#include "system/filesys.h"
#include "system/locale.h"
#include <stdlib.h>
-#include "../include/ctdb_private.h"
+#include "include/ctdb_protocol.h"
+#include "include/ctdb_private.h"
#include "lib/util/dlinklist.h"
-/*
- allocate a packet for use in client<->daemon communication
- */
-struct ctdb_req_header *_ctdbd_allocate_pkt(struct ctdb_context *ctdb,
- TALLOC_CTX *mem_ctx,
- enum ctdb_operation operation,
- size_t length, size_t slength,
- const char *type)
-{
- int size;
- struct ctdb_req_header *hdr;
-
- length = MAX(length, slength);
- size = (length+(CTDB_DS_ALIGNMENT-1)) & ~(CTDB_DS_ALIGNMENT-1);
-
- hdr = (struct ctdb_req_header *)talloc_size(mem_ctx, size);
- if (hdr == NULL) {
- DEBUG(DEBUG_ERR,("Unable to allocate packet for operation %u of length %u\n",
- operation, (unsigned)length));
- return NULL;
- }
- talloc_set_name_const(hdr, type);
- memset(hdr, 0, slength);
- hdr->length = length;
- hdr->operation = operation;
- hdr->ctdb_magic = CTDB_MAGIC;
- hdr->ctdb_version = CTDB_VERSION;
- hdr->srcnode = ctdb->pnn;
- if (ctdb->vnn_map) {
- hdr->generation = ctdb->vnn_map->generation;
- }
-
- return hdr;
-}
-
-/*
- local version of ctdb_call
-*/
-int ctdb_call_local(struct ctdb_db_context *ctdb_db, struct ctdb_call *call,
- struct ctdb_ltdb_header *header, TALLOC_CTX *mem_ctx,
- TDB_DATA *data, uint32_t caller)
-{
- struct ctdb_call_info *c;
- struct ctdb_registered_call *fn;
- struct ctdb_context *ctdb = ctdb_db->ctdb;
-
- c = talloc(ctdb, struct ctdb_call_info);
- CTDB_NO_MEMORY(ctdb, c);
-
- c->key = call->key;
- c->call_data = &call->call_data;
- c->record_data.dptr = talloc_memdup(c, data->dptr, data->dsize);
- c->record_data.dsize = data->dsize;
- CTDB_NO_MEMORY(ctdb, c->record_data.dptr);
- c->new_data = NULL;
- c->reply_data = NULL;
- c->status = 0;
-
- for (fn=ctdb_db->calls;fn;fn=fn->next) {
- if (fn->id == call->call_id) break;
- }
- if (fn == NULL) {
- ctdb_set_error(ctdb, "Unknown call id %u\n", call->call_id);
- talloc_free(c);
- return -1;
- }
-
- if (fn->fn(c) != 0) {
- ctdb_set_error(ctdb, "ctdb_call %u failed\n", call->call_id);
- talloc_free(c);
- return -1;
- }
-
- if (header->laccessor != caller) {
- header->lacount = 0;
- }
- header->laccessor = caller;
- header->lacount++;
-
- /* we need to force the record to be written out if this was a remote access,
- so that the lacount is updated */
- if (c->new_data == NULL && header->laccessor != ctdb->pnn) {
- c->new_data = &c->record_data;
- }
-
- if (c->new_data) {
- /* XXX check that we always have the lock here? */
- if (ctdb_ltdb_store(ctdb_db, call->key, header, *c->new_data) != 0) {
- ctdb_set_error(ctdb, "ctdb_call tdb_store failed\n");
- talloc_free(c);
- return -1;
- }
- }
-
- if (c->reply_data) {
- call->reply_data = *c->reply_data;
-
- talloc_steal(call, call->reply_data.dptr);
- talloc_set_name_const(call->reply_data.dptr, __location__);
- } else {
- call->reply_data.dptr = NULL;
- call->reply_data.dsize = 0;
- }
- call->status = c->status;
-
- talloc_free(c);
-
- return 0;
-}
-
-
-/*
- queue a packet for sending from client to daemon
-*/
-static int ctdb_client_queue_pkt(struct ctdb_context *ctdb, struct ctdb_req_header *hdr)
-{
- return ctdb_queue_send(ctdb->daemon.queue, (uint8_t *)hdr, hdr->length);
-}
-
-
-/*
- called when a CTDB_REPLY_CALL packet comes in in the client
-
- This packet comes in response to a CTDB_REQ_CALL request packet. It
- contains any reply data from the call
-*/
-static void ctdb_client_reply_call(struct ctdb_context *ctdb, struct ctdb_req_header *hdr)
-{
- struct ctdb_reply_call *c = (struct ctdb_reply_call *)hdr;
- struct ctdb_client_call_state *state;
-
- state = ctdb_reqid_find(ctdb, hdr->reqid, struct ctdb_client_call_state);
- if (state == NULL) {
- DEBUG(DEBUG_ERR,(__location__ " reqid %u not found\n", hdr->reqid));
- return;
- }
-
- if (hdr->reqid != state->reqid) {
- /* we found a record but it was the wrong one */
- DEBUG(DEBUG_ERR, ("Dropped client call reply with reqid:%u\n",hdr->reqid));
- return;
- }
-
- state->call->reply_data.dptr = c->data;
- state->call->reply_data.dsize = c->datalen;
- state->call->status = c->status;
-
- talloc_steal(state, c);
-
- state->state = CTDB_CALL_DONE;
-
- if (state->async.fn) {
- state->async.fn(state);
- }
-}
-
-static void ctdb_client_reply_control(struct ctdb_context *ctdb, struct ctdb_req_header *hdr);
-
-/*
- this is called in the client, when data comes in from the daemon
- */
-static void ctdb_client_read_cb(uint8_t *data, size_t cnt, void *args)
-{
- struct ctdb_context *ctdb = talloc_get_type(args, struct ctdb_context);
- struct ctdb_req_header *hdr = (struct ctdb_req_header *)data;
- TALLOC_CTX *tmp_ctx;
-
- /* place the packet as a child of a tmp_ctx. We then use
- talloc_free() below to free it. If any of the calls want
- to keep it, then they will steal it somewhere else, and the
- talloc_free() will be a no-op */
- tmp_ctx = talloc_new(ctdb);
- talloc_steal(tmp_ctx, hdr);
-
- if (cnt == 0) {
- DEBUG(DEBUG_INFO,("Daemon has exited - shutting down client\n"));
- exit(0);
- }
-
- if (cnt < sizeof(*hdr)) {
- DEBUG(DEBUG_CRIT,("Bad packet length %u in client\n", (unsigned)cnt));
- goto done;
- }
- if (cnt != hdr->length) {
- ctdb_set_error(ctdb, "Bad header length %u expected %u in client\n",
- (unsigned)hdr->length, (unsigned)cnt);
- goto done;
- }
-
- if (hdr->ctdb_magic != CTDB_MAGIC) {
- ctdb_set_error(ctdb, "Non CTDB packet rejected in client\n");
- goto done;
- }
-
- if (hdr->ctdb_version != CTDB_VERSION) {
- ctdb_set_error(ctdb, "Bad CTDB version 0x%x rejected in client\n", hdr->ctdb_version);
- goto done;
- }
-
- switch (hdr->operation) {
- case CTDB_REPLY_CALL:
- ctdb_client_reply_call(ctdb, hdr);
- break;
-
- case CTDB_REQ_MESSAGE:
- ctdb_request_message(ctdb, hdr);
- break;
-
- case CTDB_REPLY_CONTROL:
- ctdb_client_reply_control(ctdb, hdr);
- break;
-
- default:
- DEBUG(DEBUG_CRIT,("bogus operation code:%u\n",hdr->operation));
- }
-
-done:
- talloc_free(tmp_ctx);
-}
-
-/*
- connect to a unix domain socket
-*/
-int ctdb_socket_connect(struct ctdb_context *ctdb)
-{
- struct sockaddr_un addr;
-
- memset(&addr, 0, sizeof(addr));
- addr.sun_family = AF_UNIX;
- strncpy(addr.sun_path, ctdb->daemon.name, sizeof(addr.sun_path));
-
- ctdb->daemon.sd = socket(AF_UNIX, SOCK_STREAM, 0);
- if (ctdb->daemon.sd == -1) {
- return -1;
- }
-
- set_nonblocking(ctdb->daemon.sd);
- set_close_on_exec(ctdb->daemon.sd);
-
- if (connect(ctdb->daemon.sd, (struct sockaddr *)&addr, sizeof(addr)) == -1) {
- close(ctdb->daemon.sd);
- ctdb->daemon.sd = -1;
- return -1;
- }
-
- ctdb->daemon.queue = ctdb_queue_setup(ctdb, ctdb, ctdb->daemon.sd,
- CTDB_DS_ALIGNMENT,
- ctdb_client_read_cb, ctdb);
- return 0;
-}
struct ctdb_record_handle {
-/*
- destroy a ctdb_call in client
-*/
-static int ctdb_client_call_destructor(struct ctdb_client_call_state *state)
-{
- ctdb_reqid_remove(state->ctdb_db->ctdb, state->reqid);
- return 0;
-}
-
-/*
- construct an event driven local ctdb_call
-
- this is used so that locally processed ctdb_call requests are processed
- in an event driven manner
-*/
-static struct ctdb_client_call_state *ctdb_client_call_local_send(struct ctdb_db_context *ctdb_db,
- struct ctdb_call *call,
- struct ctdb_ltdb_header *header,
- TDB_DATA *data)
-{
- struct ctdb_client_call_state *state;
- struct ctdb_context *ctdb = ctdb_db->ctdb;
- int ret;
-
- state = talloc_zero(ctdb_db, struct ctdb_client_call_state);
- CTDB_NO_MEMORY_NULL(ctdb, state);
- state->call = talloc_zero(state, struct ctdb_call);
- CTDB_NO_MEMORY_NULL(ctdb, state->call);
-
- talloc_steal(state, data->dptr);
-
- state->state = CTDB_CALL_DONE;
- *(state->call) = *call;
- state->ctdb_db = ctdb_db;
-
- ret = ctdb_call_local(ctdb_db, state->call, header, state, data, ctdb->pnn);
-
- return state;
-}
-
-/*
- make a ctdb call to the local daemon - async send. Called from client context.
