ctdb/server/ctdb_recover.c

   1 /*
   2    ctdb recovery code
   3
   4    Copyright (C) Andrew Tridgell  2007
   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 #include "replace.h"
  21 #include "system/time.h"
  22 #include "system/network.h"
  23 #include "system/filesys.h"
  24 #include "system/wait.h"
  25
  26 #include <talloc.h>
  27 #include <tevent.h>
  28 #include <tdb.h>
  29
  30 #include "lib/tdb_wrap/tdb_wrap.h"
  31 #include "lib/util/dlinklist.h"
  32 #include "lib/util/debug.h"
  33 #include "lib/util/time.h"
  34 #include "lib/util/util_process.h"
  35
  36 #include "ctdb_private.h"
  37 #include "ctdb_client.h"
  38
  39 #include "common/system.h"
  40 #include "common/common.h"
  41 #include "common/logging.h"
  42
  43 #include "ctdb_cluster_mutex.h"
  44
  45 int
  46 ctdb_control_getvnnmap(struct ctdb_context *ctdb, uint32_t opcode, TDB_DATA indata, TDB_DATA *outdata)
  47 {
  48         struct ctdb_vnn_map_wire *map;
  49         size_t len;
  50
  51         CHECK_CONTROL_DATA_SIZE(0);
  52
  53         len = offsetof(struct ctdb_vnn_map_wire, map) + sizeof(uint32_t)*ctdb->vnn_map->size;
  54         map = talloc_size(outdata, len);
  55         CTDB_NO_MEMORY(ctdb, map);
  56
  57         map->generation = ctdb->vnn_map->generation;
  58         map->size = ctdb->vnn_map->size;
  59         memcpy(map->map, ctdb->vnn_map->map, sizeof(uint32_t)*map->size);
  60
  61         outdata->dsize = len;
  62         outdata->dptr  = (uint8_t *)map;
  63
  64         return 0;
  65 }
  66
  67 int
  68 ctdb_control_setvnnmap(struct ctdb_context *ctdb, uint32_t opcode, TDB_DATA indata, TDB_DATA *outdata)
  69 {
  70         struct ctdb_vnn_map_wire *map = (struct ctdb_vnn_map_wire *)indata.dptr;
  71
  72         if (ctdb->recovery_mode != CTDB_RECOVERY_ACTIVE) {
  73                 DEBUG(DEBUG_ERR, ("Attempt to set vnnmap when not in recovery\n"));
  74                 return -1;
  75         }
  76
  77         talloc_free(ctdb->vnn_map);
  78
  79         ctdb->vnn_map = talloc(ctdb, struct ctdb_vnn_map);
  80         CTDB_NO_MEMORY(ctdb, ctdb->vnn_map);
  81
  82         ctdb->vnn_map->generation = map->generation;
  83         ctdb->vnn_map->size       = map->size;
  84         ctdb->vnn_map->map = talloc_array(ctdb->vnn_map, uint32_t, map->size);
  85         CTDB_NO_MEMORY(ctdb, ctdb->vnn_map->map);
  86
  87         memcpy(ctdb->vnn_map->map, map->map, sizeof(uint32_t)*map->size);
  88
  89         return 0;
  90 }
  91
  92 int
  93 ctdb_control_getdbmap(struct ctdb_context *ctdb, uint32_t opcode, TDB_DATA indata, TDB_DATA *outdata)
  94 {
  95         uint32_t i, len;
  96         struct ctdb_db_context *ctdb_db;
  97         struct ctdb_dbid_map_old *dbid_map;
  98
  99         CHECK_CONTROL_DATA_SIZE(0);
 100
 101         len = 0;
 102         for(ctdb_db=ctdb->db_list;ctdb_db;ctdb_db=ctdb_db->next){
 103                 len++;
 104         }
 105
 106
 107         outdata->dsize = offsetof(struct ctdb_dbid_map_old, dbs) + sizeof(dbid_map->dbs[0])*len;
 108         outdata->dptr  = (unsigned char *)talloc_zero_size(outdata, outdata->dsize);
 109         if (!outdata->dptr) {
 110                 DEBUG(DEBUG_ALERT, (__location__ " Failed to allocate dbmap array\n"));
 111                 exit(1);
 112         }
 113
 114         dbid_map = (struct ctdb_dbid_map_old *)outdata->dptr;
 115         dbid_map->num = len;
 116         for (i=0,ctdb_db=ctdb->db_list;ctdb_db;i++,ctdb_db=ctdb_db->next){
 117                 dbid_map->dbs[i].db_id       = ctdb_db->db_id;
 118                 dbid_map->dbs[i].flags       = ctdb_db->db_flags;
 119         }
 120
 121         return 0;
 122 }
 123
 124 int
 125 ctdb_control_getnodemap(struct ctdb_context *ctdb, uint32_t opcode, TDB_DATA indata, TDB_DATA *outdata)
 126 {
 127         CHECK_CONTROL_DATA_SIZE(0);
 128
 129         outdata->dptr  = (unsigned char *)ctdb_node_list_to_map(ctdb->nodes,
 130                                                                 ctdb->num_nodes,
 131                                                                 outdata);
 132         if (outdata->dptr == NULL) {
 133                 return -1;
 134         }
 135
 136         outdata->dsize = talloc_get_size(outdata->dptr);
 137
 138         return 0;
 139 }
 140
 141 /*
 142   reload the nodes file
 143 */
 144 int
 145 ctdb_control_reload_nodes_file(struct ctdb_context *ctdb, uint32_t opcode)
 146 {
 147         int i, num_nodes;
 148         TALLOC_CTX *tmp_ctx;
 149         struct ctdb_node **nodes;
 150
 151         tmp_ctx = talloc_new(ctdb);
 152
 153         /* steal the old nodes file for a while */
 154         talloc_steal(tmp_ctx, ctdb->nodes);
 155         nodes = ctdb->nodes;
 156         ctdb->nodes = NULL;
 157         num_nodes = ctdb->num_nodes;
 158         ctdb->num_nodes = 0;
 159
 160         /* load the new nodes file */
 161         ctdb_load_nodes_file(ctdb);
 162
 163         for (i=0; i<ctdb->num_nodes; i++) {
 164                 /* keep any identical pre-existing nodes and connections */
 165                 if ((i < num_nodes) && ctdb_same_address(&ctdb->nodes[i]->address, &nodes[i]->address)) {
 166                         talloc_free(ctdb->nodes[i]);
 167                         ctdb->nodes[i] = talloc_steal(ctdb->nodes, nodes[i]);
 168                         continue;
 169                 }
 170
 171                 if (ctdb->nodes[i]->flags & NODE_FLAGS_DELETED) {
 172                         continue;
 173                 }
 174
 175                 /* any new or different nodes must be added */
 176                 if (ctdb->methods->add_node(ctdb->nodes[i]) != 0) {
 177                         DEBUG(DEBUG_CRIT, (__location__ " methods->add_node failed at %d\n", i));
 178                         ctdb_fatal(ctdb, "failed to add node. shutting down\n");
 179                 }
 180                 if (ctdb->methods->connect_node(ctdb->nodes[i]) != 0) {
 181                         DEBUG(DEBUG_CRIT, (__location__ " methods->add_connect failed at %d\n", i));
 182                         ctdb_fatal(ctdb, "failed to connect to node. shutting down\n");
 183                 }
 184         }
 185
 186         /* tell the recovery daemon to reaload the nodes file too */
 187         ctdb_daemon_send_message(ctdb, ctdb->pnn, CTDB_SRVID_RELOAD_NODES, tdb_null);
 188
 189         talloc_free(tmp_ctx);
 190
 191         return 0;
 192 }
 193
 194 /*
 195    a traverse function for pulling all relevent records from pulldb
 196  */
 197 struct pulldb_data {
 198         struct ctdb_context *ctdb;
 199         struct ctdb_db_context *ctdb_db;
 200         struct ctdb_marshall_buffer *pulldata;
 201         uint32_t len;
 202         uint32_t allocated_len;
 203         bool failed;
 204 };
 205
 206 static int traverse_pulldb(struct tdb_context *tdb, TDB_DATA key, TDB_DATA data, void *p)
 207 {
 208         struct pulldb_data *params = (struct pulldb_data *)p;
 209         struct ctdb_rec_data_old *rec;
 210         struct ctdb_context *ctdb = params->ctdb;
 211         struct ctdb_db_context *ctdb_db = params->ctdb_db;
 212
 213         /* add the record to the blob */
 214         rec = ctdb_marshall_record(params->pulldata, 0, key, NULL, data);
 215         if (rec == NULL) {
 216                 params->failed = true;
 217                 return -1;
