tools/ctdb: Fix documentation string for ban command
[Samba/wip.git] / ctdb / tools / ctdb.c
blobc5d4a4ad6cc7fe81058e117f99d88496f7b307f2
1 /*
2 ctdb control tool
4 Copyright (C) Andrew Tridgell 2007
5 Copyright (C) Ronnie Sahlberg 2007
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.
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.
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/>.
21 #include "includes.h"
22 #include "system/time.h"
23 #include "system/filesys.h"
24 #include "system/network.h"
25 #include "system/locale.h"
26 #include "popt.h"
27 #include "cmdline.h"
28 #include "../include/ctdb_version.h"
29 #include "../include/ctdb_client.h"
30 #include "../include/ctdb_private.h"
31 #include "../common/rb_tree.h"
32 #include "db_wrap.h"
34 #define ERR_TIMEOUT 20 /* timed out trying to reach node */
35 #define ERR_NONODE 21 /* node does not exist */
36 #define ERR_DISNODE 22 /* node is disconnected */
38 static void usage(void);
40 static struct {
41 int timelimit;
42 uint32_t pnn;
43 uint32_t *nodes;
44 int machinereadable;
45 int verbose;
46 int maxruntime;
47 int printemptyrecords;
48 int printdatasize;
49 int printlmaster;
50 int printhash;
51 int printrecordflags;
52 } options;
54 #define LONGTIMEOUT options.timelimit*10
56 #define TIMELIMIT() timeval_current_ofs(options.timelimit, 0)
57 #define LONGTIMELIMIT() timeval_current_ofs(LONGTIMEOUT, 0)
59 static int control_version(struct ctdb_context *ctdb, int argc, const char **argv)
61 printf("CTDB version: %s\n", CTDB_VERSION_STRING);
62 return 0;
65 #define CTDB_NOMEM_ABORT(p) do { if (!(p)) { \
66 DEBUG(DEBUG_ALERT,("ctdb fatal error: %s\n", \
67 "Out of memory in " __location__ )); \
68 abort(); \
69 }} while (0)
71 static uint32_t getpnn(struct ctdb_context *ctdb)
73 if ((options.pnn == CTDB_BROADCAST_ALL) ||
74 (options.pnn == CTDB_MULTICAST)) {
75 DEBUG(DEBUG_ERR,
76 ("Cannot get PNN for node %u\n", options.pnn));
77 exit(1);
80 if (options.pnn == CTDB_CURRENT_NODE) {
81 return ctdb_get_pnn(ctdb);
82 } else {
83 return options.pnn;
87 static void assert_single_node_only(void)
89 if ((options.pnn == CTDB_BROADCAST_ALL) ||
90 (options.pnn == CTDB_MULTICAST)) {
91 DEBUG(DEBUG_ERR,
92 ("This control can not be applied to multiple PNNs\n"));
93 exit(1);
97 /* Pretty print the flags to a static buffer in human-readable format.
98 * This never returns NULL!
