?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
ctdb-daemon: Fix CID 1363233 Resource leak (RESOURCE_LEAK)
[Samba.git] / ctdb / server / ctdb_recover.c
blobb42f00173109f7d87f04bdb2723c5ab81b4c305e
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 if (ctdb_db->persistent != 0) {
119 dbid_map->dbs[i].flags |= CTDB_DB_FLAGS_PERSISTENT;
120 }
121 if (ctdb_db->readonly != 0) {
122 dbid_map->dbs[i].flags |= CTDB_DB_FLAGS_READONLY;
123 }
124 if (ctdb_db->sticky != 0) {
125 dbid_map->dbs[i].flags |= CTDB_DB_FLAGS_STICKY;
126 }
127 }
128
129 return 0;
130 }
131
132 int
133 ctdb_control_getnodemap(struct ctdb_context *ctdb, uint32_t opcode, TDB_DATA indata, TDB_DATA *outdata)
134 {
135 CHECK_CONTROL_DATA_SIZE(0);
136
137 outdata->dptr = (unsigned char *)ctdb_node_list_to_map(ctdb->nodes,
138 ctdb->num_nodes,
139 outdata);
140 if (outdata->dptr == NULL) {
141 return -1;
142 }
143
144 outdata->dsize = talloc_get_size(outdata->dptr);
145
146 return 0;
147 }
148
149 /*
150 reload the nodes file
151 */
152 int
153 ctdb_control_reload_nodes_file(struct ctdb_context *ctdb, uint32_t opcode)
154 {
155 int i, num_nodes;
156 TALLOC_CTX *tmp_ctx;
157 struct ctdb_node **nodes;
158
159 tmp_ctx = talloc_new(ctdb);
160
161 /* steal the old nodes file for a while */
162 talloc_steal(tmp_ctx, ctdb->nodes);
163 nodes = ctdb->nodes;
164 ctdb->nodes = NULL;
165 num_nodes = ctdb->num_nodes;
166 ctdb->num_nodes = 0;
167
168 /* load the new nodes file */
169 ctdb_load_nodes_file(ctdb);
170
171 for (i=0; i<ctdb->num_nodes; i++) {
172 /* keep any identical pre-existing nodes and connections */
173 if ((i < num_nodes) && ctdb_same_address(&ctdb->nodes[i]->address, &nodes[i]->address)) {
174 talloc_free(ctdb->nodes[i]);
175 ctdb->nodes[i] = talloc_steal(ctdb->nodes, nodes[i]);
176 continue;
177 }
178
179 if (ctdb->nodes[i]->flags & NODE_FLAGS_DELETED) {
180 continue;
181 }
182
183 /* any new or different nodes must be added */
184 if (ctdb->methods->add_node(ctdb->nodes[i]) != 0) {
185 DEBUG(DEBUG_CRIT, (__location__ " methods->add_node failed at %d\n", i));
186 ctdb_fatal(ctdb, "failed to add node. shutting down\n");
187 }
188 if (ctdb->methods->connect_node(ctdb->nodes[i]) != 0) {
189 DEBUG(DEBUG_CRIT, (__location__ " methods->add_connect failed at %d\n", i));
190 ctdb_fatal(ctdb, "failed to connect to node. shutting down\n");
191 }
192 }
193
194 /* tell the recovery daemon to reaload the nodes file too */
195 ctdb_daemon_send_message(ctdb, ctdb->pnn, CTDB_SRVID_RELOAD_NODES, tdb_null);
196
197 talloc_free(tmp_ctx);
198
199 return 0;
200 }
201
202 /*
203 a traverse function for pulling all relevent records from pulldb
204 */
205 struct pulldb_data {
206 struct ctdb_context *ctdb;
207 struct ctdb_db_context *ctdb_db;
208 struct ctdb_marshall_buffer *pulldata;
209 uint32_t len;
210 uint32_t allocated_len;
211 bool failed;
212 };
213
214 static int traverse_pulldb(struct tdb_context *tdb, TDB_DATA key, TDB_DATA data, void *p)
215 {
216 struct pulldb_data *params = (struct pulldb_data *)p;
217 struct ctdb_rec_data_old *rec;
218 struct ctdb_context *ctdb = params->ctdb;
219 struct ctdb_db_context *ctdb_db = params->ctdb_db;
220
221 /* add the record to the blob */
222 rec = ctdb_marshall_record(params->pulldata, 0, key, NULL, data);
223 if (rec == NULL) {
224 params->failed = true;
225 return -1;
226 }
227 if (params->len + rec->length >= params->allocated_len) {
228 params->allocated_len = rec->length + params->len + ctdb->tunable.pulldb_preallocation_size;
229 params->pulldata = talloc_realloc_size(NULL, params->pulldata, params->allocated_len);
230 }
231 if (params->pulldata == NULL) {
232 DEBUG(DEBUG_CRIT,(__location__ " Failed to expand pulldb_data to %u\n", rec->length + params->len));
233 ctdb_fatal(params->ctdb, "failed to allocate memory for recovery. shutting down\n");
234 }
235 params->pulldata->count++;
236 memcpy(params->len+(uint8_t *)params->pulldata, rec, rec->length);
237 params->len += rec->length;
238
239 if (ctdb->tunable.db_record_size_warn != 0 && rec->length > ctdb->tunable.db_record_size_warn) {
240 DEBUG(DEBUG_ERR,("Data record in %s is big. Record size is %d bytes\n", ctdb_db->db_name, (int)rec->length));
241 }
242
243 talloc_free(rec);
244
245 return 0;
246 }
247
248 /*
249 pull a bunch of records from a ltdb, filtering by lmaster
250 */
251 int32_t ctdb_control_pull_db(struct ctdb_context *ctdb, TDB_DATA indata, TDB_DATA *outdata)
252 {
253 struct ctdb_pulldb *pull;
254 struct ctdb_db_context *ctdb_db;
255 struct pulldb_data params;
256 struct ctdb_marshall_buffer *reply;
257
258 pull = (struct ctdb_pulldb *)indata.dptr;
259
260 ctdb_db = find_ctdb_db(ctdb, pull->db_id);
261 if (!ctdb_db) {
262 DEBUG(DEBUG_ERR,(__location__ " Unknown db 0x%08x\n", pull->db_id));
263 return -1;
264 }
265
266 if (!ctdb_db_frozen(ctdb_db)) {
267 DEBUG(DEBUG_ERR,
268 ("rejecting ctdb_control_pull_db when not frozen\n"));
269 return -1;
270 }
271
272 reply = talloc_zero(outdata, struct ctdb_marshall_buffer);
273 CTDB_NO_MEMORY(ctdb, reply);
274
275 reply->db_id = pull->db_id;
276
277 params.ctdb = ctdb;
278 params.ctdb_db = ctdb_db;
279 params.pulldata = reply;
280 params.len = offsetof(struct ctdb_marshall_buffer, data);
281 params.allocated_len = params.len;
282 params.failed = false;
283
284 if (ctdb_db->unhealthy_reason) {
285 /* this is just a warning, as the tdb should be empty anyway */
286 DEBUG(DEBUG_WARNING,("db(%s) unhealty in ctdb_control_pull_db: %s\n",
287 ctdb_db->db_name, ctdb_db->unhealthy_reason));
288 }
289
290 if (ctdb_lockdb_mark(ctdb_db) != 0) {
291 DEBUG(DEBUG_ERR,(__location__ " Failed to get lock on entire db - failing\n"));
292 return -1;
293 }
294
295 if (tdb_traverse_read(ctdb_db->ltdb->tdb, traverse_pulldb, &params) == -1) {