 218         }
 219         if (params->len + rec->length >= params->allocated_len) {
 220                 params->allocated_len = rec->length + params->len + ctdb->tunable.pulldb_preallocation_size;
 221                 params->pulldata = talloc_realloc_size(NULL, params->pulldata, params->allocated_len);
 222         }
 223         if (params->pulldata == NULL) {
 224                 DEBUG(DEBUG_CRIT,(__location__ " Failed to expand pulldb_data to %u\n", rec->length + params->len));
 225                 ctdb_fatal(params->ctdb, "failed to allocate memory for recovery. shutting down\n");
 226         }
 227         params->pulldata->count++;
 228         memcpy(params->len+(uint8_t *)params->pulldata, rec, rec->length);
 229         params->len += rec->length;
 230
 231         if (ctdb->tunable.db_record_size_warn != 0 && rec->length > ctdb->tunable.db_record_size_warn) {
 232                 DEBUG(DEBUG_ERR,("Data record in %s is big. Record size is %d bytes\n", ctdb_db->db_name, (int)rec->length));
 233         }
 234
 235         talloc_free(rec);
 236
 237         return 0;
 238 }
 239
 240 /*
 241   pull a bunch of records from a ltdb, filtering by lmaster
 242  */
 243 int32_t ctdb_control_pull_db(struct ctdb_context *ctdb, TDB_DATA indata, TDB_DATA *outdata)
 244 {
 245         struct ctdb_pulldb *pull;
 246         struct ctdb_db_context *ctdb_db;
 247         struct pulldb_data params;
 248         struct ctdb_marshall_buffer *reply;
 249
 250         pull = (struct ctdb_pulldb *)indata.dptr;
 251
 252         ctdb_db = find_ctdb_db(ctdb, pull->db_id);
 253         if (!ctdb_db) {
 254                 DEBUG(DEBUG_ERR,(__location__ " Unknown db 0x%08x\n", pull->db_id));
 255                 return -1;
 256         }
 257
 258         if (!ctdb_db_frozen(ctdb_db)) {
 259                 DEBUG(DEBUG_ERR,
 260                       ("rejecting ctdb_control_pull_db when not frozen\n"));
 261                 return -1;
 262         }
 263
 264         reply = talloc_zero(outdata, struct ctdb_marshall_buffer);
 265         CTDB_NO_MEMORY(ctdb, reply);
 266
 267         reply->db_id = pull->db_id;
 268
 269         params.ctdb = ctdb;
 270         params.ctdb_db = ctdb_db;
 271         params.pulldata = reply;
 272         params.len = offsetof(struct ctdb_marshall_buffer, data);
 273         params.allocated_len = params.len;
 274         params.failed = false;
 275
 276         if (ctdb_db->unhealthy_reason) {
 277                 /* this is just a warning, as the tdb should be empty anyway */
 278                 DEBUG(DEBUG_WARNING,("db(%s) unhealty in ctdb_control_pull_db: %s\n",
 279                                      ctdb_db->db_name, ctdb_db->unhealthy_reason));
 280         }
 281
 282         /* If the records are invalid, we are done */
 283         if (ctdb_db->invalid_records) {
 284                 goto done;
 285         }
 286
 287         if (ctdb_lockdb_mark(ctdb_db) != 0) {
 288                 DEBUG(DEBUG_ERR,(__location__ " Failed to get lock on entire db - failing\n"));
 289                 return -1;
 290         }
 291
 292         if (tdb_traverse_read(ctdb_db->ltdb->tdb, traverse_pulldb, &params) == -1) {
 293                 DEBUG(DEBUG_ERR,(__location__ " Failed to get traverse db '%s'\n", ctdb_db->db_name));
 294                 ctdb_lockdb_unmark(ctdb_db);
 295                 talloc_free(params.pulldata);
 296                 return -1;
 297         }
 298
 299         ctdb_lockdb_unmark(ctdb_db);
 300
 301 done:
 302         outdata->dptr = (uint8_t *)params.pulldata;
 303         outdata->dsize = params.len;
 304
 305         if (ctdb->tunable.db_record_count_warn != 0 && params.pulldata->count > ctdb->tunable.db_record_count_warn) {
 306                 DEBUG(DEBUG_ERR,("Database %s is big. Contains %d records\n", ctdb_db->db_name, params.pulldata->count));
 307         }
 308         if (ctdb->tunable.db_size_warn != 0 && outdata->dsize > ctdb->tunable.db_size_warn) {
 309                 DEBUG(DEBUG_ERR,("Database %s is big. Contains %d bytes\n", ctdb_db->db_name, (int)outdata->dsize));
 310         }
 311
 312
 313         return 0;
 314 }
 315
 316 struct db_pull_state {
 317         struct ctdb_context *ctdb;
 318         struct ctdb_db_context *ctdb_db;
 319         struct ctdb_marshall_buffer *recs;
 320         uint32_t pnn;
 321         uint64_t srvid;
 322         uint32_t num_records;
 323 };
 324
 325 static int traverse_db_pull(struct tdb_context *tdb, TDB_DATA key,
 326                             TDB_DATA data, void *private_data)
 327 {
 328         struct db_pull_state *state = (struct db_pull_state *)private_data;
 329         struct ctdb_marshall_buffer *recs;
 330
 331         recs = ctdb_marshall_add(state->ctdb, state->recs,
 332                                  state->ctdb_db->db_id, 0, key, NULL, data);
 333         if (recs == NULL) {
 334                 TALLOC_FREE(state->recs);
 335                 return -1;
 336         }
 337         state->recs = recs;
 338
 339         if (talloc_get_size(state->recs) >=
 340                         state->ctdb->tunable.rec_buffer_size_limit) {
 341                 TDB_DATA buffer;
 342                 int ret;
 343
 344                 buffer = ctdb_marshall_finish(state->recs);
 345                 ret = ctdb_daemon_send_message(state->ctdb, state->pnn,
 346                                                state->srvid, buffer);
 347                 if (ret != 0) {
 348                         TALLOC_FREE(state->recs);
 349                         return -1;
 350                 }
 351
 352                 state->num_records += state->recs->count;
 353                 TALLOC_FREE(state->recs);
 354         }
 355
 356         return 0;
 357 }
 358
 359 int32_t ctdb_control_db_pull(struct ctdb_context *ctdb,
 360                              struct ctdb_req_control_old *c,
 361                              TDB_DATA indata, TDB_DATA *outdata)
 362 {
 363         struct ctdb_pulldb_ext *pulldb_ext;
 364         struct ctdb_db_context *ctdb_db;
 365         struct db_pull_state state;
 366         int ret;
 367
 368         pulldb_ext = (struct ctdb_pulldb_ext *)indata.dptr;
 369
 370         ctdb_db = find_ctdb_db(ctdb, pulldb_ext->db_id);
 371         if (ctdb_db == NULL) {
 372                 DEBUG(DEBUG_ERR,(__location__ " Unknown db 0x%08x\n",
 373                                  pulldb_ext->db_id));
 374                 return -1;
 375         }
 376
 377         if (!ctdb_db_frozen(ctdb_db)) {
 378                 DEBUG(DEBUG_ERR,
 379                       ("rejecting ctdb_control_pull_db when not frozen\n"));
 380                 return -1;
 381         }
 382
 383         if (ctdb_db->unhealthy_reason) {
 384                 /* this is just a warning, as the tdb should be empty anyway */
 385                 DEBUG(DEBUG_WARNING,
 386                       ("db(%s) unhealty in ctdb_control_db_pull: %s\n",
 387                        ctdb_db->db_name, ctdb_db->unhealthy_reason));
 388         }
 389
 390         state.ctdb = ctdb;
 391         state.ctdb_db = ctdb_db;
 392         state.recs = NULL;
 393         state.pnn = c->hdr.srcnode;
 394         state.srvid = pulldb_ext->srvid;
 395         state.num_records = 0;
 396
 397         /* If the records are invalid, we are done */