100 static const char *pretty_print_flags(uint32_t flags)
102 int j;
103 static const struct {
104 uint32_t flag;
105 const char *name;
106 } flag_names[] = {
107 { NODE_FLAGS_DISCONNECTED, "DISCONNECTED" },
108 { NODE_FLAGS_PERMANENTLY_DISABLED, "DISABLED" },
109 { NODE_FLAGS_BANNED, "BANNED" },
110 { NODE_FLAGS_UNHEALTHY, "UNHEALTHY" },
111 { NODE_FLAGS_DELETED, "DELETED" },
112 { NODE_FLAGS_STOPPED, "STOPPED" },
113 { NODE_FLAGS_INACTIVE, "INACTIVE" },
115 static char flags_str[512]; /* Big enough to contain all flag names */
117 flags_str[0] = '\0';
118 for (j=0;j<ARRAY_SIZE(flag_names);j++) {
119 if (flags & flag_names[j].flag) {
120 if (flags_str[0] == '\0') {
121 (void) strcpy(flags_str, flag_names[j].name);
122 } else {
123 (void) strcat(flags_str, "|");
124 (void) strcat(flags_str, flag_names[j].name);
128 if (flags_str[0] == '\0') {
129 (void) strcpy(flags_str, "OK");
132 return flags_str;
135 static int h2i(char h)
137 if (h >= 'a' && h <= 'f') return h - 'a' + 10;
138 if (h >= 'A' && h <= 'F') return h - 'f' + 10;
139 return h - '0';
142 static TDB_DATA hextodata(TALLOC_CTX *mem_ctx, const char *str)
144 int i, len;
145 TDB_DATA key = {NULL, 0};
147 len = strlen(str);
148 if (len & 0x01) {
149 DEBUG(DEBUG_ERR,("Key specified with odd number of hexadecimal digits\n"));
150 return key;
153 key.dsize = len>>1;
154 key.dptr = talloc_size(mem_ctx, key.dsize);
156 for (i=0; i < len/2; i++) {
157 key.dptr[i] = h2i(str[i*2]) << 4 | h2i(str[i*2+1]);
159 return key;
162 /* Parse a nodestring. Parameter dd_ok controls what happens to nodes
163 * that are disconnected or deleted. If dd_ok is true those nodes are
164 * included in the output list of nodes. If dd_ok is false, those
165 * nodes are filtered from the "all" case and cause an error if
166 * explicitly specified.
168 static bool parse_nodestring(struct ctdb_context *ctdb,
169 TALLOC_CTX *mem_ctx,
170 const char * nodestring,
171 uint32_t current_pnn,
172 bool dd_ok,
173 uint32_t **nodes,
174 uint32_t *pnn_mode)
176 TALLOC_CTX *tmp_ctx = talloc_new(mem_ctx);
177 int n;
178 uint32_t i;
179 struct ctdb_node_map *nodemap;
180 int ret;
182 *nodes = NULL;
184 ret = ctdb_ctrl_getnodemap(ctdb, TIMELIMIT(), CTDB_CURRENT_NODE, tmp_ctx, &nodemap);
185 if (ret != 0) {
186 DEBUG(DEBUG_ERR, ("Unable to get nodemap from local node\n"));
187 talloc_free(tmp_ctx);
188 exit(10);
191 if (nodestring != NULL) {
192 *nodes = talloc_array(mem_ctx, uint32_t, 0);
193 if (*nodes == NULL) {
194 goto failed;
197 n = 0;
199 if (strcmp(nodestring, "all") == 0) {
200 *pnn_mode = CTDB_BROADCAST_ALL;
202 /* all */
203 for (i = 0; i < nodemap->num; i++) {
204 if ((nodemap->nodes[i].flags &
205 (NODE_FLAGS_DISCONNECTED |
206 NODE_FLAGS_DELETED)) && !dd_ok) {
207 continue;
209 *nodes = talloc_realloc(mem_ctx, *nodes,
210 uint32_t, n+1);
211 if (*nodes == NULL) {
212 goto failed;
214 (*nodes)[n] = i;
215 n++;
217 } else {
218 /* x{,y...} */
219 char *ns, *tok;
221 ns = talloc_strdup(tmp_ctx, nodestring);
222 tok = strtok(ns, ",");
223 while (tok != NULL) {
224 uint32_t pnn;
225 i = (uint32_t)strtoul(tok, NULL, 0);
226 if (i >= nodemap->num) {