-
- This constructs a ctdb_call request and queues it for processing.
- This call never blocks.
-*/
-struct ctdb_client_call_state *ctdb_call_send(struct ctdb_db_context *ctdb_db,
- struct ctdb_call *call)
-{
- struct ctdb_client_call_state *state;
- struct ctdb_context *ctdb = ctdb_db->ctdb;
- struct ctdb_ltdb_header header;
- TDB_DATA data;
- int ret;
- size_t len;
- struct ctdb_req_call *c;
-
- /* if the domain socket is not yet open, open it */
- if (ctdb->daemon.sd==-1) {
- ctdb_socket_connect(ctdb);
- }
-
- ret = ctdb_ltdb_lock(ctdb_db, call->key);
- if (ret != 0) {
- DEBUG(DEBUG_ERR,(__location__ " Failed to get chainlock\n"));
- return NULL;
- }
-
- ret = ctdb_ltdb_fetch(ctdb_db, call->key, &header, ctdb_db, &data);
-
- if (ret == 0 && header.dmaster == ctdb->pnn) {
- state = ctdb_client_call_local_send(ctdb_db, call, &header, &data);
- talloc_free(data.dptr);
- ctdb_ltdb_unlock(ctdb_db, call->key);
- return state;
- }
-
- ctdb_ltdb_unlock(ctdb_db, call->key);
- talloc_free(data.dptr);
-
- state = talloc_zero(ctdb_db, struct ctdb_client_call_state);
- if (state == NULL) {
- DEBUG(DEBUG_ERR, (__location__ " failed to allocate state\n"));
- return NULL;
- }
- state->call = talloc_zero(state, struct ctdb_call);
- if (state->call == NULL) {
- DEBUG(DEBUG_ERR, (__location__ " failed to allocate state->call\n"));
- return NULL;
- }
-
- len = offsetof(struct ctdb_req_call, data) + call->key.dsize + call->call_data.dsize;
- c = ctdbd_allocate_pkt(ctdb, state, CTDB_REQ_CALL, len, struct ctdb_req_call);
- if (c == NULL) {
- DEBUG(DEBUG_ERR, (__location__ " failed to allocate packet\n"));
- return NULL;
- }
-
- state->reqid = ctdb_reqid_new(ctdb, state);
- state->ctdb_db = ctdb_db;
- talloc_set_destructor(state, ctdb_client_call_destructor);
-
- c->hdr.reqid = state->reqid;
- c->flags = call->flags;
- c->db_id = ctdb_db->db_id;
- c->callid = call->call_id;
- c->hopcount = 0;
- c->keylen = call->key.dsize;
- c->calldatalen = call->call_data.dsize;
- memcpy(&c->data[0], call->key.dptr, call->key.dsize);
- memcpy(&c->data[call->key.dsize],
- call->call_data.dptr, call->call_data.dsize);
- *(state->call) = *call;
- state->call->call_data.dptr = &c->data[call->key.dsize];
- state->call->key.dptr = &c->data[0];
-
- state->state = CTDB_CALL_WAIT;
-
-
- ctdb_client_queue_pkt(ctdb, &c->hdr);
- return state;
-}
/*
}
-/*
- tell the daemon what messaging srvid we will use, and register the message
- handler function in the client
-*/
-int ctdb_set_message_handler(struct ctdb_context *ctdb, uint64_t srvid,
- ctdb_message_fn_t handler,
- void *private_data)
-
-{
- int res;
- int32_t status;
-
- res = ctdb_control(ctdb, CTDB_CURRENT_NODE, srvid, CTDB_CONTROL_REGISTER_SRVID, 0,
- tdb_null, NULL, NULL, &status, NULL, NULL);
- if (res != 0 || status != 0) {
- DEBUG(DEBUG_ERR,("Failed to register srvid %llu\n", (unsigned long long)srvid));
- return -1;
- }
-
- /* also need to register the handler with our own ctdb structure */
- return ctdb_register_message_handler(ctdb, ctdb, srvid, handler, private_data);
-}
-
-/*
- tell the daemon we no longer want a srvid
-*/
-int ctdb_remove_message_handler(struct ctdb_context *ctdb, uint64_t srvid, void *private_data)
-{
- int res;
- int32_t status;
-
- res = ctdb_control(ctdb, CTDB_CURRENT_NODE, srvid, CTDB_CONTROL_DEREGISTER_SRVID, 0,
- tdb_null, NULL, NULL, &status, NULL, NULL);
- if (res != 0 || status != 0) {
- DEBUG(DEBUG_ERR,("Failed to deregister srvid %llu\n", (unsigned long long)srvid));
- return -1;
- }
-
- /* also need to register the handler with our own ctdb structure */
- ctdb_deregister_message_handler(ctdb, srvid, private_data);
- return 0;
-}
-
-
-/*
- send a message - from client context
- */
-int ctdb_send_message(struct ctdb_context *ctdb, uint32_t pnn,
- uint64_t srvid, TDB_DATA data)
-{
- struct ctdb_req_message *r;
- int len, res;
-
- len = offsetof(struct ctdb_req_message, data) + data.dsize;
- r = ctdbd_allocate_pkt(ctdb, ctdb, CTDB_REQ_MESSAGE,
- len, struct ctdb_req_message);
- CTDB_NO_MEMORY(ctdb, r);
- r->hdr.destnode = pnn;
- r->srvid = srvid;
- r->datalen = data.dsize;
- memcpy(&r->data[0], data.dptr, data.dsize);
-
- res = ctdb_client_queue_pkt(ctdb, &r->hdr);
- if (res != 0) {
- return res;
- }
-
- talloc_free(r);
- return 0;
-}
/*
*/
int ctdb_record_store(struct ctdb_record_handle *h, TDB_DATA data)
{
- int ret;
- int32_t status;
- struct ctdb_rec_data *rec;
- TDB_DATA recdata;
-
if (h->ctdb_db->persistent) {
- h->header.rsn++;
- }
-
- ret = ctdb_ltdb_store(h->ctdb_db, h->key, &h->header, data);
- if (ret != 0) {
- return ret;
- }
-
- /* don't need the persistent_store control for non-persistent databases */
- if (!h->ctdb_db->persistent) {
- return 0;
- }
-
- rec = ctdb_marshall_record(h, h->ctdb_db->db_id, h->key, &h->header, data);
- if (rec == NULL) {
- DEBUG(DEBUG_ERR,("Unable to marshall record in ctdb_record_store\n"));
- return -1;
- }
-
- recdata.dptr = (uint8_t *)rec;
- recdata.dsize = rec->length;
-
- ret = ctdb_control(h->ctdb_db->ctdb, CTDB_CURRENT_NODE, 0,
- CTDB_CONTROL_PERSISTENT_STORE, 0,
- recdata, NULL, NULL, &status, NULL, NULL);
-
- talloc_free(rec);
-
- if (ret != 0 || status != 0) {
- DEBUG(DEBUG_ERR,("Failed persistent store in ctdb_record_store\n"));
+ DEBUG(DEBUG_ERR, (__location__ " ctdb_record_store prohibited for persistent dbs\n"));
return -1;
}
- return 0;
+ return ctdb_ltdb_store(h->ctdb_db, h->key, &h->header, data);
}
/*
-/*
- called when a control completes or timesout to invoke the callback
- function the user provided
-*/
-static void invoke_control_callback(struct event_context *ev, struct timed_event *te,
- struct timeval t, void *private_data)
-{
- struct ctdb_client_control_state *state;
- TALLOC_CTX *tmp_ctx = talloc_new(NULL);
- int ret;
- state = talloc_get_type(private_data, struct ctdb_client_control_state);
- talloc_steal(tmp_ctx, state);
- ret = ctdb_control_recv(state->ctdb, state, state,
- NULL,
- NULL,
- NULL);
- talloc_free(tmp_ctx);
-}
-/*
- called when a CTDB_REPLY_CONTROL packet comes in in the client
- This packet comes in response to a CTDB_REQ_CONTROL request packet. It
- contains any reply data from the control
-*/
-static void ctdb_client_reply_control(struct ctdb_context *ctdb,
- struct ctdb_req_header *hdr)
+/*
+ send a ctdb control message
+ timeout specifies how long we should wait for a reply.