 398         if (ctdb_db->invalid_records) {
 399                 goto done;
 400         }
 401
 402         if (ctdb_lockdb_mark(ctdb_db) != 0) {
 403                 DEBUG(DEBUG_ERR,
 404                       (__location__ " Failed to get lock on entire db - failing\n"));
 405                 return -1;
 406         }
 407
 408         ret = tdb_traverse_read(ctdb_db->ltdb->tdb, traverse_db_pull, &state);
 409         if (ret == -1) {
 410                 DEBUG(DEBUG_ERR,
 411                       (__location__ " Failed to get traverse db '%s'\n",
 412                        ctdb_db->db_name));
 413                 ctdb_lockdb_unmark(ctdb_db);
 414                 return -1;
 415         }
 416
 417         /* Last few records */
 418         if (state.recs != NULL) {
 419                 TDB_DATA buffer;
 420
 421                 buffer = ctdb_marshall_finish(state.recs);
 422                 ret = ctdb_daemon_send_message(state.ctdb, state.pnn,
 423                                                state.srvid, buffer);
 424                 if (ret != 0) {
 425                         TALLOC_FREE(state.recs);
 426                         ctdb_lockdb_unmark(ctdb_db);
 427                         return -1;
 428                 }
 429
 430                 state.num_records += state.recs->count;
 431                 TALLOC_FREE(state.recs);
 432         }
 433
 434         ctdb_lockdb_unmark(ctdb_db);
 435
 436 done:
 437         outdata->dptr = talloc_size(outdata, sizeof(uint32_t));
 438         if (outdata->dptr == NULL) {
 439                 DEBUG(DEBUG_ERR, (__location__ " Memory allocation error\n"));
 440                 return -1;
 441         }
 442
 443         memcpy(outdata->dptr, (uint8_t *)&state.num_records, sizeof(uint32_t));
 444         outdata->dsize = sizeof(uint32_t);
 445
 446         return 0;
 447 }
 448
 449 /*
 450   push a bunch of records into a ltdb, filtering by rsn
 451  */
 452 int32_t ctdb_control_push_db(struct ctdb_context *ctdb, TDB_DATA indata)
 453 {
 454         struct ctdb_marshall_buffer *reply = (struct ctdb_marshall_buffer *)indata.dptr;
 455         struct ctdb_db_context *ctdb_db;
 456         int i, ret;
 457         struct ctdb_rec_data_old *rec;
 458
 459         if (indata.dsize < offsetof(struct ctdb_marshall_buffer, data)) {
 460                 DEBUG(DEBUG_ERR,(__location__ " invalid data in pulldb reply\n"));
 461                 return -1;
 462         }
 463
 464         ctdb_db = find_ctdb_db(ctdb, reply->db_id);
 465         if (!ctdb_db) {
 466                 DEBUG(DEBUG_ERR,(__location__ " Unknown db 0x%08x\n", reply->db_id));
 467                 return -1;
 468         }
 469
 470         if (!ctdb_db_frozen(ctdb_db)) {
 471                 DEBUG(DEBUG_ERR,
 472                       ("rejecting ctdb_control_push_db when not frozen\n"));
 473                 return -1;
 474         }
 475
 476         if (ctdb_lockdb_mark(ctdb_db) != 0) {
 477                 DEBUG(DEBUG_ERR,(__location__ " Failed to get lock on entire db - failing\n"));
 478                 return -1;
 479         }
 480
 481         rec = (struct ctdb_rec_data_old *)&reply->data[0];
 482
 483         DEBUG(DEBUG_INFO,("starting push of %u records for dbid 0x%x\n",
 484                  reply->count, reply->db_id));
 485
 486         for (i=0;i<reply->count;i++) {
 487                 TDB_DATA key, data;
 488                 struct ctdb_ltdb_header *hdr;
 489
 490                 key.dptr = &rec->data[0];
 491                 key.dsize = rec->keylen;
 492                 data.dptr = &rec->data[key.dsize];
 493                 data.dsize = rec->datalen;
 494
 495                 if (data.dsize < sizeof(struct ctdb_ltdb_header)) {
 496                         DEBUG(DEBUG_CRIT,(__location__ " bad ltdb record\n"));
 497                         goto failed;
 498                 }
 499                 hdr = (struct ctdb_ltdb_header *)data.dptr;
 500                 /* strip off any read only record flags. All readonly records
 501                    are revoked implicitely by a recovery
 502                 */
 503                 hdr->flags &= ~CTDB_REC_RO_FLAGS;
 504
 505                 data.dptr += sizeof(*hdr);
 506                 data.dsize -= sizeof(*hdr);
 507
 508                 ret = ctdb_ltdb_store(ctdb_db, key, hdr, data);
 509                 if (ret != 0) {
 510                         DEBUG(DEBUG_CRIT, (__location__ " Unable to store record\n"));
 511                         goto failed;
 512                 }
 513
 514                 rec = (struct ctdb_rec_data_old *)(rec->length + (uint8_t *)rec);
 515         }
 516
 517         DEBUG(DEBUG_DEBUG,("finished push of %u records for dbid 0x%x\n",
 518                  reply->count, reply->db_id));
 519
 520         if (ctdb_db_readonly(ctdb_db)) {
 521                 DEBUG(DEBUG_CRIT,("Clearing the tracking database for dbid 0x%x\n",
 522                                   ctdb_db->db_id));
 523                 if (tdb_wipe_all(ctdb_db->rottdb) != 0) {
 524                         DEBUG(DEBUG_ERR,("Failed to wipe tracking database for 0x%x. Dropping read-only delegation support\n", ctdb_db->db_id));
 525                         tdb_close(ctdb_db->rottdb);
 526                         ctdb_db->rottdb = NULL;
 527                         ctdb_db_reset_readonly(ctdb_db);
 528                 }
 529                 while (ctdb_db->revokechild_active != NULL) {
 530                         talloc_free(ctdb_db->revokechild_active);
 531                 }
 532         }
 533
 534         ctdb_lockdb_unmark(ctdb_db);
 535         return 0;
 536
 537 failed:
 538         ctdb_lockdb_unmark(ctdb_db);
 539         return -1;
 540 }
 541
 542 struct db_push_state {
 543         struct ctdb_context *ctdb;
 544         struct ctdb_db_context *ctdb_db;
 545         uint64_t srvid;
 546         uint32_t num_records;
 547         bool failed;
 548 };
 549
 550 static void db_push_msg_handler(uint64_t srvid, TDB_DATA indata,
 551                                 void *private_data)
 552 {
 553         struct db_push_state *state = talloc_get_type(
 554                 private_data, struct db_push_state);
 555         struct ctdb_marshall_buffer *recs;
 556         struct ctdb_rec_data_old *rec;
 557         int i, ret;
 558
 559         if (state->failed) {
 560                 return;
 561         }
 562
 563         recs = (struct ctdb_marshall_buffer *)indata.dptr;
 564         rec = (struct ctdb_rec_data_old *)&recs->data[0];
 565
 566         DEBUG(DEBUG_INFO, ("starting push of %u records for dbid 0x%x\n",
 567                            recs->count, recs->db_id));
 568
 569         for (i=0; i<recs->count; i++) {
 570                 TDB_DATA key, data;
 571                 struct ctdb_ltdb_header *hdr;
 572
 573                 key.dptr = &rec->data[0];
 574                 key.dsize = rec->keylen;
 575                 data.dptr = &rec->data[key.dsize];
 576                 data.dsize = rec->datalen;
 577
 578                 if (data.dsize < sizeof(struct ctdb_ltdb_header)) {
 579                         DEBUG(DEBUG_CRIT,(__location__ " bad ltdb record\n"));
 580                         goto failed;
 581                 }
 582
 583                 hdr = (struct ctdb_ltdb_header *)data.dptr;
 584                 /* Strip off any read only record flags.