296 DEBUG(DEBUG_ERR,(__location__ " Failed to get traverse db '%s'\n", ctdb_db->db_name));
297 ctdb_lockdb_unmark(ctdb_db);
298 talloc_free(params.pulldata);
299 return -1;
300 }
301
302 ctdb_lockdb_unmark(ctdb_db);
303
304 outdata->dptr = (uint8_t *)params.pulldata;
305 outdata->dsize = params.len;
306
307 if (ctdb->tunable.db_record_count_warn != 0 && params.pulldata->count > ctdb->tunable.db_record_count_warn) {
308 DEBUG(DEBUG_ERR,("Database %s is big. Contains %d records\n", ctdb_db->db_name, params.pulldata->count));
309 }
310 if (ctdb->tunable.db_size_warn != 0 && outdata->dsize > ctdb->tunable.db_size_warn) {
311 DEBUG(DEBUG_ERR,("Database %s is big. Contains %d bytes\n", ctdb_db->db_name, (int)outdata->dsize));
312 }
313
314
315 return 0;
316 }
317
318 struct db_pull_state {
319 struct ctdb_context *ctdb;
320 struct ctdb_db_context *ctdb_db;
321 struct ctdb_marshall_buffer *recs;
322 uint32_t pnn;
323 uint64_t srvid;
324 uint32_t num_records;
325 };
326
327 static int traverse_db_pull(struct tdb_context *tdb, TDB_DATA key,
328 TDB_DATA data, void *private_data)
329 {
330 struct db_pull_state *state = (struct db_pull_state *)private_data;
331 struct ctdb_marshall_buffer *recs;
332
333 recs = ctdb_marshall_add(state->ctdb, state->recs,
334 state->ctdb_db->db_id, 0, key, NULL, data);
335 if (recs == NULL) {
336 TALLOC_FREE(state->recs);
337 return -1;
338 }
339 state->recs = recs;
340
341 if (talloc_get_size(state->recs) >=
342 state->ctdb->tunable.rec_buffer_size_limit) {
343 TDB_DATA buffer;
344 int ret;
345
346 buffer = ctdb_marshall_finish(state->recs);
347 ret = ctdb_daemon_send_message(state->ctdb, state->pnn,
348 state->srvid, buffer);
349 if (ret != 0) {
350 TALLOC_FREE(state->recs);
351 return -1;
352 }
353
354 state->num_records += state->recs->count;
355 TALLOC_FREE(state->recs);
356 }
357
358 return 0;
359 }
360
361 int32_t ctdb_control_db_pull(struct ctdb_context *ctdb,
362 struct ctdb_req_control_old *c,
363 TDB_DATA indata, TDB_DATA *outdata)
364 {
365 struct ctdb_pulldb_ext *pulldb_ext;
366 struct ctdb_db_context *ctdb_db;
367 struct db_pull_state state;
368 int ret;
369
370 pulldb_ext = (struct ctdb_pulldb_ext *)indata.dptr;
371
372 ctdb_db = find_ctdb_db(ctdb, pulldb_ext->db_id);
373 if (ctdb_db == NULL) {
374 DEBUG(DEBUG_ERR,(__location__ " Unknown db 0x%08x\n",
375 pulldb_ext->db_id));
376 return -1;
377 }
378
379 if (!ctdb_db_frozen(ctdb_db)) {
380 DEBUG(DEBUG_ERR,
381 ("rejecting ctdb_control_pull_db when not frozen\n"));
382 return -1;
383 }
384
385 if (ctdb_db->unhealthy_reason) {
386 /* this is just a warning, as the tdb should be empty anyway */
387 DEBUG(DEBUG_WARNING,
388 ("db(%s) unhealty in ctdb_control_db_pull: %s\n",
389 ctdb_db->db_name, ctdb_db->unhealthy_reason));
390 }
391
392 state.ctdb = ctdb;
393 state.ctdb_db = ctdb_db;
394 state.recs = NULL;
395 state.pnn = c->hdr.srcnode;
396 state.srvid = pulldb_ext->srvid;
397 state.num_records = 0;
398
399 if (ctdb_lockdb_mark(ctdb_db) != 0) {
400 DEBUG(DEBUG_ERR,
401 (__location__ " Failed to get lock on entire db - failing\n"));
402 return -1;
403 }
404
405 ret = tdb_traverse_read(ctdb_db->ltdb->tdb, traverse_db_pull, &state);
406 if (ret == -1) {
407 DEBUG(DEBUG_ERR,
408 (__location__ " Failed to get traverse db '%s'\n",
409 ctdb_db->db_name));
410 ctdb_lockdb_unmark(ctdb_db);
411 return -1;
412 }
413
414 /* Last few records */
415 if (state.recs != NULL) {
416 TDB_DATA buffer;
417
418 buffer = ctdb_marshall_finish(state.recs);
419 ret = ctdb_daemon_send_message(state.ctdb, state.pnn,
420 state.srvid, buffer);
421 if (ret != 0) {
422 TALLOC_FREE(state.recs);
423 ctdb_lockdb_unmark(ctdb_db);
424 return -1;
425 }
426
427 state.num_records += state.recs->count;
428 TALLOC_FREE(state.recs);
429 }
430
431 ctdb_lockdb_unmark(ctdb_db);
432
433 outdata->dptr = talloc_size(outdata, sizeof(uint32_t));
434 if (outdata->dptr == NULL) {
435 DEBUG(DEBUG_ERR, (__location__ " Memory allocation error\n"));
436 return -1;
437 }
438
439 memcpy(outdata->dptr, (uint8_t *)&state.num_records, sizeof(uint32_t));
440 outdata->dsize = sizeof(uint32_t);
441
442 return 0;
443 }
444
445 /*
446 push a bunch of records into a ltdb, filtering by rsn
447 */
448 int32_t ctdb_control_push_db(struct ctdb_context *ctdb, TDB_DATA indata)
449 {
450 struct ctdb_marshall_buffer *reply = (struct ctdb_marshall_buffer *)indata.dptr;
451 struct ctdb_db_context *ctdb_db;
452 int i, ret;
453 struct ctdb_rec_data_old *rec;
454
455 if (indata.dsize < offsetof(struct ctdb_marshall_buffer, data)) {
456 DEBUG(DEBUG_ERR,(__location__ " invalid data in pulldb reply\n"));
457 return -1;
458 }
459
460 ctdb_db = find_ctdb_db(ctdb, reply->db_id);
461 if (!ctdb_db) {
462 DEBUG(DEBUG_ERR,(__location__ " Unknown db 0x%08x\n", reply->db_id));
463 return -1;
464 }
465
466 if (!ctdb_db_frozen(ctdb_db)) {
467 DEBUG(DEBUG_ERR,
468 ("rejecting ctdb_control_push_db when not frozen\n"));
469 return -1;
470 }
471
472 if (ctdb_lockdb_mark(ctdb_db) != 0) {
473 DEBUG(DEBUG_ERR,(__location__ " Failed to get lock on entire db - failing\n"));
474 return -1;
475 }
476
477 rec = (struct ctdb_rec_data_old *)&reply->data[0];
478
479 DEBUG(DEBUG_INFO,("starting push of %u records for dbid 0x%x\n",
480 reply->count, reply->db_id));
481
482 for (i=0;i<reply->count;i++) {
483 TDB_DATA key, data;
484 struct ctdb_ltdb_header *hdr;
485
486 key.dptr = &rec->data[0];
487 key.dsize = rec->keylen;
488 data.dptr = &rec->data[key.dsize];
489 data.dsize = rec->datalen;
490
491 if (data.dsize < sizeof(struct ctdb_ltdb_header)) {
492 DEBUG(DEBUG_CRIT,(__location__ " bad ltdb record\n"));
493 goto failed;
494 }
495 hdr = (struct ctdb_ltdb_header *)data.dptr;
496 /* strip off any read only record flags. All readonly records
497 are revoked implicitely by a recovery