+ if timeout is NULL we wait indefinitely
+ */
+int ctdb_control(struct ctdb_context *ctdb, uint32_t destnode, uint64_t srvid,
+ uint32_t opcode, uint32_t flags, TDB_DATA data,
+ TALLOC_CTX *mem_ctx, TDB_DATA *outdata, int32_t *status,
+ struct timeval *timeout,
+ char **errormsg)
{
- struct ctdb_reply_control *c = (struct ctdb_reply_control *)hdr;
struct ctdb_client_control_state *state;
- state = ctdb_reqid_find(ctdb, hdr->reqid, struct ctdb_client_control_state);
- if (state == NULL) {
- DEBUG(DEBUG_ERR,(__location__ " reqid %u not found\n", hdr->reqid));
- return;
- }
-
- if (hdr->reqid != state->reqid) {
- /* we found a record but it was the wrong one */
- DEBUG(DEBUG_ERR, ("Dropped orphaned reply control with reqid:%u\n",hdr->reqid));
- return;
- }
-
- state->outdata.dptr = c->data;
- state->outdata.dsize = c->datalen;
- state->status = c->status;
- if (c->errorlen) {
- state->errormsg = talloc_strndup(state,
- (char *)&c->data[c->datalen],
- c->errorlen);
- }
-
- /* state->outdata now uses resources from c so we dont want c
- to just dissappear from under us while state is still alive
- */
- talloc_steal(state, c);
+ state = ctdb_control_send(ctdb, destnode, srvid, opcode,
+ flags, data, mem_ctx,
+ timeout, errormsg);
+ return ctdb_control_recv(ctdb, state, mem_ctx, outdata, status,
+ errormsg);
+}
- state->state = CTDB_CONTROL_DONE;
- /* if we had a callback registered for this control, pull the response
- and call the callback.
- */
- if (state->async.fn) {
- event_add_timed(ctdb->ev, state, timeval_zero(), invoke_control_callback, state);
- }
-}
/*
- destroy a ctdb_control in client
-*/
-static int ctdb_control_destructor(struct ctdb_client_control_state *state)
-{
- ctdb_reqid_remove(state->ctdb, state->reqid);
- return 0;
-}
-
-
-/* time out handler for ctdb_control */
-static void control_timeout_func(struct event_context *ev, struct timed_event *te,
- struct timeval t, void *private_data)
-{
- struct ctdb_client_control_state *state = talloc_get_type(private_data, struct ctdb_client_control_state);
-
- DEBUG(DEBUG_ERR,("control timed out. reqid:%d opcode:%d dstnode:%d\n", state->reqid, state->c->opcode, state->c->hdr.destnode));
-
- state->state = CTDB_CONTROL_TIMEOUT;
-
- /* if we had a callback registered for this control, pull the response
- and call the callback.
- */
- if (state->async.fn) {
- event_add_timed(state->ctdb->ev, state, timeval_zero(), invoke_control_callback, state);
- }
-}
-
-/* async version of send control request */
-struct ctdb_client_control_state *ctdb_control_send(struct ctdb_context *ctdb,
- uint32_t destnode, uint64_t srvid,
- uint32_t opcode, uint32_t flags, TDB_DATA data,
- TALLOC_CTX *mem_ctx,
- struct timeval *timeout,
- char **errormsg)
-{
- struct ctdb_client_control_state *state;
- size_t len;
- struct ctdb_req_control *c;
- int ret;
-
- if (errormsg) {
- *errormsg = NULL;
- }
-
- /* if the domain socket is not yet open, open it */
- if (ctdb->daemon.sd==-1) {
- ctdb_socket_connect(ctdb);
- }
-
- state = talloc_zero(mem_ctx, struct ctdb_client_control_state);
- CTDB_NO_MEMORY_NULL(ctdb, state);
-
- state->ctdb = ctdb;
- state->reqid = ctdb_reqid_new(ctdb, state);
- state->state = CTDB_CONTROL_WAIT;
- state->errormsg = NULL;
-
- talloc_set_destructor(state, ctdb_control_destructor);
-
- len = offsetof(struct ctdb_req_control, data) + data.dsize;
- c = ctdbd_allocate_pkt(ctdb, state, CTDB_REQ_CONTROL,
- len, struct ctdb_req_control);
- state->c = c;
- CTDB_NO_MEMORY_NULL(ctdb, c);
- c->hdr.reqid = state->reqid;
- c->hdr.destnode = destnode;
- c->hdr.reqid = state->reqid;
- c->opcode = opcode;
- c->client_id = 0;
- c->flags = flags;
- c->srvid = srvid;
- c->datalen = data.dsize;
- if (data.dsize) {
- memcpy(&c->data[0], data.dptr, data.dsize);
- }
-
- /* timeout */
- if (timeout && !timeval_is_zero(timeout)) {
- event_add_timed(ctdb->ev, state, *timeout, control_timeout_func, state);
- }
-
- ret = ctdb_client_queue_pkt(ctdb, &(c->hdr));
- if (ret != 0) {
- talloc_free(state);
- return NULL;
- }
-
- if (flags & CTDB_CTRL_FLAG_NOREPLY) {
- talloc_free(state);
- return NULL;
- }
-
- return state;
-}
-
-
-/* async version of receive control reply */
-int ctdb_control_recv(struct ctdb_context *ctdb,
- struct ctdb_client_control_state *state,
- TALLOC_CTX *mem_ctx,
- TDB_DATA *outdata, int32_t *status, char **errormsg)
-{
- TALLOC_CTX *tmp_ctx;
-
- if (status != NULL) {
- *status = -1;
- }
- if (errormsg != NULL) {
- *errormsg = NULL;
- }
-
- if (state == NULL) {
- return -1;
- }
-
- /* prevent double free of state */
- tmp_ctx = talloc_new(ctdb);
- talloc_steal(tmp_ctx, state);
-
- /* loop one event at a time until we either timeout or the control
- completes.
- */
- while (state->state == CTDB_CONTROL_WAIT) {
- event_loop_once(ctdb->ev);
- }
-
- if (state->state != CTDB_CONTROL_DONE) {
- DEBUG(DEBUG_ERR,(__location__ " ctdb_control_recv failed\n"));
- if (state->async.fn) {
- state->async.fn(state);
- }
- talloc_free(tmp_ctx);
- return -1;
- }
-
- if (state->errormsg) {
- DEBUG(DEBUG_ERR,("ctdb_control error: '%s'\n", state->errormsg));
- if (errormsg) {
- (*errormsg) = talloc_move(mem_ctx, &state->errormsg);
- }
- if (state->async.fn) {
- state->async.fn(state);
- }
- talloc_free(tmp_ctx);
- return -1;
- }
-
- if (outdata) {
- *outdata = state->outdata;
- outdata->dptr = talloc_memdup(mem_ctx, outdata->dptr, outdata->dsize);
- }
-
- if (status) {
- *status = state->status;
- }
-
- if (state->async.fn) {
- state->async.fn(state);
- }
-
- talloc_free(tmp_ctx);
- return 0;
-}
-
-
-
-/*
- send a ctdb control message
- timeout specifies how long we should wait for a reply.