 585                  * All readonly records are revoked implicitely by a recovery.
 586                  */
 587                 hdr->flags &= ~CTDB_REC_RO_FLAGS;
 588
 589                 data.dptr += sizeof(*hdr);
 590                 data.dsize -= sizeof(*hdr);
 591
 592                 ret = ctdb_ltdb_store(state->ctdb_db, key, hdr, data);
 593                 if (ret != 0) {
 594                         DEBUG(DEBUG_ERR,
 595                               (__location__ " Unable to store record\n"));
 596                         goto failed;
 597                 }
 598
 599                 rec = (struct ctdb_rec_data_old *)(rec->length + (uint8_t *)rec);
 600         }
 601
 602         DEBUG(DEBUG_DEBUG, ("finished push of %u records for dbid 0x%x\n",
 603                             recs->count, recs->db_id));
 604
 605         state->num_records += recs->count;
 606         return;
 607
 608 failed:
 609         state->failed = true;
 610 }
 611
 612 int32_t ctdb_control_db_push_start(struct ctdb_context *ctdb, TDB_DATA indata)
 613 {
 614         struct ctdb_pulldb_ext *pulldb_ext;
 615         struct ctdb_db_context *ctdb_db;
 616         struct db_push_state *state;
 617         int ret;
 618
 619         pulldb_ext = (struct ctdb_pulldb_ext *)indata.dptr;
 620
 621         ctdb_db = find_ctdb_db(ctdb, pulldb_ext->db_id);
 622         if (ctdb_db == NULL) {
 623                 DEBUG(DEBUG_ERR,
 624                       (__location__ " Unknown db 0x%08x\n", pulldb_ext->db_id));
 625                 return -1;
 626         }
 627
 628         if (!ctdb_db_frozen(ctdb_db)) {
 629                 DEBUG(DEBUG_ERR,
 630                       ("rejecting ctdb_control_db_push_start when not frozen\n"));
 631                 return -1;
 632         }
 633
 634         if (ctdb_db->push_started) {
 635                 DEBUG(DEBUG_WARNING,
 636                       (__location__ " DB push already started for %s\n",
 637                        ctdb_db->db_name));
 638
 639                 /* De-register old state */
 640                 state = (struct db_push_state *)ctdb_db->push_state;
 641                 if (state != NULL) {
 642                         srvid_deregister(ctdb->srv, state->srvid, state);
 643                         talloc_free(state);
 644                         ctdb_db->push_state = NULL;
 645                 }
 646         }
 647
 648         state = talloc_zero(ctdb_db, struct db_push_state);
 649         if (state == NULL) {
 650                 DEBUG(DEBUG_ERR, (__location__ " Memory allocation error\n"));
 651                 return -1;
 652         }
 653
 654         state->ctdb = ctdb;
 655         state->ctdb_db = ctdb_db;
 656         state->srvid = pulldb_ext->srvid;
 657         state->failed = false;
 658
 659         ret = srvid_register(ctdb->srv, state, state->srvid,
 660                              db_push_msg_handler, state);
 661         if (ret != 0) {
 662                 DEBUG(DEBUG_ERR,
 663                       (__location__ " Failed to register srvid for db push\n"));
 664                 talloc_free(state);
 665                 return -1;
 666         }
 667
 668         if (ctdb_lockdb_mark(ctdb_db) != 0) {
 669                 DEBUG(DEBUG_ERR,
 670                       (__location__ " Failed to get lock on entire db - failing\n"));
 671                 srvid_deregister(ctdb->srv, state->srvid, state);
 672                 talloc_free(state);
 673                 return -1;
 674         }
 675
 676         ctdb_db->push_started = true;
 677         ctdb_db->push_state = state;
 678
 679         return 0;
 680 }
 681
 682 int32_t ctdb_control_db_push_confirm(struct ctdb_context *ctdb,
 683                                      TDB_DATA indata, TDB_DATA *outdata)
 684 {
 685         uint32_t db_id;
 686         struct ctdb_db_context *ctdb_db;
 687         struct db_push_state *state;
 688
 689         db_id = *(uint32_t *)indata.dptr;
 690
 691         ctdb_db = find_ctdb_db(ctdb, db_id);
 692         if (ctdb_db == NULL) {
 693                 DEBUG(DEBUG_ERR,(__location__ " Unknown db 0x%08x\n", db_id));
 694                 return -1;
 695         }
 696
 697         if (!ctdb_db_frozen(ctdb_db)) {
 698                 DEBUG(DEBUG_ERR,
 699                       ("rejecting ctdb_control_db_push_confirm when not frozen\n"));
 700                 return -1;
 701         }
 702
 703         if (!ctdb_db->push_started) {
 704                 DEBUG(DEBUG_ERR, (__location__ " DB push not started\n"));
 705                 return -1;
 706         }
 707
 708         if (ctdb_db_readonly(ctdb_db)) {
 709                 DEBUG(DEBUG_ERR,
 710                       ("Clearing the tracking database for dbid 0x%x\n",
 711                        ctdb_db->db_id));
 712                 if (tdb_wipe_all(ctdb_db->rottdb) != 0) {
 713                         DEBUG(DEBUG_ERR,
 714                               ("Failed to wipe tracking database for 0x%x."