498 */
499 hdr->flags &= ~CTDB_REC_RO_FLAGS;
500
501 data.dptr += sizeof(*hdr);
502 data.dsize -= sizeof(*hdr);
503
504 ret = ctdb_ltdb_store(ctdb_db, key, hdr, data);
505 if (ret != 0) {
506 DEBUG(DEBUG_CRIT, (__location__ " Unable to store record\n"));
507 goto failed;
508 }
509
510 rec = (struct ctdb_rec_data_old *)(rec->length + (uint8_t *)rec);
511 }
512
513 DEBUG(DEBUG_DEBUG,("finished push of %u records for dbid 0x%x\n",
514 reply->count, reply->db_id));
515
516 if (ctdb_db->readonly) {
517 DEBUG(DEBUG_CRIT,("Clearing the tracking database for dbid 0x%x\n",
518 ctdb_db->db_id));
519 if (tdb_wipe_all(ctdb_db->rottdb) != 0) {
520 DEBUG(DEBUG_ERR,("Failed to wipe tracking database for 0x%x. Dropping read-only delegation support\n", ctdb_db->db_id));
521 ctdb_db->readonly = false;
522 tdb_close(ctdb_db->rottdb);
523 ctdb_db->rottdb = NULL;
524 ctdb_db->readonly = false;
525 }
526 while (ctdb_db->revokechild_active != NULL) {
527 talloc_free(ctdb_db->revokechild_active);
528 }
529 }
530
531 ctdb_lockdb_unmark(ctdb_db);
532 return 0;
533
534 failed:
535 ctdb_lockdb_unmark(ctdb_db);
536 return -1;
537 }
538
539 struct db_push_state {
540 struct ctdb_context *ctdb;
541 struct ctdb_db_context *ctdb_db;
542 uint64_t srvid;
543 uint32_t num_records;
544 bool failed;
545 };
546
547 static void db_push_msg_handler(uint64_t srvid, TDB_DATA indata,
548 void *private_data)
549 {
550 struct db_push_state *state = talloc_get_type(
551 private_data, struct db_push_state);
552 struct ctdb_marshall_buffer *recs;
553 struct ctdb_rec_data_old *rec;
554 int i, ret;
555
556 if (state->failed) {
557 return;
558 }
559
560 recs = (struct ctdb_marshall_buffer *)indata.dptr;
561 rec = (struct ctdb_rec_data_old *)&recs->data[0];
562
563 DEBUG(DEBUG_INFO, ("starting push of %u records for dbid 0x%x\n",
564 recs->count, recs->db_id));
565
566 for (i=0; i<recs->count; i++) {
567 TDB_DATA key, data;
568 struct ctdb_ltdb_header *hdr;
569
570 key.dptr = &rec->data[0];
571 key.dsize = rec->keylen;
572 data.dptr = &rec->data[key.dsize];
573 data.dsize = rec->datalen;
574
575 if (data.dsize < sizeof(struct ctdb_ltdb_header)) {
576 DEBUG(DEBUG_CRIT,(__location__ " bad ltdb record\n"));
577 goto failed;
578 }
579
580 hdr = (struct ctdb_ltdb_header *)data.dptr;
581 /* Strip off any read only record flags.
582 * All readonly records are revoked implicitely by a recovery.
583 */
584 hdr->flags &= ~CTDB_REC_RO_FLAGS;
585
586 data.dptr += sizeof(*hdr);
587 data.dsize -= sizeof(*hdr);
588
589 ret = ctdb_ltdb_store(state->ctdb_db, key, hdr, data);
590 if (ret != 0) {
591 DEBUG(DEBUG_ERR,
592 (__location__ " Unable to store record\n"));
593 goto failed;
594 }
595
596 rec = (struct ctdb_rec_data_old *)(rec->length + (uint8_t *)rec);
597 }
598
599 DEBUG(DEBUG_DEBUG, ("finished push of %u records for dbid 0x%x\n",
600 recs->count, recs->db_id));
601
602 state->num_records += recs->count;
603 return;
604
605 failed:
606 state->failed = true;
607 }
608
609 int32_t ctdb_control_db_push_start(struct ctdb_context *ctdb, TDB_DATA indata)
610 {
611 struct ctdb_pulldb_ext *pulldb_ext;
612 struct ctdb_db_context *ctdb_db;
613 struct db_push_state *state;
614 int ret;
615
616 pulldb_ext = (struct ctdb_pulldb_ext *)indata.dptr;
617
618 ctdb_db = find_ctdb_db(ctdb, pulldb_ext->db_id);
619 if (ctdb_db == NULL) {
620 DEBUG(DEBUG_ERR,
621 (__location__ " Unknown db 0x%08x\n", pulldb_ext->db_id));
622 return -1;
623 }
624
625 if (!ctdb_db_frozen(ctdb_db)) {
626 DEBUG(DEBUG_ERR,
627 ("rejecting ctdb_control_db_push_start when not frozen\n"));
628 return -1;
629 }
630
631 if (ctdb_db->push_started) {
632 DEBUG(DEBUG_WARNING,
633 (__location__ " DB push already started for %s\n",
634 ctdb_db->db_name));
635
636 /* De-register old state */
637 state = (struct db_push_state *)ctdb_db->push_state;
638 if (state != NULL) {
639 srvid_deregister(ctdb->srv, state->srvid, state);
640 talloc_free(state);
641 ctdb_db->push_state = NULL;
642 }
643 }
644
645 state = talloc_zero(ctdb_db, struct db_push_state);
646 if (state == NULL) {
647 DEBUG(DEBUG_ERR, (__location__ " Memory allocation error\n"));
648 return -1;
649 }
650
651 state->ctdb = ctdb;
652 state->ctdb_db = ctdb_db;
653 state->srvid = pulldb_ext->srvid;
654 state->failed = false;
655
656 ret = srvid_register(ctdb->srv, state, state->srvid,
657 db_push_msg_handler, state);
658 if (ret != 0) {
659 DEBUG(DEBUG_ERR,
660 (__location__ " Failed to register srvid for db push\n"));
661 talloc_free(state);
662 return -1;
663 }
664
665 if (ctdb_lockdb_mark(ctdb_db) != 0) {
666 DEBUG(DEBUG_ERR,
667 (__location__ " Failed to get lock on entire db - failing\n"));
668 srvid_deregister(ctdb->srv, state->srvid, state);
669 talloc_free(state);
670 return -1;
671 }
672
673 ctdb_db->push_started = true;
674 ctdb_db->push_state = state;
675
676 return 0;
677 }
678
679 int32_t ctdb_control_db_push_confirm(struct ctdb_context *ctdb,
680 TDB_DATA indata, TDB_DATA *outdata)
681 {
682 uint32_t db_id;
683 struct ctdb_db_context *ctdb_db;
684 struct db_push_state *state;
685
686 db_id = *(uint32_t *)indata.dptr;
687
688 ctdb_db = find_ctdb_db(ctdb, db_id);
689 if (ctdb_db == NULL) {
690 DEBUG(DEBUG_ERR,(__location__ " Unknown db 0x%08x\n", db_id));
691 return -1;
692 }
693
694 if (!ctdb_db_frozen(ctdb_db)) {
695 DEBUG(DEBUG_ERR,
696 ("rejecting ctdb_control_db_push_confirm when not frozen\n"));
697 return -1;
698 }
699
700 if (!ctdb_db->push_started) {
701 DEBUG(DEBUG_ERR, (__location__ " DB push not started\n"));
702 return -1;
703 }
704
705 if (ctdb_db->readonly) {
706 DEBUG(DEBUG_ERR,
707 ("Clearing the tracking database for dbid 0x%x\n",
708 ctdb_db->db_id));
709 if (tdb_wipe_all(ctdb_db->rottdb) != 0) {
710 DEBUG(DEBUG_ERR,
711 ("Failed to wipe tracking database for 0x%x."
