uint32_t monitor_interval;
uint32_t tickle_update_interval;
uint32_t script_timeout;
- uint32_t script_ban_count; /* ban after this many consec timeouts*/
- uint32_t script_unhealthy_on_timeout; /* don't ban on timeout; set node unhealthy */
+ uint32_t script_timeout_count; /* allow dodgy scripts to hang this many times in a row before we mark the node unhealthy */
+ uint32_t script_unhealthy_on_timeout; /* obsolete */
uint32_t recovery_grace_period;
uint32_t recovery_ban_period;
uint32_t database_hash_size;
uint32_t rerecovery_timeout;
uint32_t enable_bans;
uint32_t deterministic_public_ips;
- uint32_t disable_when_unhealthy;
uint32_t reclock_ping_period;
uint32_t no_ip_failback;
uint32_t verbose_memory_names;
uint32_t vacuum_min_interval;
uint32_t vacuum_max_interval;
uint32_t max_queue_depth_drop_msg;
+ uint32_t use_status_events_for_monitoring;
+ uint32_t allow_unhealthy_db_read;
};
/*
struct ctdb_client_notify_list *notify;
};
+struct ctdb_iface;
/* state associated with a public ip address */
struct ctdb_vnn {
struct ctdb_vnn *prev, *next;
- const char *iface;
+ struct ctdb_iface *iface;
+ const char **ifaces;
ctdb_sock_addr public_address;
uint8_t public_netmask_bits;
across the nodes. it needs to know which public ip's can be handled
by each node.
*/
- struct ctdb_all_public_ips *public_ips;
+ struct ctdb_all_public_ips *known_public_ips;
+ struct ctdb_all_public_ips *available_public_ips;
/* used by the recovery dameon to track when a node should be banned */
struct ctdb_banning_state *ban_state;
};
const char *name;
const char *db_directory;
const char *db_directory_persistent;
+ const char *db_directory_state;
+ struct tdb_wrap *db_persistent_health;
+ uint32_t db_persistent_startup_generation;
+ uint64_t db_persistent_check_errors;
+ uint64_t max_persistent_check_errors;
const char *transport;
char *recovery_lock_file;
int recovery_lock_fd;
struct ctdb_node **nodes; /* array of nodes in the cluster - indexed by vnn */
struct ctdb_vnn *vnn; /* list of public ip addresses and interfaces */
struct ctdb_vnn *single_ip_vnn; /* a structure for the single ip */
+ struct ctdb_iface *ifaces; /* list of local interfaces */
char *err_msg;
const struct ctdb_methods *methods; /* transport methods */
const struct ctdb_upcalls *upcalls; /* transport upcalls */
uint32_t recovery_master;
struct ctdb_call_state *pending_calls;
struct ctdb_client_ip *client_ip_list;
- bool do_setsched;
bool do_checkpublicip;
- void *saved_scheduler_param;
- struct _trbt_tree_t *server_ids;
+ struct trbt_tree *server_ids;
const char *event_script_dir;
const char *notification_script;
const char *default_public_interface;
struct ctdb_log_state *log;
int start_as_disabled;
int start_as_stopped;
+ bool valgrinding;
uint32_t event_script_timeouts; /* counting how many consecutive times an eventscript has timedout */
- TALLOC_CTX *eventscripts_ctx; /* a context to hold data for the RUN_EVENTSCRIPTS control */
uint32_t *recd_ping_count;
TALLOC_CTX *release_ips_ctx; /* a context used to automatically drop all IPs if we fail to recover the node */
- TALLOC_CTX *script_monitor_ctx; /* a context where we store results while running the monitor event */
- TALLOC_CTX *last_monitor_ctx;
- TALLOC_CTX *event_script_ctx; /* non-monitoring events */
+
+ TALLOC_CTX *event_script_ctx;
+
+ struct ctdb_event_script_state *current_monitor;
+ struct ctdb_scripts_wire *last_status[CTDB_EVENT_MAX];
+
TALLOC_CTX *banning_ctx;
+
+ /* mapping from pid to ctdb_client * */
+ struct ctdb_client_pid_list *client_pids;