 715                                " Dropping read-only delegation support\n",
 716                                ctdb_db->db_id));
 717                         tdb_close(ctdb_db->rottdb);
 718                         ctdb_db->rottdb = NULL;
 719                         ctdb_db_reset_readonly(ctdb_db);
 720                 }
 721
 722                 while (ctdb_db->revokechild_active != NULL) {
 723                         talloc_free(ctdb_db->revokechild_active);
 724                 }
 725         }
 726
 727         ctdb_lockdb_unmark(ctdb_db);
 728
 729         state = (struct db_push_state *)ctdb_db->push_state;
 730         if (state == NULL) {
 731                 DEBUG(DEBUG_ERR, (__location__ " Missing push db state\n"));
 732                 return -1;
 733         }
 734
 735         srvid_deregister(ctdb->srv, state->srvid, state);
 736
 737         outdata->dptr = talloc_size(outdata, sizeof(uint32_t));
 738         if (outdata->dptr == NULL) {
 739                 DEBUG(DEBUG_ERR, (__location__ " Memory allocation error\n"));
 740                 talloc_free(state);
 741                 ctdb_db->push_state = NULL;
 742                 return -1;
 743         }
 744
 745         memcpy(outdata->dptr, (uint8_t *)&state->num_records, sizeof(uint32_t));
 746         outdata->dsize = sizeof(uint32_t);
 747
 748         talloc_free(state);
 749         ctdb_db->push_started = false;
 750         ctdb_db->push_state = NULL;
 751
 752         return 0;
 753 }
 754
 755 struct set_recmode_state {
 756         struct ctdb_context *ctdb;
 757         struct ctdb_req_control_old *c;
 758 };
 759
 760 static void set_recmode_handler(char status,
 761                                 double latency,
 762                                 void *private_data)
 763 {
 764         struct set_recmode_state *state = talloc_get_type_abort(
 765                 private_data, struct set_recmode_state);
 766         int s = 0;
 767         const char *err = NULL;
 768
 769         switch (status) {
 770         case '0':
 771                 /* Mutex taken */
 772                 DEBUG(DEBUG_ERR,
 773                       ("ERROR: Daemon able to take recovery lock on \"%s\" during recovery\n",
 774                        state->ctdb->recovery_lock));
 775                 s = -1;
 776                 err = "Took recovery lock from daemon during recovery - probably a cluster filesystem lock coherence problem";
 777                 break;
 778
 779         case '1':
 780                 /* Contention */
 781                 DEBUG(DEBUG_DEBUG, (__location__ " Recovery lock check OK\n"));
 782                 state->ctdb->recovery_mode = CTDB_RECOVERY_NORMAL;
 783                 ctdb_process_deferred_attach(state->ctdb);
 784
 785                 s = 0;
 786
 787                 CTDB_UPDATE_RECLOCK_LATENCY(state->ctdb, "daemon reclock",
 788                                             reclock.ctdbd, latency);
 789                 break;
 790
 791         case '2':
 792                 /* Timeout.  Consider this a success, not a failure,
 793                  * as we failed to set the recovery lock which is what
 794                  * we wanted.  This can be caused by the cluster
 795                  * filesystem being very slow to arbitrate locks
 796                  * immediately after a node failure. */
 797                 DEBUG(DEBUG_WARNING,
 798                       (__location__
 799                        "Time out getting recovery lock, allowing recmode set anyway\n"));
 800                 state->ctdb->recovery_mode = CTDB_RECOVERY_NORMAL;
 801                 ctdb_process_deferred_attach(state->ctdb);
 802
 803                 s = 0;
 804                 break;
 805
 806         default:
 807                 DEBUG(DEBUG_ERR,
 808                       ("Unexpected error when testing recovery lock\n"));
 809                 s = -1;
 810                 err = "Unexpected error when testing recovery lock";
 811         }
 812
 813         ctdb_request_control_reply(state->ctdb, state->c, NULL, s, err);
 814         talloc_free(state);
 815 }
 816
 817 static void
 818 ctdb_drop_all_ips_event(struct tevent_context *ev, struct tevent_timer *te,
 819                         struct timeval t, void *private_data)
 820 {
 821         struct ctdb_context *ctdb = talloc_get_type(private_data, struct ctdb_context);
 822
 823         DEBUG(DEBUG_ERR,(__location__ " Been in recovery mode for too long. Dropping all IPS\n"));
 824         talloc_free(ctdb->release_ips_ctx);
 825         ctdb->release_ips_ctx = NULL;
 826
 827         ctdb_release_all_ips(ctdb);
 828 }
 829
 830 /*
 831  * Set up an event to drop all public ips if we remain in recovery for too
 832  * long
 833  */
 834 int ctdb_deferred_drop_all_ips(struct ctdb_context *ctdb)
 835 {
 836         if (ctdb->release_ips_ctx != NULL) {
 837                 talloc_free(ctdb->release_ips_ctx);
 838         }
 839         ctdb->release_ips_ctx = talloc_new(ctdb);
 840         CTDB_NO_MEMORY(ctdb, ctdb->release_ips_ctx);
 841
 842         tevent_add_timer(ctdb->ev, ctdb->release_ips_ctx,
 843                          timeval_current_ofs(ctdb->tunable.recovery_drop_all_ips, 0),
 844                          ctdb_drop_all_ips_event, ctdb);
 845         return 0;
 846 }
 847
 848 /*
 849   set the recovery mode
 850  */
 851 int32_t ctdb_control_set_recmode(struct ctdb_context *ctdb,
 852                                  struct ctdb_req_control_old *c,
 853                                  TDB_DATA indata, bool *async_reply,
 854                                  const char **errormsg)
 855 {
 856         uint32_t recmode = *(uint32_t *)indata.dptr;
 857         struct ctdb_db_context *ctdb_db;
 858         struct set_recmode_state *state;
 859         struct ctdb_cluster_mutex_handle *h;
 860
 861         if (recmode == ctdb->recovery_mode) {
 862                 D_INFO("Recovery mode already set to %s\n",
 863                        recmode == CTDB_RECOVERY_NORMAL ? "NORMAL" : "ACTIVE");
 864                 return 0;
 865         }
 866
 867         D_NOTICE("Recovery mode set to %s\n",
 868                  recmode == CTDB_RECOVERY_NORMAL ? "NORMAL" : "ACTIVE");
 869
 870         /* if we enter recovery but stay in recovery for too long
 871            we will eventually drop all our ip addresses
 872         */
 873         if (recmode == CTDB_RECOVERY_ACTIVE) {
 874                 if (ctdb_deferred_drop_all_ips(ctdb) != 0) {
 875                         D_ERR("Failed to set up deferred drop all ips\n");
 876                 }
 877
 878                 ctdb->recovery_mode = CTDB_RECOVERY_ACTIVE;
 879                 return 0;
 880         }
 881
 882         /* From this point: recmode == CTDB_RECOVERY_NORMAL
 883          *
 884          * Therefore, what follows is special handling when setting
 885          * recovery mode back to normal */
 886
 887         TALLOC_FREE(ctdb->release_ips_ctx);
 888
 889         for (ctdb_db = ctdb->db_list; ctdb_db != NULL; ctdb_db = ctdb_db->next) {
 890                 if (ctdb_db->generation != ctdb->vnn_map->generation) {
 891                         DEBUG(DEBUG_ERR,
 892                               ("Inconsistent DB generation %u for %s\n",
 893                                ctdb_db->generation, ctdb_db->db_name));
 894                         DEBUG(DEBUG_ERR, ("Recovery mode set to ACTIVE\n"));
 895                         return -1;
 896                 }
 897         }
 898
 899         /* force the databases to thaw */
 900         if (ctdb_db_all_frozen(ctdb)) {
 901                 ctdb_control_thaw(ctdb, false);
 902         }
 903
 904         if (ctdb->recovery_lock == NULL) {
 905                 /* Not using recovery lock file */
 906                 ctdb->recovery_mode = CTDB_RECOVERY_NORMAL;
 907                 ctdb_process_deferred_attach(ctdb);
 908                 return 0;
 909         }
 910
 911         state = talloc_zero(ctdb, struct set_recmode_state);
 912         if (state == NULL) {
 913                 DEBUG(DEBUG_ERR, (__location__ " out of memory\n"));
 914                 return -1;
 915         }
 916         state->ctdb = ctdb;
 917         state->c = NULL;
 918
 919         h = ctdb_cluster_mutex(state, ctdb, ctdb->recovery_lock, 5,
 920                                set_recmode_handler, state, NULL, NULL);
 921         if (h == NULL) {
 922                 talloc_free(state);
 923                 return -1;
 924         }
 925
 926         state->c = talloc_steal(state, c);
 927         *async_reply = true;
 928
 929         return 0;
 930 }
 931
 932
 933 /*
 934   delete a record as part of the vacuum process
 935   only delete if we are not lmaster or dmaster, and our rsn is <= the provided rsn
 936   use non-blocking locks
 937
 938   return 0 if the record was successfully deleted (i.e. it does not exist
 939   when the function returns)
 940   or !0 is the record still exists in the tdb after returning.