712 " Dropping read-only delegation support\n",
713 ctdb_db->db_id));
714 ctdb_db->readonly = false;
715 tdb_close(ctdb_db->rottdb);
716 ctdb_db->rottdb = NULL;
717 ctdb_db->readonly = false;
718 }
719
720 while (ctdb_db->revokechild_active != NULL) {
721 talloc_free(ctdb_db->revokechild_active);
722 }
723 }
724
725 ctdb_lockdb_unmark(ctdb_db);
726
727 state = (struct db_push_state *)ctdb_db->push_state;
728 if (state == NULL) {
729 DEBUG(DEBUG_ERR, (__location__ " Missing push db state\n"));
730 return -1;
731 }
732
733 srvid_deregister(ctdb->srv, state->srvid, state);
734
735 outdata->dptr = talloc_size(outdata, sizeof(uint32_t));
736 if (outdata->dptr == NULL) {
737 DEBUG(DEBUG_ERR, (__location__ " Memory allocation error\n"));
738 talloc_free(state);
739 ctdb_db->push_state = NULL;
740 return -1;
741 }
742
743 memcpy(outdata->dptr, (uint8_t *)&state->num_records, sizeof(uint32_t));
744 outdata->dsize = sizeof(uint32_t);
745
746 talloc_free(state);
747 ctdb_db->push_started = false;
748 ctdb_db->push_state = NULL;
749
750 return 0;
751 }
752
753 struct set_recmode_state {
754 struct ctdb_context *ctdb;
755 struct ctdb_req_control_old *c;
756 };
757
758 static void set_recmode_handler(char status,
759 double latency,
760 void *private_data)
761 {
762 struct set_recmode_state *state = talloc_get_type_abort(
763 private_data, struct set_recmode_state);
764 int s = 0;
765 const char *err = NULL;
766
767 switch (status) {
768 case '0':
769 /* Mutex taken */
770 DEBUG(DEBUG_ERR,
771 ("ERROR: Daemon able to take recovery lock on \"%s\" during recovery\n",
772 state->ctdb->recovery_lock));
773 s = -1;
774 err = "Took recovery lock from daemon during recovery - probably a cluster filesystem lock coherence problem";
775 break;
776
777 case '1':
778 /* Contention */
779 DEBUG(DEBUG_DEBUG, (__location__ " Recovery lock check OK\n"));
780 state->ctdb->recovery_mode = CTDB_RECOVERY_NORMAL;
781 ctdb_process_deferred_attach(state->ctdb);
782
783 s = 0;
784
785 CTDB_UPDATE_RECLOCK_LATENCY(state->ctdb, "daemon reclock",
786 reclock.ctdbd, latency);
787 break;
788
789 case '2':
790 /* Timeout. Consider this a success, not a failure,
791 * as we failed to set the recovery lock which is what
792 * we wanted. This can be caused by the cluster
793 * filesystem being very slow to arbitrate locks
794 * immediately after a node failure. */
795 DEBUG(DEBUG_WARNING,
796 (__location__
797 "Time out getting recovery lock, allowing recmode set anyway\n"));
798 state->ctdb->recovery_mode = CTDB_RECOVERY_NORMAL;
799 ctdb_process_deferred_attach(state->ctdb);
800
801 s = 0;
802 break;
803
804 default:
805 DEBUG(DEBUG_ERR,
806 ("Unexpected error when testing recovery lock\n"));
807 s = -1;
808 err = "Unexpected error when testing recovery lock";
809 }
810
811 ctdb_request_control_reply(state->ctdb, state->c, NULL, s, err);
812 talloc_free(state);
813 }
814
815 static void
816 ctdb_drop_all_ips_event(struct tevent_context *ev, struct tevent_timer *te,
817 struct timeval t, void *private_data)
818 {
819 struct ctdb_context *ctdb = talloc_get_type(private_data, struct ctdb_context);
820
821 DEBUG(DEBUG_ERR,(__location__ " Been in recovery mode for too long. Dropping all IPS\n"));
822 talloc_free(ctdb->release_ips_ctx);
823 ctdb->release_ips_ctx = NULL;
824
825 ctdb_release_all_ips(ctdb);
826 }
827
828 /*
829 * Set up an event to drop all public ips if we remain in recovery for too
830 * long
831 */
832 int ctdb_deferred_drop_all_ips(struct ctdb_context *ctdb)
833 {
834 if (ctdb->release_ips_ctx != NULL) {
835 talloc_free(ctdb->release_ips_ctx);
836 }
837 ctdb->release_ips_ctx = talloc_new(ctdb);
838 CTDB_NO_MEMORY(ctdb, ctdb->release_ips_ctx);
839
840 tevent_add_timer(ctdb->ev, ctdb->release_ips_ctx,
841 timeval_current_ofs(ctdb->tunable.recovery_drop_all_ips, 0),
842 ctdb_drop_all_ips_event, ctdb);
843 return 0;
844 }
845
846 /*
847 set the recovery mode
848 */
849 int32_t ctdb_control_set_recmode(struct ctdb_context *ctdb,
850 struct ctdb_req_control_old *c,
851 TDB_DATA indata, bool *async_reply,
852 const char **errormsg)
853 {
854 uint32_t recmode = *(uint32_t *)indata.dptr;
855 struct ctdb_db_context *ctdb_db;
856 struct set_recmode_state *state;
857 struct ctdb_cluster_mutex_handle *h;
858
859 /* if we enter recovery but stay in recovery for too long
860 we will eventually drop all our ip addresses
861 */
862 if (recmode == CTDB_RECOVERY_NORMAL) {
863 talloc_free(ctdb->release_ips_ctx);
864 ctdb->release_ips_ctx = NULL;
865 } else {
866 if (ctdb_deferred_drop_all_ips(ctdb) != 0) {
867 DEBUG(DEBUG_ERR,("Failed to set up deferred drop all ips\n"));
868 }
869 }
870
871 if (recmode != ctdb->recovery_mode) {
872 DEBUG(DEBUG_NOTICE,(__location__ " Recovery mode set to %s\n",
873 recmode==CTDB_RECOVERY_NORMAL?"NORMAL":"ACTIVE"));
874 }
875
876 if (recmode != CTDB_RECOVERY_NORMAL ||
877 ctdb->recovery_mode != CTDB_RECOVERY_ACTIVE) {
878 ctdb->recovery_mode = recmode;
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 for (ctdb_db = ctdb->db_list; ctdb_db != NULL; ctdb_db = ctdb_db->next) {
888 if (ctdb_db->generation != ctdb->vnn_map->generation) {
889 DEBUG(DEBUG_ERR,
890 ("Inconsistent DB generation %u for %s\n",
891 ctdb_db->generation, ctdb_db->db_name));
892 DEBUG(DEBUG_ERR, ("Recovery mode set to ACTIVE\n"));
893 return -1;
894 }
895 }
896
897 /* force the databases to thaw */
898 if (ctdb_db_all_frozen(ctdb)) {
899 ctdb_control_thaw(ctdb, false);
900 }
901
902 if (ctdb->recovery_lock == NULL) {
903 /* Not using recovery lock file */
904 ctdb->recovery_mode = CTDB_RECOVERY_NORMAL;
905 ctdb_process_deferred_attach(ctdb);
906 return 0;
907 }
908
909 state = talloc_zero(ctdb, struct set_recmode_state);
910 if (state == NULL) {
911 DEBUG(DEBUG_ERR, (__location__ " out of memory\n"));
912 return -1;
913 }
914 state->ctdb = ctdb;
915 state->c = NULL;
916
917 h = ctdb_cluster_mutex(state, ctdb, ctdb->recovery_lock, 5,
918 set_recmode_handler, state, NULL, NULL);
919 if (h == NULL) {
920 talloc_free(state);
921 return -1;
922 }
923
924 state->c = talloc_steal(state, c);
925 *async_reply = true;
926
927 return 0;
928 }
929
930
931 /*
932 delete a record as part of the vacuum process
933 only delete if we are not lmaster or dmaster, and our rsn is <= the provided rsn
934 use non-blocking locks
935
936 return 0 if the record was successfully deleted (i.e. it does not exist
937 when the function returns)
938 or !0 is the record still exists in the tdb after returning.