- if timeout is NULL we wait indefinitely
- */
-int ctdb_control(struct ctdb_context *ctdb, uint32_t destnode, uint64_t srvid,
- uint32_t opcode, uint32_t flags, TDB_DATA data,
- TALLOC_CTX *mem_ctx, TDB_DATA *outdata, int32_t *status,
- struct timeval *timeout,
- char **errormsg)
-{
- struct ctdb_client_control_state *state;
-
- state = ctdb_control_send(ctdb, destnode, srvid, opcode,
- flags, data, mem_ctx,
- timeout, errormsg);
- return ctdb_control_recv(ctdb, state, mem_ctx, outdata, status,
- errormsg);
-}
-
-
-
-
-/*
- a process exists call. Returns 0 if process exists, -1 otherwise
- */
-int ctdb_ctrl_process_exists(struct ctdb_context *ctdb, uint32_t destnode, pid_t pid)
+ a process exists call. Returns 0 if process exists, -1 otherwise
+ */
+int ctdb_ctrl_process_exists(struct ctdb_context *ctdb, uint32_t destnode, pid_t pid)
{
int ret;
TDB_DATA data;
return 0;
}
+/*
+ get the health status of a db
+ */
+int ctdb_ctrl_getdbhealth(struct ctdb_context *ctdb,
+ struct timeval timeout,
+ uint32_t destnode,
+ uint32_t dbid, TALLOC_CTX *mem_ctx,
+ const char **reason)
+{
+ int ret;
+ int32_t res;
+ TDB_DATA data;
+
+ data.dptr = (uint8_t *)&dbid;
+ data.dsize = sizeof(dbid);
+
+ ret = ctdb_control(ctdb, destnode, 0,
+ CTDB_CONTROL_DB_GET_HEALTH, 0, data,
+ mem_ctx, &data, &res, &timeout, NULL);
+ if (ret != 0 || res != 0) {
+ return -1;
+ }
+
+ if (data.dsize == 0) {
+ (*reason) = NULL;
+ return 0;
+ }
+
+ (*reason) = talloc_strndup(mem_ctx, (const char *)data.dptr, data.dsize);
+ if ((*reason) == NULL) {
+ return -1;
+ }
+
+ talloc_free(data.dptr);
+
+ return 0;
+}
+
/*
create a database
*/
}
tdb_flags = persistent?TDB_DEFAULT:TDB_NOSYNC;
- if (!ctdb->do_setsched) {
+ if (ctdb->valgrinding) {
tdb_flags |= TDB_NOMMAP;
}
+ tdb_flags |= TDB_DISALLOW_NESTING;
ctdb_db->ltdb = tdb_wrap_open(ctdb, ctdb_db->db_path, 0, tdb_flags, O_RDWR, 0);
if (ctdb_db->ltdb == NULL) {
/*
called on each key during a catdb
*/
-static int dumpdb_fn(struct ctdb_context *ctdb, TDB_DATA key, TDB_DATA data, void *p)
+int ctdb_dumpdb_record(struct ctdb_context *ctdb, TDB_DATA key, TDB_DATA data, void *p)
{
int i;
FILE *f = (FILE *)p;
struct ctdb_ltdb_header *h = (struct ctdb_ltdb_header *)data.dptr;
- fprintf(f, "dmaster: %u\n", h->dmaster);
- fprintf(f, "rsn: %llu\n", (unsigned long long)h->rsn);
-
fprintf(f, "key(%u) = \"", (unsigned)key.dsize);
for (i=0;i<key.dsize;i++) {
if (ISASCII(key.dptr[i])) {
}
fprintf(f, "\"\n");
- fprintf(f, "data(%u) = \"", (unsigned)data.dsize);
+ fprintf(f, "dmaster: %u\n", h->dmaster);
+ fprintf(f, "rsn: %llu\n", (unsigned long long)h->rsn);
+
+ fprintf(f, "data(%u) = \"", (unsigned)(data.dsize - sizeof(*h)));
for (i=sizeof(*h);i<data.dsize;i++) {
if (ISASCII(data.dptr[i])) {
fprintf(f, "%c", data.dptr[i]);
}
fprintf(f, "\"\n");
+ fprintf(f, "\n");
+
return 0;
}
*/
int ctdb_dump_db(struct ctdb_db_context *ctdb_db, FILE *f)
{
- return ctdb_traverse(ctdb_db, dumpdb_fn, f);
+ return ctdb_traverse(ctdb_db, ctdb_dumpdb_record, f);
}
/*
async freeze send control
*/
struct ctdb_client_control_state *
-ctdb_ctrl_freeze_send(struct ctdb_context *ctdb, TALLOC_CTX *mem_ctx, struct timeval timeout, uint32_t destnode)
+ctdb_ctrl_freeze_send(struct ctdb_context *ctdb, TALLOC_CTX *mem_ctx, struct timeval timeout, uint32_t destnode, uint32_t priority)
{
- return ctdb_control_send(ctdb, destnode, 0,
+ return ctdb_control_send(ctdb, destnode, priority,
CTDB_CONTROL_FREEZE, 0, tdb_null,
mem_ctx, &timeout, NULL);
}
}
/*
- freeze a node
+ freeze databases of a certain priority
*/
-int ctdb_ctrl_freeze(struct ctdb_context *ctdb, struct timeval timeout, uint32_t destnode)
+int ctdb_ctrl_freeze_priority(struct ctdb_context *ctdb, struct timeval timeout, uint32_t destnode, uint32_t priority)
{
TALLOC_CTX *tmp_ctx = talloc_new(ctdb);
struct ctdb_client_control_state *state;
int ret;
- state = ctdb_ctrl_freeze_send(ctdb, tmp_ctx, timeout, destnode);
+ state = ctdb_ctrl_freeze_send(ctdb, tmp_ctx, timeout, destnode, priority);
ret = ctdb_ctrl_freeze_recv(ctdb, tmp_ctx, state);
talloc_free(tmp_ctx);
return ret;
}
+/* Freeze all databases */
+int ctdb_ctrl_freeze(struct ctdb_context *ctdb, struct timeval timeout, uint32_t destnode)
+{
+ int i;
+
+ for (i=1; i<=NUM_DB_PRIORITIES; i++) {
+ if (ctdb_ctrl_freeze_priority(ctdb, timeout, destnode, i) != 0) {
+ return -1;
+ }
+ }
+ return 0;
+}
+
/*
- thaw a node
+ thaw databases of a certain priority
*/
-int ctdb_ctrl_thaw(struct ctdb_context *ctdb, struct timeval timeout, uint32_t destnode)
+int ctdb_ctrl_thaw_priority(struct ctdb_context *ctdb, struct timeval timeout, uint32_t destnode, uint32_t priority)
{
int ret;
int32_t res;
- ret = ctdb_control(ctdb, destnode, 0,
+ ret = ctdb_control(ctdb, destnode, priority,
CTDB_CONTROL_THAW, 0, tdb_null,
NULL, NULL, &res, &timeout, NULL);
if (ret != 0 || res != 0) {
return 0;
}
+/* thaw all databases */
+int ctdb_ctrl_thaw(struct ctdb_context *ctdb, struct timeval timeout, uint32_t destnode)
+{
+ return ctdb_ctrl_thaw_priority(ctdb, timeout, destnode, 0);
+}
+
/*
get pnn of a node, or -1
*/
}
-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)
+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 ret;
TDB_DATA outdata;
int32_t res;
ret = ctdb_control(ctdb, destnode, 0,
- CTDB_CONTROL_GET_PUBLIC_IPS, 0, tdb_null,
+ CTDB_CONTROL_GET_PUBLIC_IPS, flags, tdb_null,
mem_ctx, &outdata, &res, &timeout, NULL);
if (ret == 0 && res == -1) {
DEBUG(DEBUG_ERR,(__location__ " ctdb_control to get public ips failed, falling back to ipv4-only version\n"));
return 0;
}
+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)
+{
+ return ctdb_ctrl_get_public_ips_flags(ctdb, timeout,
+ destnode, mem_ctx,
+ 0, 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)
len = offsetof(struct ctdb_all_public_ips, ips) +
ipsv4->num*sizeof(struct ctdb_public_ip);
*ips = talloc_zero_size(mem_ctx, len);
+ CTDB_NO_MEMORY(ctdb, *ips);
(*ips)->num = ipsv4->num;
for (i=0; i<ipsv4->num; i++) {
(*ips)->ips[i].pnn = ipsv4->ips[i].pnn;
return 0;
}
-/*
- set/clear the permanent disabled bit on a remote node
- */
-int ctdb_ctrl_modflags(struct ctdb_context *ctdb, struct timeval timeout, uint32_t destnode,
- uint32_t set, uint32_t clear)
+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 ret;
- TDB_DATA data;
- struct ctdb_node_modflags m;
+ TDB_DATA indata;
+ TDB_DATA outdata;
int32_t res;
+ struct ctdb_control_public_ip_info *info;
+ uint32_t len;
+ uint32_t i;
- m.set = set;
- m.clear = clear;
-
- data.dsize = sizeof(m);
- data.dptr = (unsigned char *)&m;
+ indata.dptr = discard_const_p(uint8_t, addr);
+ indata.dsize = sizeof(*addr);
- ret = ctdb_control(ctdb, destnode, 0,
- CTDB_CONTROL_MODIFY_FLAGS, 0, data,
- NULL, NULL, &res, &timeout, NULL);
+ ret = ctdb_control(ctdb, destnode, 0,
+ CTDB_CONTROL_GET_PUBLIC_IP_INFO, 0, indata,
+ mem_ctx, &outdata, &res, &timeout, NULL);
if (ret != 0 || res != 0) {
- DEBUG(DEBUG_ERR,(__location__ " ctdb_control for modflags failed\n"));
+ DEBUG(DEBUG_ERR,(__location__ " ctdb_control for get public ip info "
+ "failed ret:%d res:%d\n",
+ ret, res));
+ return -1;
+ }
+
+ len = offsetof(struct ctdb_control_public_ip_info, ifaces);
+ if (len > outdata.dsize) {
+ DEBUG(DEBUG_ERR,(__location__ " ctdb_control for get public ip info "
+ "returned invalid data with size %u > %u\n",
+ (unsigned int)outdata.dsize,
+ (unsigned int)len));
+ dump_data(DEBUG_DEBUG, outdata.dptr, outdata.dsize);
+ return -1;
+ }
+
+ info = (struct ctdb_control_public_ip_info *)outdata.dptr;
+ len += info->num*sizeof(struct ctdb_control_iface_info);
+
+ if (len > outdata.dsize) {
+ DEBUG(DEBUG_ERR,(__location__ " ctdb_control for get public ip info "
+ "returned invalid data with size %u > %u\n",
+ (unsigned int)outdata.dsize,
+ (unsigned int)len));
+ dump_data(DEBUG_DEBUG, outdata.dptr, outdata.dsize);
+ return -1;
+ }
+
+ /* make sure we null terminate the returned strings */
+ for (i=0; i < info->num; i++) {
+ info->ifaces[i].name[CTDB_IFACE_SIZE] = '\0';
+ }
+
+ *_info = (struct ctdb_control_public_ip_info *)talloc_memdup(mem_ctx,
+ outdata.dptr,
+ outdata.dsize);
+ talloc_free(outdata.dptr);
+ if (*_info == NULL) {
+ DEBUG(DEBUG_ERR,(__location__ " ctdb_control for get public ip info "