+
+ /* used in the recovery daemon to remember the ip allocation */
+ struct trbt_tree *ip_tree;
};
struct ctdb_db_context {
struct tdb_wrap *ltdb;
struct ctdb_registered_call *calls; /* list of registered calls */
uint32_t seqnum;
- struct timed_event *te;
+ struct timed_event *seqnum_update;
struct ctdb_traverse_local_handle *traverse;
bool transaction_active;
struct ctdb_vacuum_handle *vacuum_handle;
+ char *unhealthy_reason;
};
CTDB_CONTROL_TAKEOVER_IP = 89,
CTDB_CONTROL_GET_PUBLIC_IPS = 90,
CTDB_CONTROL_GET_NODEMAP = 91,
- CTDB_CONTROL_EVENT_SCRIPT_INIT = 92,
- CTDB_CONTROL_EVENT_SCRIPT_START = 93,
- CTDB_CONTROL_EVENT_SCRIPT_STOP = 94,
- CTDB_CONTROL_EVENT_SCRIPT_FINISHED = 95,
CTDB_CONTROL_GET_EVENT_SCRIPT_STATUS = 96,
CTDB_CONTROL_TRAVERSE_KILL = 97,
CTDB_CONTROL_RECD_RECLOCK_LATENCY = 98,
CTDB_CONTROL_SET_NATGWSTATE = 103,
CTDB_CONTROL_SET_LMASTERROLE = 104,
CTDB_CONTROL_SET_RECMASTERROLE = 105,
- CTDB_CONTROL_EVENT_SCRIPT_DISABLED = 106,
CTDB_CONTROL_ENABLE_SCRIPT = 107,
CTDB_CONTROL_DISABLE_SCRIPT = 108,
CTDB_CONTROL_SET_BAN_STATE = 109,
CTDB_CONTROL_REGISTER_NOTIFY = 114,
CTDB_CONTROL_DEREGISTER_NOTIFY = 115,
CTDB_CONTROL_TRANS2_ACTIVE = 116,
+ CTDB_CONTROL_GET_LOG = 117,
+ CTDB_CONTROL_CLEAR_LOG = 118,
+ CTDB_CONTROL_TRANS3_COMMIT = 119,
+ CTDB_CONTROL_GET_DB_SEQNUM = 120,
+ CTDB_CONTROL_DB_SET_HEALTHY = 121,
+ CTDB_CONTROL_DB_GET_HEALTH = 122,
+ CTDB_CONTROL_GET_PUBLIC_IP_INFO = 123,
+ CTDB_CONTROL_GET_IFACES = 124,
+ CTDB_CONTROL_SET_IFACE_LINK_STATE = 125,
};
/*
uint64_t srvid;
uint32_t client_id;
#define CTDB_CTRL_FLAG_NOREPLY 1
+#define CTDB_CTRL_FLAG_OPCODE_SPECIFIC 0xFFFF0000
uint32_t flags;
uint32_t datalen;
uint8_t data[1];
CTDB_TRANS2_COMMIT_SOMEFAIL=3 /* some nodes failed the commit, some allowed it */
};
-
/* internal prototypes */
void ctdb_set_error(struct ctdb_context *ctdb, const char *fmt, ...) PRINTF_ATTRIBUTE(2,3);
void ctdb_fatal(struct ctdb_context *ctdb, const char *msg);
void ctdb_node_dead(struct ctdb_node *node);
void ctdb_node_connected(struct ctdb_node *node);
bool ctdb_blocking_freeze(struct ctdb_context *ctdb);
-void ctdb_set_scheduler(struct ctdb_context *ctdb);
-void ctdb_restore_scheduler(struct ctdb_context *ctdb);
+void ctdb_high_priority(struct ctdb_context *ctdb);
int32_t ctdb_control_takeover_ip(struct ctdb_context *ctdb,
struct ctdb_req_control *c,
TDB_DATA indata,
int32_t ctdb_control_get_public_ipsv4(struct ctdb_context *ctdb, struct ctdb_req_control *c, TDB_DATA *outdata);
int32_t ctdb_control_get_public_ips(struct ctdb_context *ctdb, struct ctdb_req_control *c, TDB_DATA *outdata);
int ctdb_ctrl_get_public_ips(struct ctdb_context *ctdb,
- struct timeval timeout, uint32_t destnode,
- TALLOC_CTX *mem_ctx, struct ctdb_all_public_ips **ips);
+ struct timeval timeout,
+ uint32_t destnode,
+ TALLOC_CTX *mem_ctx,
+ struct ctdb_all_public_ips **ips);
+#define CTDB_PUBLIC_IP_FLAGS_ONLY_AVAILABLE 0x00010000
+int ctdb_ctrl_get_public_ips_flags(struct ctdb_context *ctdb,
+ struct timeval timeout, uint32_t destnode,
+ TALLOC_CTX *mem_ctx,
+ uint32_t flags,
+ struct ctdb_all_public_ips **ips);
int ctdb_ctrl_get_public_ipsv4(struct ctdb_context *ctdb,
struct timeval timeout, uint32_t destnode,
TALLOC_CTX *mem_ctx, struct ctdb_all_public_ips **ips);
+#ifdef IFNAMSIZ
+#define CTDB_IFACE_SIZE IFNAMSIZ
+#else
+#define CTDB_IFACE_SIZE 16
+#endif
+
+struct ctdb_control_iface_info {
+ char name[CTDB_IFACE_SIZE+2];
+ uint16_t link_state;
+ uint32_t references;
+};
+
+struct ctdb_control_public_ip_info {