939 */
940 static int delete_tdb_record(struct ctdb_context *ctdb, struct ctdb_db_context *ctdb_db, struct ctdb_rec_data_old *rec)
941 {
942 TDB_DATA key, data, data2;
943 struct ctdb_ltdb_header *hdr, *hdr2;
944
945 /* these are really internal tdb functions - but we need them here for
946 non-blocking lock of the freelist */
947 int tdb_lock_nonblock(struct tdb_context *tdb, int list, int ltype);
948 int tdb_unlock(struct tdb_context *tdb, int list, int ltype);
949
950
951 key.dsize = rec->keylen;
952 key.dptr = &rec->data[0];
953 data.dsize = rec->datalen;
954 data.dptr = &rec->data[rec->keylen];
955
956 if (ctdb_lmaster(ctdb, &key) == ctdb->pnn) {
957 DEBUG(DEBUG_INFO,(__location__ " Called delete on record where we are lmaster\n"));
958 return -1;
959 }
960
961 if (data.dsize != sizeof(struct ctdb_ltdb_header)) {
962 DEBUG(DEBUG_ERR,(__location__ " Bad record size\n"));
963 return -1;
964 }
965
966 hdr = (struct ctdb_ltdb_header *)data.dptr;
967
968 /* use a non-blocking lock */
969 if (tdb_chainlock_nonblock(ctdb_db->ltdb->tdb, key) != 0) {
970 return -1;
971 }
972
973 data2 = tdb_fetch(ctdb_db->ltdb->tdb, key);
974 if (data2.dptr == NULL) {
975 tdb_chainunlock(ctdb_db->ltdb->tdb, key);
976 return 0;
977 }
978
979 if (data2.dsize < sizeof(struct ctdb_ltdb_header)) {
980 if (tdb_lock_nonblock(ctdb_db->ltdb->tdb, -1, F_WRLCK) == 0) {
981 if (tdb_delete(ctdb_db->ltdb->tdb, key) != 0) {
982 DEBUG(DEBUG_CRIT,(__location__ " Failed to delete corrupt record\n"));
983 }
984 tdb_unlock(ctdb_db->ltdb->tdb, -1, F_WRLCK);
985 DEBUG(DEBUG_CRIT,(__location__ " Deleted corrupt record\n"));
986 }
987 tdb_chainunlock(ctdb_db->ltdb->tdb, key);
988 free(data2.dptr);
989 return 0;
990 }
991
992 hdr2 = (struct ctdb_ltdb_header *)data2.dptr;
993
994 if (hdr2->rsn > hdr->rsn) {
995 tdb_chainunlock(ctdb_db->ltdb->tdb, key);
996 DEBUG(DEBUG_INFO,(__location__ " Skipping record with rsn=%llu - called with rsn=%llu\n",
997 (unsigned long long)hdr2->rsn, (unsigned long long)hdr->rsn));
998 free(data2.dptr);
999 return -1;
1000 }
1001
1002 /* do not allow deleting record that have readonly flags set. */
1003 if (hdr->flags & CTDB_REC_RO_FLAGS) {
1004 tdb_chainunlock(ctdb_db->ltdb->tdb, key);
1005 DEBUG(DEBUG_INFO,(__location__ " Skipping record with readonly flags set\n"));
1006 free(data2.dptr);
1007 return -1;
1008 }
1009 if (hdr2->flags & CTDB_REC_RO_FLAGS) {
1010 tdb_chainunlock(ctdb_db->ltdb->tdb, key);
1011 DEBUG(DEBUG_INFO,(__location__ " Skipping record with readonly flags set\n"));
1012 free(data2.dptr);
1013 return -1;
1014 }
1015
1016 if (hdr2->dmaster == ctdb->pnn) {
1017 tdb_chainunlock(ctdb_db->ltdb->tdb, key);
1018 DEBUG(DEBUG_INFO,(__location__ " Attempted delete record where we are the dmaster\n"));
1019 free(data2.dptr);
1020 return -1;
1021 }
1022
1023 if (tdb_lock_nonblock(ctdb_db->ltdb->tdb, -1, F_WRLCK) != 0) {
1024 tdb_chainunlock(ctdb_db->ltdb->tdb, key);
1025 free(data2.dptr);
1026 return -1;
1027 }
1028
1029 if (tdb_delete(ctdb_db->ltdb->tdb, key) != 0) {
1030 tdb_unlock(ctdb_db->ltdb->tdb, -1, F_WRLCK);
1031 tdb_chainunlock(ctdb_db->ltdb->tdb, key);
1032 DEBUG(DEBUG_INFO,(__location__ " Failed to delete record\n"));
1033 free(data2.dptr);
1034 return -1;
1035 }
1036
1037 tdb_unlock(ctdb_db->ltdb->tdb, -1, F_WRLCK);
1038 tdb_chainunlock(ctdb_db->ltdb->tdb, key);
1039 free(data2.dptr);
1040 return 0;
1041 }
1042
1043
1044
1045 struct recovery_callback_state {
1046 struct ctdb_req_control_old *c;
1047 };
1048
1049
1050 /*
1051 called when the 'recovered' event script has finished
1052 */
1053 static void ctdb_end_recovery_callback(struct ctdb_context *ctdb, int status, void *p)
1054 {
1055 struct recovery_callback_state *state = talloc_get_type(p, struct recovery_callback_state);
1056
1057 ctdb_enable_monitoring(ctdb);
1058 CTDB_INCREMENT_STAT(ctdb, num_recoveries);
1059
1060 if (status != 0) {
1061 DEBUG(DEBUG_ERR,(__location__ " recovered event script failed (status %d)\n", status));
1062 if (status == -ETIME) {
1063 ctdb_ban_self(ctdb);
1064 }
1065 }
1066
1067 ctdb_request_control_reply(ctdb, state->c, NULL, status, NULL);
1068 talloc_free(state);
1069
1070 gettimeofday(&ctdb->last_recovery_finished, NULL);
1071
1072 if (ctdb->runstate == CTDB_RUNSTATE_FIRST_RECOVERY) {
1073 ctdb_set_runstate(ctdb, CTDB_RUNSTATE_STARTUP);
1074 }
1075 }
1076
1077 /*
1078 recovery has finished
1079 */
1080 int32_t ctdb_control_end_recovery(struct ctdb_context *ctdb,
1081 struct ctdb_req_control_old *c,
1082 bool *async_reply)
1083 {
1084 int ret;
1085 struct recovery_callback_state *state;
1086
1087 DEBUG(DEBUG_NOTICE,("Recovery has finished\n"));
1088
1089 ctdb_persistent_finish_trans3_commits(ctdb);
1090
1091 state = talloc(ctdb, struct recovery_callback_state);
1092 CTDB_NO_MEMORY(ctdb, state);
1093
1094 state->c = c;
1095
1096 ctdb_disable_monitoring(ctdb);
1097
1098 ret = ctdb_event_script_callback(ctdb, state,
1099 ctdb_end_recovery_callback,
1100 state,
1101 CTDB_EVENT_RECOVERED, "%s", "");
1102
1103 if (ret != 0) {
1104 ctdb_enable_monitoring(ctdb);
1105
1106 DEBUG(DEBUG_ERR,(__location__ " Failed to end recovery\n"));
1107 talloc_free(state);
1108 return -1;