+ "talloc_memdup size %u failed\n",
+ (unsigned int)outdata.dsize));
return -1;
}
return 0;
}
-
-/*
- get all tunables
- */
-int ctdb_ctrl_get_all_tunables(struct ctdb_context *ctdb,
- struct timeval timeout,
- uint32_t destnode,
- struct ctdb_tunable *tunables)
+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)
{
- TDB_DATA outdata;
int ret;
+ TDB_DATA outdata;
int32_t res;
+ struct ctdb_control_get_ifaces *ifaces;
+ uint32_t len;
+ uint32_t i;
- ret = ctdb_control(ctdb, destnode, 0, CTDB_CONTROL_GET_ALL_TUNABLES, 0, tdb_null, ctdb,
- &outdata, &res, &timeout, NULL);
+ ret = ctdb_control(ctdb, destnode, 0,
+ CTDB_CONTROL_GET_IFACES, 0, tdb_null,
+ mem_ctx, &outdata, &res, &timeout, NULL);
if (ret != 0 || res != 0) {
- DEBUG(DEBUG_ERR,(__location__ " ctdb_control for get all tunables failed\n"));
+ DEBUG(DEBUG_ERR,(__location__ " ctdb_control for get ifaces "
+ "failed ret:%d res:%d\n",
+ ret, res));
return -1;
}
- if (outdata.dsize != sizeof(*tunables)) {
- DEBUG(DEBUG_ERR,(__location__ " bad data size %u in ctdb_ctrl_get_all_tunables should be %u\n",
- (unsigned)outdata.dsize, (unsigned)sizeof(*tunables)));
- return -1;
+ len = offsetof(struct ctdb_control_get_ifaces, ifaces);
+ if (len > outdata.dsize) {
+ DEBUG(DEBUG_ERR,(__location__ " ctdb_control for get ifaces "
+ "returned invalid data with size %u > %u\n",
+ (unsigned int)outdata.dsize,
+ (unsigned int)len));
+ dump_data(DEBUG_DEBUG, outdata.dptr, outdata.dsize);
+ return -1;
}
- *tunables = *(struct ctdb_tunable *)outdata.dptr;
- talloc_free(outdata.dptr);
- return 0;
-}
+ ifaces = (struct ctdb_control_get_ifaces *)outdata.dptr;
+ len += ifaces->num*sizeof(struct ctdb_control_iface_info);
-/*
- add a public address to a node
- */
-int ctdb_ctrl_add_public_ip(struct ctdb_context *ctdb,
- struct timeval timeout,
- uint32_t destnode,
- struct ctdb_control_ip_iface *pub)
-{
- TDB_DATA data;
- int32_t res;
- int ret;
+ if (len > outdata.dsize) {
+ DEBUG(DEBUG_ERR,(__location__ " ctdb_control for get ifaces "
+ "returned invalid data with size %u > %u\n",
+ (unsigned int)outdata.dsize,
+ (unsigned int)len));
+ dump_data(DEBUG_DEBUG, outdata.dptr, outdata.dsize);
+ return -1;
+ }
- data.dsize = offsetof(struct ctdb_control_ip_iface, iface) + pub->len;
- data.dptr = (unsigned char *)pub;
+ /* make sure we null terminate the returned strings */
+ for (i=0; i < ifaces->num; i++) {
+ ifaces->ifaces[i].name[CTDB_IFACE_SIZE] = '\0';
+ }
- ret = ctdb_control(ctdb, destnode, 0, CTDB_CONTROL_ADD_PUBLIC_IP, 0, data, NULL,
- NULL, &res, &timeout, NULL);
- if (ret != 0 || res != 0) {
- DEBUG(DEBUG_ERR,(__location__ " ctdb_control for add_public_ip failed\n"));
+ *_ifaces = (struct ctdb_control_get_ifaces *)talloc_memdup(mem_ctx,
+ outdata.dptr,
+ outdata.dsize);
+ talloc_free(outdata.dptr);
+ if (*_ifaces == NULL) {
+ DEBUG(DEBUG_ERR,(__location__ " ctdb_control for get ifaces "
+ "talloc_memdup size %u failed\n",
+ (unsigned int)outdata.dsize));
return -1;
}
return 0;
}
-/*
- delete a public address from a node
- */
-int ctdb_ctrl_del_public_ip(struct ctdb_context *ctdb,
- struct timeval timeout,
- uint32_t destnode,
- struct ctdb_control_ip_iface *pub)
+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)
{
- TDB_DATA data;
- int32_t res;
int ret;
+ TDB_DATA indata;
+ int32_t res;
- data.dsize = offsetof(struct ctdb_control_ip_iface, iface) + pub->len;
- data.dptr = (unsigned char *)pub;
+ indata.dptr = discard_const_p(uint8_t, info);
+ indata.dsize = sizeof(*info);
- ret = ctdb_control(ctdb, destnode, 0, CTDB_CONTROL_DEL_PUBLIC_IP, 0, data, NULL,
- NULL, &res, &timeout, NULL);
+ ret = ctdb_control(ctdb, destnode, 0,
+ CTDB_CONTROL_SET_IFACE_LINK_STATE, 0, indata,
+ mem_ctx, NULL, &res, &timeout, NULL);
if (ret != 0 || res != 0) {
- DEBUG(DEBUG_ERR,(__location__ " ctdb_control for del_public_ip failed\n"));
+ DEBUG(DEBUG_ERR,(__location__ " ctdb_control for set iface link "
+ "failed ret:%d res:%d\n",
+ ret, res));
return -1;
}
}
/*
- kill a tcp connection
+ set/clear the permanent disabled bit on a remote node
*/
-int ctdb_ctrl_killtcp(struct ctdb_context *ctdb,
- struct timeval timeout,
+int ctdb_ctrl_modflags(struct ctdb_context *ctdb, struct timeval timeout, uint32_t destnode,
+ uint32_t set, uint32_t clear)
+{
+ int ret;
+ TDB_DATA data;
+ struct ctdb_node_map *nodemap=NULL;
+ struct ctdb_node_flag_change c;
+ TALLOC_CTX *tmp_ctx = talloc_new(ctdb);
+ uint32_t recmaster;
+ uint32_t *nodes;
+
+
+ /* find the recovery master */
+ ret = ctdb_ctrl_getrecmaster(ctdb, tmp_ctx, timeout, CTDB_CURRENT_NODE, &recmaster);
+ if (ret != 0) {
+ DEBUG(DEBUG_ERR, (__location__ " Unable to get recmaster from local node\n"));
+ talloc_free(tmp_ctx);
+ return ret;
+ }
+
+
+ /* read the node flags from the recmaster */
+ ret = ctdb_ctrl_getnodemap(ctdb, timeout, recmaster, tmp_ctx, &nodemap);
+ if (ret != 0) {
+ DEBUG(DEBUG_ERR, (__location__ " Unable to get nodemap from node %u\n", destnode));
+ talloc_free(tmp_ctx);
+ return -1;
+ }
+ if (destnode >= nodemap->num) {
+ DEBUG(DEBUG_ERR,(__location__ " Nodemap from recmaster does not contain node %d\n", destnode));
+ talloc_free(tmp_ctx);
+ return -1;
+ }
+
+ c.pnn = destnode;
+ c.old_flags = nodemap->nodes[destnode].flags;
+ c.new_flags = c.old_flags;
+ c.new_flags |= set;
+ c.new_flags &= ~clear;
+
+ data.dsize = sizeof(c);
+ data.dptr = (unsigned char *)&c;
+
+ /* send the flags update to all connected nodes */
+ nodes = list_of_connected_nodes(ctdb, nodemap, tmp_ctx, true);
+
+ if (ctdb_client_async_control(ctdb, CTDB_CONTROL_MODIFY_FLAGS,
+ nodes, 0,
+ timeout, false, data,
+ NULL, NULL,
+ NULL) != 0) {
+ DEBUG(DEBUG_ERR, (__location__ " Unable to update nodeflags on remote nodes\n"));
+
+ talloc_free(tmp_ctx);
+ return -1;
+ }
+
+ talloc_free(tmp_ctx);
+ return 0;
+}
+
+
+/*
+ get all tunables
+ */
+int ctdb_ctrl_get_all_tunables(struct ctdb_context *ctdb,
+ struct timeval timeout,
+ uint32_t destnode,
+ struct ctdb_tunable *tunables)
+{
+ TDB_DATA outdata;
+ int ret;
+ int32_t res;
+
+ ret = ctdb_control(ctdb, destnode, 0, CTDB_CONTROL_GET_ALL_TUNABLES, 0, tdb_null, ctdb,
+ &outdata, &res, &timeout, NULL);
+ if (ret != 0 || res != 0) {
+ DEBUG(DEBUG_ERR,(__location__ " ctdb_control for get all tunables failed\n"));
+ return -1;
+ }
+
+ if (outdata.dsize != sizeof(*tunables)) {
+ DEBUG(DEBUG_ERR,(__location__ " bad data size %u in ctdb_ctrl_get_all_tunables should be %u\n",
+ (unsigned)outdata.dsize, (unsigned)sizeof(*tunables)));
+ return -1;
+ }
+
+ *tunables = *(struct ctdb_tunable *)outdata.dptr;
+ talloc_free(outdata.dptr);
+ return 0;
+}
+
+/*
+ add a public address to a node
+ */
+int ctdb_ctrl_add_public_ip(struct ctdb_context *ctdb,
+ struct timeval timeout,
+ uint32_t destnode,
+ struct ctdb_control_ip_iface *pub)
+{
+ TDB_DATA data;
+ int32_t res;
+ int ret;
+
+ data.dsize = offsetof(struct ctdb_control_ip_iface, iface) + pub->len;
+ data.dptr = (unsigned char *)pub;
+
+ ret = ctdb_control(ctdb, destnode, 0, CTDB_CONTROL_ADD_PUBLIC_IP, 0, data, NULL,
+ NULL, &res, &timeout, NULL);
+ if (ret != 0 || res != 0) {
+ DEBUG(DEBUG_ERR,(__location__ " ctdb_control for add_public_ip failed\n"));
+ return -1;
+ }
+
+ return 0;
+}
+
+/*
+ delete a public address from a node
+ */
+int ctdb_ctrl_del_public_ip(struct ctdb_context *ctdb,
+ struct timeval timeout,
+ uint32_t destnode,
+ struct ctdb_control_ip_iface *pub)
+{
+ TDB_DATA data;
+ int32_t res;
+ int ret;
+
+ data.dsize = offsetof(struct ctdb_control_ip_iface, iface) + pub->len;
+ data.dptr = (unsigned char *)pub;
+
+ ret = ctdb_control(ctdb, destnode, 0, CTDB_CONTROL_DEL_PUBLIC_IP, 0, data, NULL,
+ NULL, &res, &timeout, NULL);
+ if (ret != 0 || res != 0) {
+ DEBUG(DEBUG_ERR,(__location__ " ctdb_control for del_public_ip failed\n"));
+ return -1;
+ }
+
+ return 0;
+}
+
+/*
+ kill a tcp connection
+ */
+int ctdb_ctrl_killtcp(struct ctdb_context *ctdb,
+ struct timeval timeout,
uint32_t destnode,
struct ctdb_control_killtcp *killtcp)
{
return 0;
}
-/*
- initialise the ctdb daemon for client applications
-
- NOTE: In current code the daemon does not fork. This is for testing purposes only
- and to simplify the code.
