, as under very heavy load
# the number of smbd processes will mean that it leaves few cycles for
# anything else
nice service smb start
service winbind start
# wait for the Samba tcp ports to become available

4
ctdb/include/ctdb_private.h
View File

@ -50,6 +50,7 @@ struct ctdb_tunable {
uint32_t recover_timeout;
uint32_t monitor_frequency;
uint32_t election_timeout;
uint32_t takeover_timeout;
};
/*
@ -109,6 +110,7 @@ struct ctdb_node {
void *private_data; /* private to transport */
uint32_t vnn;
#define NODE_FLAGS_CONNECTED 0x00000001
#define NODE_FLAGS_DISABLED 0x00000002
uint32_t flags;
/* used by the dead node monitoring */
@ -905,7 +907,7 @@ int32_t ctdb_control_thaw(struct ctdb_context *ctdb);
int ctdb_start_recoverd(struct ctdb_context *ctdb);
uint32_t ctdb_get_num_connected_nodes(struct ctdb_context *ctdb);
uint32_t ctdb_get_num_enabled_nodes(struct ctdb_context *ctdb);
int ctdb_start_monitoring(struct ctdb_context *ctdb);
void ctdb_send_keepalive(struct ctdb_context *ctdb, uint32_t destnode);

6
ctdb/takeover/ctdb_takeover.c
View File

@ -27,7 +27,7 @@
#include "../include/ctdb_private.h"
#define TAKEOVER_TIMEOUT() timeval_current_ofs(5,0)
#define TAKEOVER_TIMEOUT() timeval_current_ofs(ctdb->tunable.takeover_timeout,0)
#define CTDB_ARP_INTERVAL 1
#define CTDB_ARP_REPEAT 3
@ -403,7 +403,8 @@ int ctdb_takeover_run(struct ctdb_context *ctdb, struct ctdb_node_map *nodemap)
/* work out which node will look after each public IP */
for (i=0;i<nodemap->num;i++) {
if (nodemap->nodes[i].flags & NODE_FLAGS_CONNECTED) {
if ((nodemap->nodes[i].flags & NODE_FLAGS_CONNECTED) &&
!(nodemap->nodes[i].flags & NODE_FLAGS_DISABLED)) {
ctdb->nodes[i]->takeover_vnn = nodemap->nodes[i].vnn;
} else {
/* assign this dead nodes IP to the next higher node */
@ -411,6 +412,7 @@ int ctdb_takeover_run(struct ctdb_context *ctdb, struct ctdb_node_map *nodemap)
j != i;
j=(j+1)%nodemap->num) {
if ((nodemap->nodes[j].flags & NODE_FLAGS_CONNECTED) &&
!(nodemap->nodes[j].flags & NODE_FLAGS_DISABLED) &&
ctdb_same_subnet(ctdb->nodes[j]->public_address,
ctdb->nodes[i]->public_address,
ctdb->nodes[j]->public_netmask_bits)) {

2
ctdb/tools/ctdb_control.c
View File

@ -383,7 +383,7 @@ static int control_shutdown(struct ctdb_context *ctdb, int argc, const char **ar
{
int ret;
ret = ctdb_ctrl_shutdown(ctdb, timeval_current_ofs(1, 0), options.vnn);
ret = ctdb_ctrl_shutdown(ctdb, TIMELIMIT(), options.vnn);
if (ret != 0) {
printf("Unable to shutdown node %u\n", options.vnn);
return ret;