1109 }
1110
1111 /* tell the control that we will be reply asynchronously */
1112 state->c = talloc_steal(state, c);
1113 *async_reply = true;
1114 return 0;
1115 }
1116
1117 /*
1118 called when the 'startrecovery' event script has finished
1119 */
1120 static void ctdb_start_recovery_callback(struct ctdb_context *ctdb, int status, void *p)
1121 {
1122 struct recovery_callback_state *state = talloc_get_type(p, struct recovery_callback_state);
1123
1124 if (status != 0) {
1125 DEBUG(DEBUG_ERR,(__location__ " startrecovery event script failed (status %d)\n", status));
1126 }
1127
1128 ctdb_request_control_reply(ctdb, state->c, NULL, status, NULL);
1129 talloc_free(state);
1130 }
1131
1132 static void run_start_recovery_event(struct ctdb_context *ctdb,
1133 struct recovery_callback_state *state)
1134 {
1135 int ret;
1136
1137 ctdb_disable_monitoring(ctdb);
1138
1139 ret = ctdb_event_script_callback(ctdb, state,
1140 ctdb_start_recovery_callback,
1141 state,
1142 CTDB_EVENT_START_RECOVERY,
1143 "%s", "");
1144
1145 if (ret != 0) {
1146 DEBUG(DEBUG_ERR,("Unable to run startrecovery event\n"));
1147 ctdb_request_control_reply(ctdb, state->c, NULL, -1, NULL);
1148 talloc_free(state);
1149 return;
1150 }
1151
1152 return;
1153 }
1154
1155 static bool reclock_strings_equal(const char *a, const char *b)
1156 {
1157 return (a == NULL && b == NULL) ||
1158 (a != NULL && b != NULL && strcmp(a, b) == 0);
1159 }
1160
1161 static void start_recovery_reclock_callback(struct ctdb_context *ctdb,
1162 int32_t status,
1163 TDB_DATA data,
1164 const char *errormsg,
1165 void *private_data)
1166 {
1167 struct recovery_callback_state *state = talloc_get_type_abort(
1168 private_data, struct recovery_callback_state);
1169 const char *local = ctdb->recovery_lock;
1170 const char *remote = NULL;
1171
1172 if (status != 0) {
1173 DEBUG(DEBUG_ERR, (__location__ " GET_RECLOCK failed\n"));
1174 ctdb_request_control_reply(ctdb, state->c, NULL,
1175 status, errormsg);
1176 talloc_free(state);
1177 return;
1178 }
1179
1180 /* Check reclock consistency */
1181 if (data.dsize > 0) {
1182 /* Ensure NUL-termination */
1183 data.dptr[data.dsize-1] = '\0';
1184 remote = (const char *)data.dptr;
1185 }
1186 if (! reclock_strings_equal(local, remote)) {
1187 /* Inconsistent */
1188 ctdb_request_control_reply(ctdb, state->c, NULL, -1, NULL);
1189 DEBUG(DEBUG_ERR,
1190 ("Recovery lock configuration inconsistent: "
1191 "recmaster has %s, this node has %s, shutting down\n",
1192 remote == NULL ? "NULL" : remote,
1193 local == NULL ? "NULL" : local));
1194 talloc_free(state);
1195 ctdb_shutdown_sequence(ctdb, 1);
1196 }
1197 DEBUG(DEBUG_INFO,
1198 ("Recovery lock consistency check successful\n"));
1199
1200 run_start_recovery_event(ctdb, state);
1201 }
1202
1203 /* Check recovery lock consistency and run eventscripts for the
1204 * "startrecovery" event */
1205 int32_t ctdb_control_start_recovery(struct ctdb_context *ctdb,
1206 struct ctdb_req_control_old *c,
1207 bool *async_reply)
1208 {
1209 int ret;
1210 struct recovery_callback_state *state;
1211 uint32_t recmaster = c->hdr.srcnode;
1212
1213 DEBUG(DEBUG_NOTICE, ("Recovery has started\n"));
1214 gettimeofday(&ctdb->last_recovery_started, NULL);
1215
1216 state = talloc(ctdb, struct recovery_callback_state);
1217 CTDB_NO_MEMORY(ctdb, state);
1218
1219 state->c = c;
1220
1221 /* Although the recovery master sent this node a start
1222 * recovery control, this node might still think the recovery
1223 * master is disconnected. In this case defer the recovery
1224 * lock consistency check. */
1225 if (ctdb->nodes[recmaster]->flags & NODE_FLAGS_DISCONNECTED) {
1226 run_start_recovery_event(ctdb, state);
1227 } else {
1228 /* Ask the recovery master about its reclock setting */
1229 ret = ctdb_daemon_send_control(ctdb,
1230 recmaster,
1231 0,
1232 CTDB_CONTROL_GET_RECLOCK_FILE,
1233 0, 0,
1234 tdb_null,
1235 start_recovery_reclock_callback,
1236 state);
1237
1238 if (ret != 0) {
1239 DEBUG(DEBUG_ERR, (__location__ " GET_RECLOCK failed\n"));
1240 talloc_free(state);
1241 return -1;
1242 }
1243 }
1244
1245 /* tell the control that we will be reply asynchronously */
1246 state->c = talloc_steal(state, c);
1247 *async_reply = true;
1248
1249 return 0;
1250 }
1251
1252 /*
1253 try to delete all these records as part of the vacuuming process
1254 and return the records we failed to delete
1255 */
1256 int32_t ctdb_control_try_delete_records(struct ctdb_context *ctdb, TDB_DATA indata, TDB_DATA *outdata)
1257 {
1258 struct ctdb_marshall_buffer *reply = (struct ctdb_marshall_buffer *)indata.dptr;
1259 struct ctdb_db_context *ctdb_db;
1260 int i;
1261 struct ctdb_rec_data_old *rec;
1262 struct ctdb_marshall_buffer *records;
1263
1264 if (indata.dsize < offsetof(struct ctdb_marshall_buffer, data)) {
1265 DEBUG(DEBUG_ERR,(__location__ " invalid data in try_delete_records\n"));
1266 return -1;
1267 }
1268
1269 ctdb_db = find_ctdb_db(ctdb, reply->db_id);
1270 if (!ctdb_db) {
1271 DEBUG(DEBUG_ERR,(__location__ " Unknown db 0x%08x\n", reply->db_id));
1272 return -1;
1273 }
1274
1275
1276 DEBUG(DEBUG_DEBUG,("starting try_delete_records of %u records for dbid 0x%x\n",