-*/
-struct ctdb_context *ctdb_init(struct event_context *ev)
-{
- struct ctdb_context *ctdb;
-
- ctdb = talloc_zero(ev, struct ctdb_context);
- ctdb->ev = ev;
- ctdb->idr = idr_init(ctdb);
- CTDB_NO_MEMORY_NULL(ctdb, ctdb->idr);
-
- ctdb_set_socketname(ctdb, CTDB_PATH);
-
- return ctdb;
-}
-
-
/*
set some ctdb flags
*/
ctdb->flags |= flags;
}
-/*
- setup the local socket name
-*/
-int ctdb_set_socketname(struct ctdb_context *ctdb, const char *socketname)
-{
- ctdb->daemon.name = talloc_strdup(ctdb, socketname);
- return 0;
-}
/*
return the pnn of this node
*/
if (state->state != CTDB_CONTROL_DONE) {
if ( !data->dont_log_errors) {
- DEBUG(DEBUG_ERR,("Async operation failed with state %d\n opcode:%u", state->state, data->opcode));
+ DEBUG(DEBUG_ERR,("Async operation failed with state %d, opcode:%u\n", state->state, data->opcode));
}
data->fail_count++;
if (data->fail_callback) {
int ctdb_client_async_control(struct ctdb_context *ctdb,
enum ctdb_controls opcode,
uint32_t *nodes,
+ uint64_t srvid,
struct timeval timeout,
bool dont_log_errors,
TDB_DATA data,
for (j=0; j<num_nodes; j++) {
uint32_t pnn = nodes[j];
- state = ctdb_control_send(ctdb, pnn, 0, opcode,
+ state = ctdb_control_send(ctdb, pnn, srvid, opcode,
0, data, async_data, &timeout, NULL);
if (state == NULL) {
DEBUG(DEBUG_ERR,(__location__ " Failed to call async control %u\n", (unsigned)opcode));
return nodes;
}
+uint32_t *list_of_active_nodes_except_pnn(struct ctdb_context *ctdb,
+ struct ctdb_node_map *node_map,
+ TALLOC_CTX *mem_ctx,
+ uint32_t pnn)
+{
+ int i, j, num_nodes;
+ uint32_t *nodes;
+
+ for (i=num_nodes=0;i<node_map->num;i++) {
+ if (node_map->nodes[i].flags & NODE_FLAGS_INACTIVE) {
+ continue;
+ }
+ if (node_map->nodes[i].pnn == pnn) {
+ continue;
+ }
+ num_nodes++;
+ }
+
+ nodes = talloc_array(mem_ctx, uint32_t, num_nodes);
+ CTDB_NO_MEMORY_FATAL(ctdb, nodes);
+
+ for (i=j=0;i<node_map->num;i++) {
+ if (node_map->nodes[i].flags & NODE_FLAGS_INACTIVE) {
+ continue;
+ }
+ if (node_map->nodes[i].pnn == pnn) {
+ continue;
+ }
+ nodes[j++] = node_map->nodes[i].pnn;
+ }
+
+ return nodes;
+}
+
+uint32_t *list_of_connected_nodes(struct ctdb_context *ctdb,
+ struct ctdb_node_map *node_map,
+ TALLOC_CTX *mem_ctx,
+ bool include_self)
+{
+ int i, j, num_nodes;
+ uint32_t *nodes;
+
+ for (i=num_nodes=0;i<node_map->num;i++) {
+ if (node_map->nodes[i].flags & NODE_FLAGS_DISCONNECTED) {
+ continue;
+ }
+ if (node_map->nodes[i].pnn == ctdb->pnn && !include_self) {
+ continue;
+ }
+ num_nodes++;
+ }
+
+ nodes = talloc_array(mem_ctx, uint32_t, num_nodes);
+ CTDB_NO_MEMORY_FATAL(ctdb, nodes);
+
+ for (i=j=0;i<node_map->num;i++) {
+ if (node_map->nodes[i].flags & NODE_FLAGS_DISCONNECTED) {
+ continue;
+ }
+ if (node_map->nodes[i].pnn == ctdb->pnn && !include_self) {
+ continue;
+ }
+ nodes[j++] = node_map->nodes[i].pnn;
+ }
+
+ return nodes;
+}
+
/*
this is used to test if a pnn lock exists and if it exists will return
the number of connections that pnn has reported or -1 if that recovery
return ret;
}
+/**
+ * check whether a transaction is active on a given db on a given node
+ */
+int32_t ctdb_ctrl_transaction_active(struct ctdb_context *ctdb,
+ uint32_t destnode,
+ uint32_t db_id)
+{
+ int32_t status;
+ int ret;
+ TDB_DATA indata;
+
+ indata.dptr = (uint8_t *)&db_id;
+ indata.dsize = sizeof(db_id);
+
+ ret = ctdb_control(ctdb, destnode, 0,
+ CTDB_CONTROL_TRANS2_ACTIVE,
+ 0, indata, NULL, NULL, &status,
+ NULL, NULL);
+
+ if (ret != 0) {
+ DEBUG(DEBUG_ERR, (__location__ " ctdb control for transaction_active failed\n"));
+ return -1;
+ }
+
+ return status;
+}
+
+
struct ctdb_transaction_handle {
struct ctdb_db_context *ctdb_db;
bool in_replay;
- /* we store the reads and writes done under a transaction one
- list stores both reads and writes, the other just writes
- */
+ /*
+ * we store the reads and writes done under a transaction:
+ * - one list stores both reads and writes (m_all),
+ * - the other just writes (m_write)
+ */
struct ctdb_marshall_buffer *m_all;
struct ctdb_marshall_buffer *m_write;
};
{
struct ctdb_record_handle *rh;
TDB_DATA key;
+ TDB_DATA data;
struct ctdb_ltdb_header header;
TALLOC_CTX *tmp_ctx;
const char *keyname = CTDB_TRANSACTION_LOCK_KEY;
int ret;
struct ctdb_db_context *ctdb_db = h->ctdb_db;
+ pid_t pid;
+ int32_t status;
key.dptr = discard_const(keyname);
key.dsize = strlen(keyname);
rh = ctdb_fetch_lock(ctdb_db, tmp_ctx, key, NULL);
if (rh == NULL) {
- DEBUG(DEBUG_ERR,(__location__ " Failed to fetch_lock database\n"));
+ DEBUG(DEBUG_ERR,(__location__ " Failed to fetch_lock database\n"));
+ talloc_free(tmp_ctx);
+ return -1;
+ }
+
+ status = ctdb_ctrl_transaction_active(ctdb_db->ctdb,
+ CTDB_CURRENT_NODE,
+ ctdb_db->db_id);
+ if (status == 1) {
+ unsigned long int usec = (1000 + random()) % 100000;
+ DEBUG(DEBUG_DEBUG, (__location__ " transaction is active "
+ "on db_id[0x%08x]. waiting for %lu "
+ "microseconds\n",
+ ctdb_db->db_id, usec));
+ talloc_free(tmp_ctx);
+ usleep(usec);
+ goto again;
+ }
+
+ /*
+ * store the pid in the database:
+ * it is not enough that the node is dmaster...