 941  */
 942 static int delete_tdb_record(struct ctdb_context *ctdb, struct ctdb_db_context *ctdb_db, struct ctdb_rec_data_old *rec)
 943 {
 944         TDB_DATA key, data, data2;
 945         struct ctdb_ltdb_header *hdr, *hdr2;
 946
 947         /* these are really internal tdb functions - but we need them here for
 948            non-blocking lock of the freelist */
 949         int tdb_lock_nonblock(struct tdb_context *tdb, int list, int ltype);
 950         int tdb_unlock(struct tdb_context *tdb, int list, int ltype);
 951
 952
 953         key.dsize = rec->keylen;
 954         key.dptr  = &rec->data[0];
 955         data.dsize = rec->datalen;
 956         data.dptr = &rec->data[rec->keylen];
 957
 958         if (ctdb_lmaster(ctdb, &key) == ctdb->pnn) {
 959                 DEBUG(DEBUG_INFO,(__location__ " Called delete on record where we are lmaster\n"));
 960                 return -1;
 961         }
 962
 963         if (data.dsize != sizeof(struct ctdb_ltdb_header)) {
 964                 DEBUG(DEBUG_ERR,(__location__ " Bad record size\n"));
 965                 return -1;
 966         }
 967
 968         hdr = (struct ctdb_ltdb_header *)data.dptr;
 969
 970         /* use a non-blocking lock */
 971         if (tdb_chainlock_nonblock(ctdb_db->ltdb->tdb, key) != 0) {
 972                 return -1;
 973         }
 974
 975         data2 = tdb_fetch(ctdb_db->ltdb->tdb, key);
 976         if (data2.dptr == NULL) {
 977                 tdb_chainunlock(ctdb_db->ltdb->tdb, key);
 978                 return 0;
 979         }
 980
 981         if (data2.dsize < sizeof(struct ctdb_ltdb_header)) {
 982                 if (tdb_lock_nonblock(ctdb_db->ltdb->tdb, -1, F_WRLCK) == 0) {
 983                         if (tdb_delete(ctdb_db->ltdb->tdb, key) != 0) {
 984                                 DEBUG(DEBUG_CRIT,(__location__ " Failed to delete corrupt record\n"));
 985                         }
 986                         tdb_unlock(ctdb_db->ltdb->tdb, -1, F_WRLCK);
 987                         DEBUG(DEBUG_CRIT,(__location__ " Deleted corrupt record\n"));
 988                 }
 989                 tdb_chainunlock(ctdb_db->ltdb->tdb, key);
 990                 free(data2.dptr);
 991                 return 0;
 992         }
 993
 994         hdr2 = (struct ctdb_ltdb_header *)data2.dptr;
 995
 996         if (hdr2->rsn > hdr->rsn) {
 997                 tdb_chainunlock(ctdb_db->ltdb->tdb, key);
 998                 DEBUG(DEBUG_INFO,(__location__ " Skipping record with rsn=%llu - called with rsn=%llu\n",
 999                          (unsigned long long)hdr2->rsn, (unsigned long long)hdr->rsn));
1000                 free(data2.dptr);
1001                 return -1;
1002         }
1003
1004         /* do not allow deleting record that have readonly flags set. */
1005         if (hdr->flags & CTDB_REC_RO_FLAGS) {
1006                 tdb_chainunlock(ctdb_db->ltdb->tdb, key);
1007                 DEBUG(DEBUG_INFO,(__location__ " Skipping record with readonly flags set\n"));
1008                 free(data2.dptr);
1009                 return -1;
1010         }
1011         if (hdr2->flags & CTDB_REC_RO_FLAGS) {
1012                 tdb_chainunlock(ctdb_db->ltdb->tdb, key);
1013                 DEBUG(DEBUG_INFO,(__location__ " Skipping record with readonly flags set\n"));
1014                 free(data2.dptr);
1015                 return -1;
1016         }
1017
1018         if (hdr2->dmaster == ctdb->pnn) {
1019                 tdb_chainunlock(ctdb_db->ltdb->tdb, key);
1020                 DEBUG(DEBUG_INFO,(__location__ " Attempted delete record where we are the dmaster\n"));
1021                 free(data2.dptr);
1022                 return -1;
1023         }
1024
1025         if (tdb_lock_nonblock(ctdb_db->ltdb->tdb, -1, F_WRLCK) != 0) {
1026                 tdb_chainunlock(ctdb_db->ltdb->tdb, key);
1027                 free(data2.dptr);
1028                 return -1;
1029         }
1030
1031         if (tdb_delete(ctdb_db->ltdb->tdb, key) != 0) {
1032                 tdb_unlock(ctdb_db->ltdb->tdb, -1, F_WRLCK);
1033                 tdb_chainunlock(ctdb_db->ltdb->tdb, key);
1034                 DEBUG(DEBUG_INFO,(__location__ " Failed to delete record\n"));
1035                 free(data2.dptr);
1036                 return -1;
1037         }
1038
1039         tdb_unlock(ctdb_db->ltdb->tdb, -1, F_WRLCK);
1040         tdb_chainunlock(ctdb_db->ltdb->tdb, key);
1041         free(data2.dptr);
1042         return 0;
1043 }
1044
1045
1046
1047 struct recovery_callback_state {
1048         struct ctdb_req_control_old *c;
1049 };
1050
1051
1052 /*
1053   called when the 'recovered' event script has finished
1054  */
1055 static void ctdb_end_recovery_callback(struct ctdb_context *ctdb, int status, void *p)
1056 {
1057         struct recovery_callback_state *state = talloc_get_type(p, struct recovery_callback_state);
1058
1059         CTDB_INCREMENT_STAT(ctdb, num_recoveries);
1060
1061         if (status != 0) {
1062                 DEBUG(DEBUG_ERR,(__location__ " recovered event script failed (status %d)\n", status));
1063                 if (status == -ETIMEDOUT) {
1064                         ctdb_ban_self(ctdb);
1065                 }
1066         }
1067
1068         ctdb_request_control_reply(ctdb, state->c, NULL, status, NULL);
1069         talloc_free(state);
1070
1071         gettimeofday(&ctdb->last_recovery_finished, NULL);
1072
1073         if (ctdb->runstate == CTDB_RUNSTATE_FIRST_RECOVERY) {
1074                 ctdb_set_runstate(ctdb, CTDB_RUNSTATE_STARTUP);
1075         }
1076 }
1077
1078 /*
1079   recovery has finished
1080  */
1081 int32_t ctdb_control_end_recovery(struct ctdb_context *ctdb,
1082                                 struct ctdb_req_control_old *c,
1083                                 bool *async_reply)
1084 {
1085         int ret;
1086         struct recovery_callback_state *state;
1087
1088         DEBUG(DEBUG_ERR,("Recovery has finished\n"));
1089
1090         ctdb_persistent_finish_trans3_commits(ctdb);
1091
1092         state = talloc(ctdb, struct recovery_callback_state);
1093         CTDB_NO_MEMORY(ctdb, state);
1094
1095         state->c    = c;
1096
1097         ret = ctdb_event_script_callback(ctdb, state,
1098                                          ctdb_end_recovery_callback,
1099                                          state,
1100                                          CTDB_EVENT_RECOVERED, "%s", "");
1101
1102         if (ret != 0) {
1103                 DEBUG(DEBUG_ERR,(__location__ " Failed to end recovery\n"));
1104                 talloc_free(state);
1105                 return -1;
1106         }
1107