1277 reply->count, reply->db_id));
1278
1279
1280 /* create a blob to send back the records we couldnt delete */
1281 records = (struct ctdb_marshall_buffer *)
1282 talloc_zero_size(outdata,
1283 offsetof(struct ctdb_marshall_buffer, data));
1284 if (records == NULL) {
1285 DEBUG(DEBUG_ERR,(__location__ " Out of memory\n"));
1286 return -1;
1287 }
1288 records->db_id = ctdb_db->db_id;
1289
1290
1291 rec = (struct ctdb_rec_data_old *)&reply->data[0];
1292 for (i=0;i<reply->count;i++) {
1293 TDB_DATA key, data;
1294
1295 key.dptr = &rec->data[0];
1296 key.dsize = rec->keylen;
1297 data.dptr = &rec->data[key.dsize];
1298 data.dsize = rec->datalen;
1299
1300 if (data.dsize < sizeof(struct ctdb_ltdb_header)) {
1301 DEBUG(DEBUG_CRIT,(__location__ " bad ltdb record in indata\n"));
1302 talloc_free(records);
1303 return -1;
1304 }
1305
1306 /* If we cant delete the record we must add it to the reply
1307 so the lmaster knows it may not purge this record
1308 */
1309 if (delete_tdb_record(ctdb, ctdb_db, rec) != 0) {
1310 size_t old_size;
1311 struct ctdb_ltdb_header *hdr;
1312
1313 hdr = (struct ctdb_ltdb_header *)data.dptr;
1314 data.dptr += sizeof(*hdr);
1315 data.dsize -= sizeof(*hdr);
1316
1317 DEBUG(DEBUG_INFO, (__location__ " Failed to vacuum delete record with hash 0x%08x\n", ctdb_hash(&key)));
1318
1319 old_size = talloc_get_size(records);
1320 records = talloc_realloc_size(outdata, records, old_size + rec->length);
1321 if (records == NULL) {
1322 DEBUG(DEBUG_ERR,(__location__ " Failed to expand\n"));
1323 return -1;
1324 }
1325 records->count++;
1326 memcpy(old_size+(uint8_t *)records, rec, rec->length);
1327 }
1328
1329 rec = (struct ctdb_rec_data_old *)(rec->length + (uint8_t *)rec);
1330 }
1331
1332
1333 *outdata = ctdb_marshall_finish(records);
1334
1335 return 0;
1336 }
1337
1338 /**
1339 * Store a record as part of the vacuum process:
1340 * This is called from the RECEIVE_RECORD control which
1341 * the lmaster uses to send the current empty copy
1342 * to all nodes for storing, before it lets the other
1343 * nodes delete the records in the second phase with
1344 * the TRY_DELETE_RECORDS control.
1345 *
1346 * Only store if we are not lmaster or dmaster, and our
1347 * rsn is <= the provided rsn. Use non-blocking locks.
1348 *
1349 * return 0 if the record was successfully stored.
1350 * return !0 if the record still exists in the tdb after returning.
1351 */
1352 static int store_tdb_record(struct ctdb_context *ctdb,
1353 struct ctdb_db_context *ctdb_db,
1354 struct ctdb_rec_data_old *rec)
1355 {
1356 TDB_DATA key, data, data2;
1357 struct ctdb_ltdb_header *hdr, *hdr2;
1358 int ret;
1359
1360 key.dsize = rec->keylen;
1361 key.dptr = &rec->data[0];
1362 data.dsize = rec->datalen;
1363 data.dptr = &rec->data[rec->keylen];
1364
1365 if (ctdb_lmaster(ctdb, &key) == ctdb->pnn) {
1366 DEBUG(DEBUG_INFO, (__location__ " Called store_tdb_record "
1367 "where we are lmaster\n"));
1368 return -1;
1369 }
1370
1371 if (data.dsize != sizeof(struct ctdb_ltdb_header)) {
1372 DEBUG(DEBUG_ERR, (__location__ " Bad record size\n"));
1373 return -1;
1374 }
1375
1376 hdr = (struct ctdb_ltdb_header *)data.dptr;
1377
1378 /* use a non-blocking lock */
1379 if (tdb_chainlock_nonblock(ctdb_db->ltdb->tdb, key) != 0) {
1380 DEBUG(DEBUG_INFO, (__location__ " Failed to lock chain in non-blocking mode\n"));
1381 return -1;
1382 }
1383
1384 data2 = tdb_fetch(ctdb_db->ltdb->tdb, key);
1385 if (data2.dptr == NULL || data2.dsize < sizeof(struct ctdb_ltdb_header)) {
1386 if (tdb_store(ctdb_db->ltdb->tdb, key, data, 0) == -1) {
1387 DEBUG(DEBUG_ERR, (__location__ "Failed to store record\n"));
1388 ret = -1;
1389 goto done;
1390 }
1391 DEBUG(DEBUG_INFO, (__location__ " Stored record\n"));
1392 ret = 0;
1393 goto done;
1394 }
1395
1396 hdr2 = (struct ctdb_ltdb_header *)data2.dptr;
1397
1398 if (hdr2->rsn > hdr->rsn) {
1399 DEBUG(DEBUG_INFO, (__location__ " Skipping record with "
1400 "rsn=%llu - called with rsn=%llu\n",
1401 (unsigned long long)hdr2->rsn,
1402 (unsigned long long)hdr->rsn));
1403 ret = -1;
1404 goto done;
1405 }
1406
1407 /* do not allow vacuuming of records that have readonly flags set. */
1408 if (hdr->flags & CTDB_REC_RO_FLAGS) {
1409 DEBUG(DEBUG_INFO,(__location__ " Skipping record with readonly "
1410 "flags set\n"));
1411 ret = -1;
1412 goto done;
1413 }
1414 if (hdr2->flags & CTDB_REC_RO_FLAGS) {
1415 DEBUG(DEBUG_INFO,(__location__ " Skipping record with readonly "
1416 "flags set\n"));
1417 ret = -1;
1418 goto done;
1419 }
1420
1421 if (hdr2->dmaster == ctdb->pnn) {
1422 DEBUG(DEBUG_INFO, (__location__ " Attempted to store record "
1423 "where we are the dmaster\n"));
1424 ret = -1;
1425 goto done;
1426 }
1427
1428 if (tdb_store(ctdb_db->ltdb->tdb, key, data, 0) != 0) {
1429 DEBUG(DEBUG_INFO,(__location__ " Failed to store record\n"));
1430 ret = -1;
1431 goto done;
1432 }
1433
1434 ret = 0;
1435
1436 done:
1437 tdb_chainunlock(ctdb_db->ltdb->tdb, key);
1438 free(data2.dptr);
1439 return ret;
1440 }
1441
1442
1443
1444 /**
1445 * Try to store all these records as part of the vacuuming process
1446 * and return the records we failed to store.