+ */
+ pid = getpid();
+ data.dptr = (unsigned char *)&pid;
+ data.dsize = sizeof(pid_t);
+ rh->header.rsn++;
+ rh->header.dmaster = ctdb_db->ctdb->pnn;
+ ret = ctdb_ltdb_store(ctdb_db, key, &(rh->header), data);
+ if (ret != 0) {
+ DEBUG(DEBUG_ERR, (__location__ " Failed to store pid in "
+ "transaction record\n"));
talloc_free(tmp_ctx);
return -1;
}
+
talloc_free(rh);
ret = tdb_transaction_start(ctdb_db->ltdb->tdb);
return -1;
}
- ret = ctdb_ltdb_fetch(ctdb_db, key, &header, tmp_ctx, NULL);
- if (ret != 0 || header.dmaster != ctdb_db->ctdb->pnn) {
+ ret = ctdb_ltdb_fetch(ctdb_db, key, &header, tmp_ctx, &data);
+ if (ret != 0) {
+ DEBUG(DEBUG_ERR,(__location__ " Failed to re-fetch transaction "
+ "lock record inside transaction\n"));
+ tdb_transaction_cancel(ctdb_db->ltdb->tdb);
+ talloc_free(tmp_ctx);
+ goto again;
+ }
+
+ if (header.dmaster != ctdb_db->ctdb->pnn) {
+ DEBUG(DEBUG_DEBUG,(__location__ " not dmaster any more on "
+ "transaction lock record\n"));
+ tdb_transaction_cancel(ctdb_db->ltdb->tdb);
+ talloc_free(tmp_ctx);
+ goto again;
+ }
+
+ if ((data.dsize != sizeof(pid_t)) || (*(pid_t *)(data.dptr) != pid)) {
+ DEBUG(DEBUG_DEBUG, (__location__ " my pid is not stored in "
+ "the transaction lock record\n"));
tdb_transaction_cancel(ctdb_db->ltdb->tdb);
talloc_free(tmp_ctx);
goto again;
/* the record doesn't exist - create one with us as dmaster.
This is only safe because we are in a transaction and this
is a persistent database */
- header.dmaster = h->ctdb_db->ctdb->pnn;
- header.rsn = 0;
+ ZERO_STRUCT(header);
} else if (ret != 0) {
DEBUG(DEBUG_ERR,(__location__ " Failed to fetch record\n"));
talloc_free(tmp_ctx);
return 0;
}
+ header.dmaster = h->ctdb_db->ctdb->pnn;
header.rsn++;
if (!h->in_replay) {
&timeout, NULL);
if (ret != 0 || status != 0) {
tdb_transaction_cancel(h->ctdb_db->ltdb->tdb);
+ DEBUG(DEBUG_NOTICE, (__location__ " transaction commit%s failed"
+ ", retrying after 1 second...\n",
+ (retries==0)?"":"retry "));
sleep(1);
if (ret != 0) {
}
}
- if (++retries == 10) {
+ if (++retries == 100) {
DEBUG(DEBUG_ERR,(__location__ " Giving up transaction on db 0x%08x after %d retries failure_control=%u\n",
h->ctdb_db->db_id, retries, (unsigned)failure_control));
ctdb_control(ctdb, CTDB_CURRENT_NODE, h->ctdb_db->db_id,
}
if (ctdb_replay_transaction(h) != 0) {
- DEBUG(DEBUG_ERR,(__location__ " Failed to replay transaction\n"));
+ DEBUG(DEBUG_ERR, (__location__ " Failed to replay "
+ "transaction on db 0x%08x, "
+ "failure control =%u\n",
+ h->ctdb_db->db_id,
+ (unsigned)failure_control));
ctdb_control(ctdb, CTDB_CURRENT_NODE, h->ctdb_db->db_id,
failure_control, CTDB_CTRL_FLAG_NOREPLY,
tdb_null, NULL, NULL, NULL, NULL, NULL);
/* do the real commit locally */
ret = tdb_transaction_commit(h->ctdb_db->ltdb->tdb);
if (ret != 0) {
- DEBUG(DEBUG_ERR,(__location__ " Failed to commit transaction\n"));
+ DEBUG(DEBUG_ERR, (__location__ " Failed to commit transaction "
+ "on db id 0x%08x locally, "
+ "failure_control=%u\n",
+ h->ctdb_db->db_id,
+ (unsigned)failure_control));
ctdb_control(ctdb, CTDB_CURRENT_NODE, h->ctdb_db->db_id,
failure_control, CTDB_CTRL_FLAG_NOREPLY,
tdb_null, NULL, NULL, NULL, NULL, NULL);
return 0;
}
+
+/* when forking the main daemon and the child process needs to connect back
+ * to the daemon as a client process, this function can be used to change
+ * the ctdb context from daemon into client mode
+ */
+int switch_from_server_to_client(struct ctdb_context *ctdb)
+{
+ int ret;
+
+ /* shutdown the transport */
+ if (ctdb->methods) {
+ ctdb->methods->shutdown(ctdb);
+ }
+
+ /* get a new event context */
+ talloc_free(ctdb->ev);
+ ctdb->ev = event_context_init(ctdb);
+
+ close(ctdb->daemon.sd);
+ ctdb->daemon.sd = -1;
+
+ /* initialise ctdb */
+ ret = ctdb_socket_connect(ctdb);
+ if (ret != 0) {
+ DEBUG(DEBUG_ALERT, (__location__ " Failed to init ctdb client\n"));
+ return -1;
+ }
+
+ return 0;
+}
+
+/*
+ get the status of running the monitor eventscripts: NULL means never run.
+ */
+int ctdb_ctrl_getscriptstatus(struct ctdb_context *ctdb,
+ struct timeval timeout, uint32_t destnode,
+ TALLOC_CTX *mem_ctx, enum ctdb_eventscript_call type,
+ struct ctdb_scripts_wire **script_status)
+{
+ int ret;
+ TDB_DATA outdata, indata;
+ int32_t res;
+ uint32_t uinttype = type;
+
+ indata.dptr = (uint8_t *)&uinttype;
+ indata.dsize = sizeof(uinttype);
+
+ ret = ctdb_control(ctdb, destnode, 0,
+ CTDB_CONTROL_GET_EVENT_SCRIPT_STATUS, 0, indata,
+ mem_ctx, &outdata, &res, &timeout, NULL);
+ if (ret != 0 || res != 0) {
+ DEBUG(DEBUG_ERR,(__location__ " ctdb_control for getscriptstatus failed ret:%d res:%d\n", ret, res));
+ return -1;
+ }
+
+ if (outdata.dsize == 0) {
+ *script_status = NULL;
+ } else {
+ *script_status = (struct ctdb_scripts_wire *)talloc_memdup(mem_ctx, outdata.dptr, outdata.dsize);
+ talloc_free(outdata.dptr);
+ }
+
+ return 0;
+}
+
+/*
+ tell the main daemon how long it took to lock the reclock file
+ */
+int ctdb_ctrl_report_recd_lock_latency(struct ctdb_context *ctdb, struct timeval timeout, double latency)
+{
+ int ret;
+ int32_t res;
+ TDB_DATA data;
+
+ data.dptr = (uint8_t *)&latency;
+ data.dsize = sizeof(latency);
+
+ ret = ctdb_control(ctdb, CTDB_CURRENT_NODE, 0, CTDB_CONTROL_RECD_RECLOCK_LATENCY, 0, data,
+ ctdb, NULL, &res, NULL, NULL);
+ if (ret != 0 || res != 0) {
+ DEBUG(DEBUG_ERR,("Failed to send recd reclock latency\n"));
+ return -1;
+ }
+
+ return 0;
+}
+
+/*
+ get the name of the reclock file
+ */
+int ctdb_ctrl_getreclock(struct ctdb_context *ctdb, struct timeval timeout,
+ uint32_t destnode, TALLOC_CTX *mem_ctx,
+ const char **name)
+{
+ int ret;
+ int32_t res;
+ TDB_DATA data;
+
+ ret = ctdb_control(ctdb, destnode, 0,
+ CTDB_CONTROL_GET_RECLOCK_FILE, 0, tdb_null,
+ mem_ctx, &data, &res, &timeout, NULL);
+ if (ret != 0 || res != 0) {
+ return -1;
+ }
+
+ if (data.dsize == 0) {
+ *name = NULL;
+ } else {
+ *name = talloc_strdup(mem_ctx, discard_const(data.dptr));
+ }
+ talloc_free(data.dptr);
+
+ return 0;
+}
+
+/*
+ set the reclock filename for a node
+ */
+int ctdb_ctrl_setreclock(struct ctdb_context *ctdb, struct timeval timeout, uint32_t destnode, const char *reclock)
+{
+ int ret;
+ TDB_DATA data;
+ int32_t res;
+
+ if (reclock == NULL) {
+ data.dsize = 0;