1108         /* tell the control that we will be reply asynchronously */
1109         state->c    = talloc_steal(state, c);
1110         *async_reply = true;
1111         return 0;
1112 }
1113
1114 /*
1115   called when the 'startrecovery' event script has finished
1116  */
1117 static void ctdb_start_recovery_callback(struct ctdb_context *ctdb, int status, void *p)
1118 {
1119         struct recovery_callback_state *state = talloc_get_type(p, struct recovery_callback_state);
1120
1121         if (status != 0) {
1122                 DEBUG(DEBUG_ERR,(__location__ " startrecovery event script failed (status %d)\n", status));
1123         }
1124
1125         ctdb_request_control_reply(ctdb, state->c, NULL, status, NULL);
1126         talloc_free(state);
1127 }
1128
1129 static void run_start_recovery_event(struct ctdb_context *ctdb,
1130                                      struct recovery_callback_state *state)
1131 {
1132         int ret;
1133
1134         ret = ctdb_event_script_callback(ctdb, state,
1135                                          ctdb_start_recovery_callback,
1136                                          state,
1137                                          CTDB_EVENT_START_RECOVERY,
1138                                          "%s", "");
1139
1140         if (ret != 0) {
1141                 DEBUG(DEBUG_ERR,("Unable to run startrecovery event\n"));
1142                 ctdb_request_control_reply(ctdb, state->c, NULL, -1, NULL);
1143                 talloc_free(state);
1144                 return;
1145         }
1146
1147         return;
1148 }
1149
1150 static bool reclock_strings_equal(const char *a, const char *b)
1151 {
1152         return (a == NULL && b == NULL) ||
1153                 (a != NULL && b != NULL && strcmp(a, b) == 0);
1154 }
1155
1156 static void start_recovery_reclock_callback(struct ctdb_context *ctdb,
1157                                                 int32_t status,
1158                                                 TDB_DATA data,
1159                                                 const char *errormsg,
1160                                                 void *private_data)
1161 {
1162         struct recovery_callback_state *state = talloc_get_type_abort(
1163                 private_data, struct recovery_callback_state);
1164         const char *local = ctdb->recovery_lock;
1165         const char *remote = NULL;
1166
1167         if (status != 0) {
1168                 DEBUG(DEBUG_ERR, (__location__ " GET_RECLOCK failed\n"));
1169                 ctdb_request_control_reply(ctdb, state->c, NULL,
1170                                            status, errormsg);
1171                 talloc_free(state);
1172                 return;
1173         }
1174
1175         /* Check reclock consistency */
1176         if (data.dsize > 0) {
1177                 /* Ensure NUL-termination */
1178                 data.dptr[data.dsize-1] = '\0';
1179                 remote = (const char *)data.dptr;
1180         }
1181         if (! reclock_strings_equal(local, remote)) {
1182                 /* Inconsistent */
1183                 ctdb_request_control_reply(ctdb, state->c, NULL, -1, NULL);
1184                 DEBUG(DEBUG_ERR,
1185                       ("Recovery lock configuration inconsistent: "
1186                        "recmaster has %s, this node has %s, shutting down\n",
1187                        remote == NULL ? "NULL" : remote,
1188                        local == NULL ? "NULL" : local));
1189                 talloc_free(state);
1190                 ctdb_shutdown_sequence(ctdb, 1);
1191         }
1192         DEBUG(DEBUG_INFO,
1193               ("Recovery lock consistency check successful\n"));
1194
1195         run_start_recovery_event(ctdb, state);
1196 }
1197
1198 /* Check recovery lock consistency and run eventscripts for the
1199  * "startrecovery" event */
1200 int32_t ctdb_control_start_recovery(struct ctdb_context *ctdb,
1201                                     struct ctdb_req_control_old *c,
1202                                     bool *async_reply)
1203 {
1204         int ret;
1205         struct recovery_callback_state *state;
1206         uint32_t recmaster = c->hdr.srcnode;
1207
1208         DEBUG(DEBUG_ERR, ("Recovery has started\n"));
1209         gettimeofday(&ctdb->last_recovery_started, NULL);
1210
1211         state = talloc(ctdb, struct recovery_callback_state);
1212         CTDB_NO_MEMORY(ctdb, state);
1213
1214         state->c = c;
1215
1216         /* Although the recovery master sent this node a start
1217          * recovery control, this node might still think the recovery
1218          * master is disconnected.  In this case defer the recovery
1219          * lock consistency check. */
1220         if (ctdb->nodes[recmaster]->flags & NODE_FLAGS_DISCONNECTED) {
1221                 run_start_recovery_event(ctdb, state);
1222         } else {
1223                 /* Ask the recovery master about its reclock setting */
1224                 ret = ctdb_daemon_send_control(ctdb,
1225                                                recmaster,
1226                                                0,
1227                                                CTDB_CONTROL_GET_RECLOCK_FILE,
1228                                                0, 0,
1229                                                tdb_null,
1230                                                start_recovery_reclock_callback,
1231                                                state);
1232
1233                 if (ret != 0) {
1234                         DEBUG(DEBUG_ERR, (__location__ " GET_RECLOCK failed\n"));
1235                         talloc_free(state);
1236                         return -1;
1237                 }
1238         }
1239
1240         /* tell the control that we will be reply asynchronously */
1241         state->c = talloc_steal(state, c);
1242         *async_reply = true;
1243
1244         return 0;
1245 }
1246
1247 /*
1248  try to delete all these records as part of the vacuuming process
1249  and return the records we failed to delete
1250 */
1251 int32_t ctdb_control_try_delete_records(struct ctdb_context *ctdb, TDB_DATA indata, TDB_DATA *outdata)
1252 {
1253         struct ctdb_marshall_buffer *reply = (struct ctdb_marshall_buffer *)indata.dptr;
1254         struct ctdb_db_context *ctdb_db;
1255         int i;
1256         struct ctdb_rec_data_old *rec;
1257         struct ctdb_marshall_buffer *records;
1258
1259         if (indata.dsize < offsetof(struct ctdb_marshall_buffer, data)) {
1260                 DEBUG(DEBUG_ERR,(__location__ " invalid data in try_delete_records\n"));
1261                 return -1;
1262         }
1263