+ struct ctdb_public_ip ip;
+ uint32_t active_idx;
+ uint32_t num;
+ struct ctdb_control_iface_info ifaces[1];
+};
+
+struct ctdb_control_get_ifaces {
+ uint32_t num;
+ struct ctdb_control_iface_info ifaces[1];
+};
+
+int32_t ctdb_control_get_public_ip_info(struct ctdb_context *ctdb,
+ struct ctdb_req_control *c,
+ TDB_DATA indata,
+ TDB_DATA *outdata);
+int32_t ctdb_control_get_ifaces(struct ctdb_context *ctdb,
+ struct ctdb_req_control *c,
+ TDB_DATA *outdata);
+int32_t ctdb_control_set_iface_link(struct ctdb_context *ctdb,
+ struct ctdb_req_control *c,
+ TDB_DATA indata);
+int ctdb_ctrl_get_public_ip_info(struct ctdb_context *ctdb,
+ struct timeval timeout, uint32_t destnode,
+ TALLOC_CTX *mem_ctx,
+ const ctdb_sock_addr *addr,
+ struct ctdb_control_public_ip_info **info);
+int ctdb_ctrl_get_ifaces(struct ctdb_context *ctdb,
+ struct timeval timeout, uint32_t destnode,
+ TALLOC_CTX *mem_ctx,
+ struct ctdb_control_get_ifaces **ifaces);
+int ctdb_ctrl_set_iface_link(struct ctdb_context *ctdb,
+ struct timeval timeout, uint32_t destnode,
+ TALLOC_CTX *mem_ctx,
+ const struct ctdb_control_iface_info *info);
/* from takeover/system.c */
uint32_t uint16_checksum(uint16_t *data, size_t n);
uint32_t seq, uint32_t ack, int rst);
int ctdb_set_public_addresses(struct ctdb_context *ctdb, const char *alist);
+int ctdb_set_single_public_ip(struct ctdb_context *ctdb,
+ const char *iface,
+ const char *ip);
int ctdb_set_event_script(struct ctdb_context *ctdb, const char *script);
int ctdb_set_event_script_dir(struct ctdb_context *ctdb, const char *script_dir);
int ctdb_set_notification_script(struct ctdb_context *ctdb, const char *script);
int32_t ctdb_control_set_tcp_tickle_list(struct ctdb_context *ctdb, TDB_DATA indata);
void ctdb_takeover_client_destructor_hook(struct ctdb_client *client);
-int ctdb_event_script(struct ctdb_context *ctdb, const char *fmt, ...) PRINTF_ATTRIBUTE(2,3);
+int ctdb_event_script(struct ctdb_context *ctdb, enum ctdb_eventscript_call call);
+int ctdb_event_script_args(struct ctdb_context *ctdb, enum ctdb_eventscript_call call,
+ const char *fmt, ...) PRINTF_ATTRIBUTE(3,4);
int ctdb_event_script_callback(struct ctdb_context *ctdb,
- struct timeval timeout,
TALLOC_CTX *mem_ctx,
void (*callback)(struct ctdb_context *, int, void *),
void *private_data,
- const char *fmt, ...) PRINTF_ATTRIBUTE(6,7);
+ bool from_user,
+ enum ctdb_eventscript_call call,
+ const char *fmt, ...) PRINTF_ATTRIBUTE(7,8);
void ctdb_release_all_ips(struct ctdb_context *ctdb);
void set_nonblocking(int fd);
int32_t ctdb_control_uptime(struct ctdb_context *ctdb,
TDB_DATA *outdata);
-int ctdb_attach_persistent(struct ctdb_context *ctdb);
+int ctdb_attach_databases(struct ctdb_context *ctdb);
int32_t ctdb_control_persistent_store(struct ctdb_context *ctdb,
struct ctdb_req_control *c,
struct ctdb_req_control *c,
TDB_DATA recdata, bool *async_reply);
+int32_t ctdb_control_trans3_commit(struct ctdb_context *ctdb,
+ struct ctdb_req_control *c,
+ TDB_DATA recdata, bool *async_reply);
+
int32_t ctdb_control_transaction_start(struct ctdb_context *ctdb, uint32_t id);
int32_t ctdb_control_transaction_commit(struct ctdb_context *ctdb, uint32_t id);
int32_t ctdb_control_transaction_cancel(struct ctdb_context *ctdb);
int32_t ctdb_control_wipe_database(struct ctdb_context *ctdb, TDB_DATA indata);
+int32_t ctdb_control_db_set_healthy(struct ctdb_context *ctdb, TDB_DATA indata);
+int32_t ctdb_control_db_get_health(struct ctdb_context *ctdb,
+ TDB_DATA indata,
+ TDB_DATA *outdata);