1447 */
1448 int32_t ctdb_control_receive_records(struct ctdb_context *ctdb,
1449 TDB_DATA indata, TDB_DATA *outdata)
1450 {
1451 struct ctdb_marshall_buffer *reply = (struct ctdb_marshall_buffer *)indata.dptr;
1452 struct ctdb_db_context *ctdb_db;
1453 int i;
1454 struct ctdb_rec_data_old *rec;
1455 struct ctdb_marshall_buffer *records;
1456
1457 if (indata.dsize < offsetof(struct ctdb_marshall_buffer, data)) {
1458 DEBUG(DEBUG_ERR,
1459 (__location__ " invalid data in receive_records\n"));
1460 return -1;
1461 }
1462
1463 ctdb_db = find_ctdb_db(ctdb, reply->db_id);
1464 if (!ctdb_db) {
1465 DEBUG(DEBUG_ERR, (__location__ " Unknown db 0x%08x\n",
1466 reply->db_id));
1467 return -1;
1468 }
1469
1470 DEBUG(DEBUG_DEBUG, ("starting receive_records of %u records for "
1471 "dbid 0x%x\n", reply->count, reply->db_id));
1472
1473 /* create a blob to send back the records we could not store */
1474 records = (struct ctdb_marshall_buffer *)
1475 talloc_zero_size(outdata,
1476 offsetof(struct ctdb_marshall_buffer, data));
1477 if (records == NULL) {
1478 DEBUG(DEBUG_ERR, (__location__ " Out of memory\n"));
1479 return -1;
1480 }
1481 records->db_id = ctdb_db->db_id;
1482
1483 rec = (struct ctdb_rec_data_old *)&reply->data[0];
1484 for (i=0; i<reply->count; i++) {
1485 TDB_DATA key, data;
1486
1487 key.dptr = &rec->data[0];
1488 key.dsize = rec->keylen;
1489 data.dptr = &rec->data[key.dsize];
1490 data.dsize = rec->datalen;
1491
1492 if (data.dsize < sizeof(struct ctdb_ltdb_header)) {
1493 DEBUG(DEBUG_CRIT, (__location__ " bad ltdb record "
1494 "in indata\n"));
1495 return -1;
1496 }
1497
1498 /*
1499 * If we can not store the record we must add it to the reply
1500 * so the lmaster knows it may not purge this record.
1501 */
1502 if (store_tdb_record(ctdb, ctdb_db, rec) != 0) {
1503 size_t old_size;
1504 struct ctdb_ltdb_header *hdr;
1505
1506 hdr = (struct ctdb_ltdb_header *)data.dptr;
1507 data.dptr += sizeof(*hdr);
1508 data.dsize -= sizeof(*hdr);
1509
1510 DEBUG(DEBUG_INFO, (__location__ " Failed to store "
1511 "record with hash 0x%08x in vacuum "
1512 "via RECEIVE_RECORDS\n",
1513 ctdb_hash(&key)));
1514
1515 old_size = talloc_get_size(records);
1516 records = talloc_realloc_size(outdata, records,
1517 old_size + rec->length);
1518 if (records == NULL) {
1519 DEBUG(DEBUG_ERR, (__location__ " Failed to "
1520 "expand\n"));
1521 return -1;
1522 }
1523 records->count++;
1524 memcpy(old_size+(uint8_t *)records, rec, rec->length);
1525 }
1526
1527 rec = (struct ctdb_rec_data_old *)(rec->length + (uint8_t *)rec);
1528 }
1529
1530 *outdata = ctdb_marshall_finish(records);
1531
1532 return 0;
1533 }
1534
1535
1536 /*
1537 report capabilities
1538 */
1539 int32_t ctdb_control_get_capabilities(struct ctdb_context *ctdb, TDB_DATA *outdata)
1540 {
1541 uint32_t *capabilities = NULL;
1542
1543 capabilities = talloc(outdata, uint32_t);
1544 CTDB_NO_MEMORY(ctdb, capabilities);
1545 *capabilities = ctdb->capabilities;
1546
1547 outdata->dsize = sizeof(uint32_t);
1548 outdata->dptr = (uint8_t *)capabilities;
1549
1550 return 0;
1551 }
1552
1553 /* The recovery daemon will ping us at regular intervals.
1554 If we havent been pinged for a while we assume the recovery
1555 daemon is inoperable and we restart.
1556 */
1557 static void ctdb_recd_ping_timeout(struct tevent_context *ev,
1558 struct tevent_timer *te,
1559 struct timeval t, void *p)
1560 {
1561 struct ctdb_context *ctdb = talloc_get_type(p, struct ctdb_context);
1562 uint32_t *count = talloc_get_type(ctdb->recd_ping_count, uint32_t);
1563
1564 DEBUG(DEBUG_ERR, ("Recovery daemon ping timeout. Count : %u\n", *count));
1565
1566 if (*count < ctdb->tunable.recd_ping_failcount) {
1567 (*count)++;
1568 tevent_add_timer(ctdb->ev, ctdb->recd_ping_count,
1569 timeval_current_ofs(ctdb->tunable.recd_ping_timeout, 0),
1570 ctdb_recd_ping_timeout, ctdb);
1571 return;
1572 }
1573
1574 DEBUG(DEBUG_ERR, ("Final timeout for recovery daemon ping. Restarting recovery daemon. (This can be caused if the cluster filesystem has hung)\n"));
1575
1576 ctdb_stop_recoverd(ctdb);
1577 ctdb_start_recoverd(ctdb);
1578 }
1579
1580 int32_t ctdb_control_recd_ping(struct ctdb_context *ctdb)
1581 {
1582 talloc_free(ctdb->recd_ping_count);
1583
1584 ctdb->recd_ping_count = talloc_zero(ctdb, uint32_t);
1585 CTDB_NO_MEMORY(ctdb, ctdb->recd_ping_count);
1586
1587 if (ctdb->tunable.recd_ping_timeout != 0) {
1588 tevent_add_timer(ctdb->ev, ctdb->recd_ping_count,
1589 timeval_current_ofs(ctdb->tunable.recd_ping_timeout, 0),
1590 ctdb_recd_ping_timeout, ctdb);
1591 }
1592
1593 return 0;
1594 }
1595
1596
1597
1598 int32_t ctdb_control_set_recmaster(struct ctdb_context *ctdb, uint32_t opcode, TDB_DATA indata)
1599 {
1600 uint32_t new_recmaster;
1601
1602 CHECK_CONTROL_DATA_SIZE(sizeof(uint32_t));
1603 new_recmaster = ((uint32_t *)(&indata.dptr[0]))[0];
1604
1605 if (ctdb->pnn != new_recmaster && ctdb->recovery_master == ctdb->pnn) {
1606 DEBUG(DEBUG_NOTICE,
1607 ("Remote node (%u) is now the recovery master\n",
1608 new_recmaster));
1609 }
1610
1611 if (ctdb->pnn == new_recmaster && ctdb->recovery_master != new_recmaster) {
1612 DEBUG(DEBUG_NOTICE,
1613 ("This node (%u) is now the recovery master\n",
1614 ctdb->pnn));
1615 }
1616
1617 ctdb->recovery_master = new_recmaster;
1618 return 0;
1619 }
1620
1621
1622 int32_t ctdb_control_stop_node(struct ctdb_context *ctdb)
1623 {
1624 DEBUG(DEBUG_NOTICE, ("Stopping node\n"));
1625 ctdb_disable_monitoring(ctdb);
1626 ctdb->nodes[ctdb->pnn]->flags |= NODE_FLAGS_STOPPED;
1627
1628 return 0;
1629 }
1630
1631 int32_t ctdb_control_continue_node(struct ctdb_context *ctdb)
1632 {
1633 DEBUG(DEBUG_NOTICE, ("Continue node\n"));
1634 ctdb->nodes[ctdb->pnn]->flags &= ~NODE_FLAGS_STOPPED;
1635
1636 return 0;
1637 }
1638
Samba GIT mirror