227 DEBUG(DEBUG_ERR, ("Node %u does not exist\n", i));
228 talloc_free(tmp_ctx);
229 exit(ERR_NONODE);
231 if ((nodemap->nodes[i].flags &
232 (NODE_FLAGS_DISCONNECTED |
233 NODE_FLAGS_DELETED)) && !dd_ok) {
234 DEBUG(DEBUG_ERR, ("Node %u has status %s\n", i, pretty_print_flags(nodemap->nodes[i].flags)));
235 talloc_free(tmp_ctx);
236 exit(ERR_DISNODE);
238 if ((pnn = ctdb_ctrl_getpnn(ctdb, TIMELIMIT(), i)) < 0) {
239 DEBUG(DEBUG_ERR, ("Can not access node %u. Node is not operational.\n", i));
240 talloc_free(tmp_ctx);
241 exit(10);
244 *nodes = talloc_realloc(mem_ctx, *nodes,
245 uint32_t, n+1);
246 if (*nodes == NULL) {
247 goto failed;
250 (*nodes)[n] = i;
251 n++;
253 tok = strtok(NULL, ",");
255 talloc_free(ns);
257 if (n == 1) {
258 *pnn_mode = (*nodes)[0];
259 } else {
260 *pnn_mode = CTDB_MULTICAST;
263 } else {
264 /* default - no nodes specified */
265 *nodes = talloc_array(mem_ctx, uint32_t, 1);
266 if (*nodes == NULL) {
267 goto failed;
269 *pnn_mode = CTDB_CURRENT_NODE;
271 if (((*nodes)[0] = ctdb_ctrl_getpnn(ctdb, TIMELIMIT(), current_pnn)) < 0) {
272 goto failed;
276 talloc_free(tmp_ctx);
277 return true;
279 failed:
280 talloc_free(tmp_ctx);
281 return false;
285 check if a database exists
287 static bool db_exists(struct ctdb_context *ctdb, const char *dbarg, uint32_t *dbid, uint8_t *flags)
289 int i, ret;
290 struct ctdb_dbid_map *dbmap=NULL;
291 bool dbid_given = false, found = false;
292 uint32_t id;
293 TALLOC_CTX *tmp_ctx = talloc_new(ctdb);
295 ret = ctdb_ctrl_getdbmap(ctdb, TIMELIMIT(), options.pnn, tmp_ctx, &dbmap);
296 if (ret != 0) {
297 DEBUG(DEBUG_ERR, ("Unable to get dbids from node %u\n", options.pnn));
298 goto fail;
301 if (strncmp(dbarg, "0x", 2) == 0) {
302 id = strtoul(dbarg, NULL, 0);
303 dbid_given = true;
306 for(i=0; i<dbmap->num; i++) {
307 if (dbid_given) {
308 if (id == dbmap->dbs[i].dbid) {
309 found = true;
310 break;
312 } else {
313 const char *name;
314 ret = ctdb_ctrl_getdbname(ctdb, TIMELIMIT(), options.pnn, dbmap->dbs[i].dbid, tmp_ctx, &name);
315 if (ret != 0) {
316 DEBUG(DEBUG_ERR, ("Unable to get dbname from dbid %u\n", dbmap->dbs[i].dbid));
317 goto fail;
320 if (strcmp(name, dbarg) == 0) {
321 id = dbmap->dbs[i].dbid;
322 found = true;
323 break;
328 if (found) {
329 if (dbid) *dbid = id;
330 if (flags) *flags = dbmap->dbs[i].flags;
331 } else {
332 DEBUG(DEBUG_ERR,("No database matching '%s' found\n", dbarg));
335 fail:
336 talloc_free(tmp_ctx);
337 return found;
341 see if a process exists
343 static int control_process_exists(struct ctdb_context *ctdb, int argc, const char **argv)
345 uint32_t pnn, pid;
346 int ret;
347 if (argc < 1) {
348 usage();
351 if (sscanf(argv[0], "%u:%u", &pnn, &pid) != 2) {
352 DEBUG(DEBUG_ERR, ("Badly formed pnn:pid\n"));
353 return -1;
356 ret = ctdb_ctrl_process_exists(ctdb, pnn, pid);
357 if (ret == 0) {
358 printf("%u:%u exists\n", pnn, pid);
359 } else {
360 printf("%u:%u does not exist\n", pnn, pid);
362 return ret;
366 display statistics structure
368 static void show_statistics(struct ctdb_statistics *s, int show_header)
370 TALLOC_CTX *tmp_ctx = talloc_new(NULL);