+ data.dptr = NULL;
+ } else {
+ data.dsize = strlen(reclock) + 1;
+ data.dptr = discard_const(reclock);
+ }
+
+ ret = ctdb_control(ctdb, destnode, 0,
+ CTDB_CONTROL_SET_RECLOCK_FILE, 0, data,
+ NULL, NULL, &res, &timeout, NULL);
+ if (ret != 0 || res != 0) {
+ DEBUG(DEBUG_ERR,(__location__ " ctdb_control for setreclock failed\n"));
+ return -1;
+ }
+
+ return 0;
+}
+
+/*
+ stop a node
+ */
+int ctdb_ctrl_stop_node(struct ctdb_context *ctdb, struct timeval timeout, uint32_t destnode)
+{
+ int ret;
+ int32_t res;
+
+ ret = ctdb_control(ctdb, destnode, 0, CTDB_CONTROL_STOP_NODE, 0, tdb_null,
+ ctdb, NULL, &res, &timeout, NULL);
+ if (ret != 0 || res != 0) {
+ DEBUG(DEBUG_ERR,("Failed to stop node\n"));
+ return -1;
+ }
+
+ return 0;
+}
+
+/*
+ continue a node
+ */
+int ctdb_ctrl_continue_node(struct ctdb_context *ctdb, struct timeval timeout, uint32_t destnode)
+{
+ int ret;
+
+ ret = ctdb_control(ctdb, destnode, 0, CTDB_CONTROL_CONTINUE_NODE, 0, tdb_null,
+ ctdb, NULL, NULL, &timeout, NULL);
+ if (ret != 0) {
+ DEBUG(DEBUG_ERR,("Failed to continue node\n"));
+ return -1;
+ }
+
+ return 0;
+}
+
+/*
+ set the natgw state for a node
+ */
+int ctdb_ctrl_setnatgwstate(struct ctdb_context *ctdb, struct timeval timeout, uint32_t destnode, uint32_t natgwstate)
+{
+ int ret;
+ TDB_DATA data;
+ int32_t res;
+
+ data.dsize = sizeof(natgwstate);
+ data.dptr = (uint8_t *)&natgwstate;
+
+ ret = ctdb_control(ctdb, destnode, 0,
+ CTDB_CONTROL_SET_NATGWSTATE, 0, data,
+ NULL, NULL, &res, &timeout, NULL);
+ if (ret != 0 || res != 0) {
+ DEBUG(DEBUG_ERR,(__location__ " ctdb_control for setnatgwstate failed\n"));
+ return -1;
+ }
+
+ return 0;
+}
+
+/*
+ set the lmaster role for a node
+ */
+int ctdb_ctrl_setlmasterrole(struct ctdb_context *ctdb, struct timeval timeout, uint32_t destnode, uint32_t lmasterrole)
+{
+ int ret;
+ TDB_DATA data;
+ int32_t res;
+
+ data.dsize = sizeof(lmasterrole);
+ data.dptr = (uint8_t *)&lmasterrole;
+
+ ret = ctdb_control(ctdb, destnode, 0,
+ CTDB_CONTROL_SET_LMASTERROLE, 0, data,
+ NULL, NULL, &res, &timeout, NULL);
+ if (ret != 0 || res != 0) {
+ DEBUG(DEBUG_ERR,(__location__ " ctdb_control for setlmasterrole failed\n"));
+ return -1;
+ }
+
+ return 0;
+}
+
+/*
+ set the recmaster role for a node
+ */
+int ctdb_ctrl_setrecmasterrole(struct ctdb_context *ctdb, struct timeval timeout, uint32_t destnode, uint32_t recmasterrole)
+{
+ int ret;
+ TDB_DATA data;
+ int32_t res;
+
+ data.dsize = sizeof(recmasterrole);
+ data.dptr = (uint8_t *)&recmasterrole;
+
+ ret = ctdb_control(ctdb, destnode, 0,
+ CTDB_CONTROL_SET_RECMASTERROLE, 0, data,
+ NULL, NULL, &res, &timeout, NULL);
+ if (ret != 0 || res != 0) {
+ DEBUG(DEBUG_ERR,(__location__ " ctdb_control for setrecmasterrole failed\n"));
+ return -1;
+ }
+
+ return 0;
+}
+
+/* enable an eventscript
+ */
+int ctdb_ctrl_enablescript(struct ctdb_context *ctdb, struct timeval timeout, uint32_t destnode, const char *script)
+{
+ int ret;
+ TDB_DATA data;
+ int32_t res;
+
+ data.dsize = strlen(script) + 1;
+ data.dptr = discard_const(script);
+
+ ret = ctdb_control(ctdb, destnode, 0,
+ CTDB_CONTROL_ENABLE_SCRIPT, 0, data,
+ NULL, NULL, &res, &timeout, NULL);
+ if (ret != 0 || res != 0) {
+ DEBUG(DEBUG_ERR,(__location__ " ctdb_control for enablescript failed\n"));
+ return -1;
+ }
+
+ return 0;
+}
+
+/* disable an eventscript
+ */
+int ctdb_ctrl_disablescript(struct ctdb_context *ctdb, struct timeval timeout, uint32_t destnode, const char *script)
+{
+ int ret;
+ TDB_DATA data;
+ int32_t res;
+
+ data.dsize = strlen(script) + 1;
+ data.dptr = discard_const(script);
+
+ ret = ctdb_control(ctdb, destnode, 0,
+ CTDB_CONTROL_DISABLE_SCRIPT, 0, data,
+ NULL, NULL, &res, &timeout, NULL);
+ if (ret != 0 || res != 0) {
+ DEBUG(DEBUG_ERR,(__location__ " ctdb_control for disablescript failed\n"));
+ return -1;
+ }
+
+ return 0;
+}
+
+
+int ctdb_ctrl_set_ban(struct ctdb_context *ctdb, struct timeval timeout, uint32_t destnode, struct ctdb_ban_time *bantime)
+{
+ int ret;
+ TDB_DATA data;
+ int32_t res;
+
+ data.dsize = sizeof(*bantime);
+ data.dptr = (uint8_t *)bantime;
+
+ ret = ctdb_control(ctdb, destnode, 0,
+ CTDB_CONTROL_SET_BAN_STATE, 0, data,
+ NULL, NULL, &res, &timeout, NULL);
+ if (ret != 0 || res != 0) {
+ DEBUG(DEBUG_ERR,(__location__ " ctdb_control for set ban state failed\n"));
+ return -1;
+ }
+
+ return 0;
+}
+
+
+int ctdb_ctrl_get_ban(struct ctdb_context *ctdb, struct timeval timeout, uint32_t destnode, TALLOC_CTX *mem_ctx, struct ctdb_ban_time **bantime)
+{
+ int ret;
+ TDB_DATA outdata;
+ int32_t res;
+ TALLOC_CTX *tmp_ctx = talloc_new(NULL);
+
+ ret = ctdb_control(ctdb, destnode, 0,
+ CTDB_CONTROL_GET_BAN_STATE, 0, tdb_null,
+ tmp_ctx, &outdata, &res, &timeout, NULL);
+ if (ret != 0 || res != 0) {
+ DEBUG(DEBUG_ERR,(__location__ " ctdb_control for set ban state failed\n"));
+ talloc_free(tmp_ctx);
+ return -1;
+ }
+
+ *bantime = (struct ctdb_ban_time *)talloc_steal(mem_ctx, outdata.dptr);
+ talloc_free(tmp_ctx);
+
+ return 0;
+}
+
+
+int ctdb_ctrl_set_db_priority(struct ctdb_context *ctdb, struct timeval timeout, uint32_t destnode, struct ctdb_db_priority *db_prio)
+{
+ int ret;
+ int32_t res;
+ TDB_DATA data;
+ TALLOC_CTX *tmp_ctx = talloc_new(NULL);
+
+ data.dptr = (uint8_t*)db_prio;
+ data.dsize = sizeof(*db_prio);
+
+ ret = ctdb_control(ctdb, destnode, 0,
+ CTDB_CONTROL_SET_DB_PRIORITY, 0, data,
+ tmp_ctx, NULL, &res, &timeout, NULL);
+ if (ret != 0 || res != 0) {
+ DEBUG(DEBUG_ERR,(__location__ " ctdb_control for set_db_priority failed\n"));
+ talloc_free(tmp_ctx);
+ return -1;
+ }
+
+ talloc_free(tmp_ctx);
+
+ return 0;
+}
+
+int ctdb_ctrl_get_db_priority(struct ctdb_context *ctdb, struct timeval timeout, uint32_t destnode, uint32_t db_id, uint32_t *priority)
+{
+ int ret;
+ int32_t res;
+ TDB_DATA data;
+ TALLOC_CTX *tmp_ctx = talloc_new(NULL);
+
+ data.dptr = (uint8_t*)&db_id;
+ data.dsize = sizeof(db_id);
+
+ ret = ctdb_control(ctdb, destnode, 0,
+ CTDB_CONTROL_GET_DB_PRIORITY, 0, data,
+ tmp_ctx, NULL, &res, &timeout, NULL);
+ if (ret != 0 || res < 0) {
+ DEBUG(DEBUG_ERR,(__location__ " ctdb_control for set_db_priority failed\n"));
+ talloc_free(tmp_ctx);
+ return -1;
+ }
+
+ if (priority) {
+ *priority = res;
+ }
+
+ talloc_free(tmp_ctx);
+
+ return 0;
+}