1264         ctdb_db = find_ctdb_db(ctdb, reply->db_id);
1265         if (!ctdb_db) {
1266                 DEBUG(DEBUG_ERR,(__location__ " Unknown db 0x%08x\n", reply->db_id));
1267                 return -1;
1268         }
1269
1270
1271         DEBUG(DEBUG_DEBUG,("starting try_delete_records of %u records for dbid 0x%x\n",
1272                  reply->count, reply->db_id));
1273
1274
1275         /* create a blob to send back the records we couldnt delete */
1276         records = (struct ctdb_marshall_buffer *)
1277                         talloc_zero_size(outdata,
1278                                     offsetof(struct ctdb_marshall_buffer, data));
1279         if (records == NULL) {
1280                 DEBUG(DEBUG_ERR,(__location__ " Out of memory\n"));
1281                 return -1;
1282         }
1283         records->db_id = ctdb_db->db_id;
1284
1285
1286         rec = (struct ctdb_rec_data_old *)&reply->data[0];
1287         for (i=0;i<reply->count;i++) {
1288                 TDB_DATA key, data;
1289
1290                 key.dptr = &rec->data[0];
1291                 key.dsize = rec->keylen;
1292                 data.dptr = &rec->data[key.dsize];
1293                 data.dsize = rec->datalen;
1294
1295                 if (data.dsize < sizeof(struct ctdb_ltdb_header)) {
1296                         DEBUG(DEBUG_CRIT,(__location__ " bad ltdb record in indata\n"));
1297                         talloc_free(records);
1298                         return -1;
1299                 }
1300
1301                 /* If we cant delete the record we must add it to the reply
1302                    so the lmaster knows it may not purge this record
1303                 */
1304                 if (delete_tdb_record(ctdb, ctdb_db, rec) != 0) {
1305                         size_t old_size;
1306                         struct ctdb_ltdb_header *hdr;
1307
1308                         hdr = (struct ctdb_ltdb_header *)data.dptr;
1309                         data.dptr += sizeof(*hdr);
1310                         data.dsize -= sizeof(*hdr);
1311
1312                         DEBUG(DEBUG_INFO, (__location__ " Failed to vacuum delete record with hash 0x%08x\n", ctdb_hash(&key)));
1313
1314                         old_size = talloc_get_size(records);
1315                         records = talloc_realloc_size(outdata, records, old_size + rec->length);
1316                         if (records == NULL) {
1317                                 DEBUG(DEBUG_ERR,(__location__ " Failed to expand\n"));
1318                                 return -1;
1319                         }
1320                         records->count++;
1321                         memcpy(old_size+(uint8_t *)records, rec, rec->length);
1322                 }
1323
1324                 rec = (struct ctdb_rec_data_old *)(rec->length + (uint8_t *)rec);
1325         }
1326
1327
1328         *outdata = ctdb_marshall_finish(records);
1329
1330         return 0;
1331 }
1332
1333 /*
1334   report capabilities
1335  */
1336 int32_t ctdb_control_get_capabilities(struct ctdb_context *ctdb, TDB_DATA *outdata)
1337 {
1338         uint32_t *capabilities = NULL;
1339
1340         capabilities = talloc(outdata, uint32_t);
1341         CTDB_NO_MEMORY(ctdb, capabilities);
1342         *capabilities = ctdb->capabilities;
1343
1344         outdata->dsize = sizeof(uint32_t);
1345         outdata->dptr = (uint8_t *)capabilities;
1346
1347         return 0;
1348 }
1349
1350 /* The recovery daemon will ping us at regular intervals.
1351    If we havent been pinged for a while we assume the recovery
1352    daemon is inoperable and we restart.
1353 */
1354 static void ctdb_recd_ping_timeout(struct tevent_context *ev,
1355                                    struct tevent_timer *te,
1356                                    struct timeval t, void *p)
1357 {
1358         struct ctdb_context *ctdb = talloc_get_type(p, struct ctdb_context);
1359         uint32_t *count = talloc_get_type(ctdb->recd_ping_count, uint32_t);
1360
1361         DEBUG(DEBUG_ERR, ("Recovery daemon ping timeout. Count : %u\n", *count));
1362
1363         if (*count < ctdb->tunable.recd_ping_failcount) {
1364                 (*count)++;
1365                 tevent_add_timer(ctdb->ev, ctdb->recd_ping_count,
1366                                  timeval_current_ofs(ctdb->tunable.recd_ping_timeout, 0),
1367                                  ctdb_recd_ping_timeout, ctdb);
1368                 return;
1369         }
1370
1371         DEBUG(DEBUG_ERR, ("Final timeout for recovery daemon ping. Restarting recovery daemon. (This can be caused if the cluster filesystem has hung)\n"));
1372
1373         ctdb_stop_recoverd(ctdb);
1374         ctdb_start_recoverd(ctdb);
1375 }
1376
1377 int32_t ctdb_control_recd_ping(struct ctdb_context *ctdb)
1378 {
1379         talloc_free(ctdb->recd_ping_count);
1380
1381         ctdb->recd_ping_count = talloc_zero(ctdb, uint32_t);
1382         CTDB_NO_MEMORY(ctdb, ctdb->recd_ping_count);
1383
1384         if (ctdb->tunable.recd_ping_timeout != 0) {
1385                 tevent_add_timer(ctdb->ev, ctdb->recd_ping_count,
1386                                  timeval_current_ofs(ctdb->tunable.recd_ping_timeout, 0),
1387                                  ctdb_recd_ping_timeout, ctdb);
1388         }
1389
1390         return 0;
1391 }
1392
1393
1394
1395 int32_t ctdb_control_set_recmaster(struct ctdb_context *ctdb, uint32_t opcode, TDB_DATA indata)
1396 {
1397         uint32_t new_recmaster;
1398
1399         CHECK_CONTROL_DATA_SIZE(sizeof(uint32_t));
1400         new_recmaster = ((uint32_t *)(&indata.dptr[0]))[0];
1401
1402         if (ctdb->pnn != new_recmaster && ctdb->recovery_master == ctdb->pnn) {
1403                 DEBUG(DEBUG_ERR,
1404                       ("Remote node (%u) is now the recovery master\n",
1405                        new_recmaster));
1406         }
1407
1408         if (ctdb->pnn == new_recmaster && ctdb->recovery_master != new_recmaster) {
1409                 DEBUG(DEBUG_ERR,
1410                       ("This node (%u) is now the recovery master\n",
1411                        ctdb->pnn));
1412         }
1413
1414         ctdb->recovery_master = new_recmaster;
1415         return 0;
1416 }
1417
1418
1419 int32_t ctdb_control_stop_node(struct ctdb_context *ctdb)
1420 {
1421         DEBUG(DEBUG_ERR, ("Stopping node\n"));
1422         ctdb->nodes[ctdb->pnn]->flags |= NODE_FLAGS_STOPPED;
1423
1424         return 0;
1425 }
1426
1427 int32_t ctdb_control_continue_node(struct ctdb_context *ctdb)
1428 {
1429         DEBUG(DEBUG_ERR, ("Continue node\n"));
1430         ctdb->nodes[ctdb->pnn]->flags &= ~NODE_FLAGS_STOPPED;
1431
1432         return 0;
1433 }
1434