int ctdb_vacuum(struct ctdb_context *ctdb, int argc, const char **argv);
void ctdb_unblock_signal(int signum);
int32_t ctdb_monitoring_mode(struct ctdb_context *ctdb);
int ctdb_set_child_logging(struct ctdb_context *ctdb);
-
+void ctdb_lockdown_memory(struct ctdb_context *ctdb);
typedef void (*client_async_callback)(struct ctdb_context *ctdb, uint32_t node_pnn, int32_t res, TDB_DATA outdata, void *callback_data);
extern int script_log_level;
-int ctdb_ctrl_event_script_init(struct ctdb_context *ctdb);
-int ctdb_ctrl_event_script_start(struct ctdb_context *ctdb, const char *name);
-int ctdb_ctrl_event_script_stop(struct ctdb_context *ctdb, int32_t res);
-int ctdb_ctrl_event_script_finished(struct ctdb_context *ctdb);
-int ctdb_ctrl_event_script_disabled(struct ctdb_context *ctdb, const char *name);
-
-int32_t ctdb_control_event_script_init(struct ctdb_context *ctdb);
-int32_t ctdb_control_event_script_start(struct ctdb_context *ctdb, TDB_DATA indata);
-int32_t ctdb_control_event_script_stop(struct ctdb_context *ctdb, TDB_DATA indata);
-int32_t ctdb_control_event_script_disabled(struct ctdb_context *ctdb, TDB_DATA indata);
-int32_t ctdb_control_event_script_finished(struct ctdb_context *ctdb);
-
-
-int32_t ctdb_control_get_event_script_status(struct ctdb_context *ctdb, TDB_DATA *outdata);
+int32_t ctdb_control_get_event_script_status(struct ctdb_context *ctdb,
+ uint32_t call_type,
+ TDB_DATA *outdata);
int ctdb_log_event_script_output(struct ctdb_context *ctdb, char *str, uint16_t len);
int ctdb_ctrl_report_recd_lock_latency(struct ctdb_context *ctdb, struct timeval timeout, double latency);
int32_t ctdb_control_set_ban_state(struct ctdb_context *ctdb, TDB_DATA indata);
int32_t ctdb_control_get_ban_state(struct ctdb_context *ctdb, TDB_DATA *outdata);
int32_t ctdb_control_set_db_priority(struct ctdb_context *ctdb, TDB_DATA indata);
+void ctdb_ban_self(struct ctdb_context *ctdb);
int32_t ctdb_control_register_notify(struct ctdb_context *ctdb, uint32_t client_id, TDB_DATA indata);
int start_syslog_daemon(struct ctdb_context *ctdb);
+/* Where to send the log messages back to */
+struct ctdb_get_log_addr {
+ uint32_t pnn;
+ uint64_t srvid;
+ int32_t level;
+};
+
+extern int log_ringbuf_size;
+
+int32_t ctdb_control_get_log(struct ctdb_context *ctdb, TDB_DATA addr);
+int32_t ctdb_control_clear_log(struct ctdb_context *ctdb);
+
+
+struct ctdb_log_state *ctdb_fork_with_logging(TALLOC_CTX *mem_ctx,
+ struct ctdb_context *ctdb,
+ void (*logfn)(const char *, uint16_t, void *),
+ void *logfn_private, pid_t *pid);
+
+int32_t ctdb_control_process_exists(struct ctdb_context *ctdb, pid_t pid);
+struct ctdb_client *ctdb_find_client_by_pid(struct ctdb_context *ctdb, pid_t pid);
+
+int32_t ctdb_control_get_db_seqnum(struct ctdb_context *ctdb,
+ TDB_DATA indata,
+ TDB_DATA *outdata);
+
+int ctdb_load_persistent_health(struct ctdb_context *ctdb,
+ struct ctdb_db_context *ctdb_db);
+int ctdb_update_persistent_health(struct ctdb_context *ctdb,
+ struct ctdb_db_context *ctdb_db,
+ const char *reason,/* NULL means healthy */
+ int num_healthy_nodes);
+int ctdb_recheck_persistent_health(struct ctdb_context *ctdb);
+
+void ctdb_run_notification_script(struct ctdb_context *ctdb, const char *event);
+
+void ctdb_fault_setup(void);
+
+int verify_remote_ip_allocation(struct ctdb_context *ctdb,
+ struct ctdb_all_public_ips *ips);
+int update_ip_assignment_tree(struct ctdb_context *ctdb,
+ struct ctdb_public_ip *ip);
+
#endif
git.samba.org / sahlberg / ctdb.git / blobdiff