371 int i;
372 const char *prefix=NULL;
373 int preflen=0;
374 int tmp, days, hours, minutes, seconds;
375 const struct {
376 const char *name;
377 uint32_t offset;
378 } fields[] = {
379 #define STATISTICS_FIELD(n) { #n, offsetof(struct ctdb_statistics, n) }
380 STATISTICS_FIELD(num_clients),
381 STATISTICS_FIELD(frozen),
382 STATISTICS_FIELD(recovering),
383 STATISTICS_FIELD(num_recoveries),
384 STATISTICS_FIELD(client_packets_sent),
385 STATISTICS_FIELD(client_packets_recv),
386 STATISTICS_FIELD(node_packets_sent),
387 STATISTICS_FIELD(node_packets_recv),
388 STATISTICS_FIELD(keepalive_packets_sent),
389 STATISTICS_FIELD(keepalive_packets_recv),
390 STATISTICS_FIELD(node.req_call),
391 STATISTICS_FIELD(node.reply_call),
392 STATISTICS_FIELD(node.req_dmaster),
393 STATISTICS_FIELD(node.reply_dmaster),
394 STATISTICS_FIELD(node.reply_error),
395 STATISTICS_FIELD(node.req_message),
396 STATISTICS_FIELD(node.req_control),
397 STATISTICS_FIELD(node.reply_control),
398 STATISTICS_FIELD(client.req_call),
399 STATISTICS_FIELD(client.req_message),
400 STATISTICS_FIELD(client.req_control),
401 STATISTICS_FIELD(timeouts.call),
402 STATISTICS_FIELD(timeouts.control),
403 STATISTICS_FIELD(timeouts.traverse),
404 STATISTICS_FIELD(locks.num_calls),
405 STATISTICS_FIELD(locks.num_current),
406 STATISTICS_FIELD(locks.num_pending),
407 STATISTICS_FIELD(locks.num_failed),
408 STATISTICS_FIELD(total_calls),
409 STATISTICS_FIELD(pending_calls),
410 STATISTICS_FIELD(childwrite_calls),
411 STATISTICS_FIELD(pending_childwrite_calls),
412 STATISTICS_FIELD(memory_used),
413 STATISTICS_FIELD(max_hop_count),
414 STATISTICS_FIELD(total_ro_delegations),
415 STATISTICS_FIELD(total_ro_revokes),
418 tmp = s->statistics_current_time.tv_sec - s->statistics_start_time.tv_sec;
419 seconds = tmp%60;
420 tmp /= 60;
421 minutes = tmp%60;
422 tmp /= 60;
423 hours = tmp%24;
424 tmp /= 24;
425 days = tmp;
427 if (options.machinereadable){
428 if (show_header) {
429 printf("CTDB version:");
430 printf("Current time of statistics:");
431 printf("Statistics collected since:");
432 for (i=0;i<ARRAY_SIZE(fields);i++) {
433 printf("%s:", fields[i].name);
435 printf("num_reclock_ctdbd_latency:");
436 printf("min_reclock_ctdbd_latency:");
437 printf("avg_reclock_ctdbd_latency:");
438 printf("max_reclock_ctdbd_latency:");
440 printf("num_reclock_recd_latency:");
441 printf("min_reclock_recd_latency:");
442 printf("avg_reclock_recd_latency:");
443 printf("max_reclock_recd_latency:");
445 printf("num_call_latency:");
446 printf("min_call_latency:");
447 printf("avg_call_latency:");
448 printf("max_call_latency:");
450 printf("num_lockwait_latency:");
451 printf("min_lockwait_latency:");
452 printf("avg_lockwait_latency:");
453 printf("max_lockwait_latency:");
455 printf("num_childwrite_latency:");
456 printf("min_childwrite_latency:");
457 printf("avg_childwrite_latency:");
458 printf("max_childwrite_latency:");
459 printf("\n");
461 printf("%d:", CTDB_VERSION);
462 printf("%d:", (int)s->statistics_current_time.tv_sec);
463 printf("%d:", (int)s->statistics_start_time.tv_sec);
464 for (i=0;i<ARRAY_SIZE(fields);i++) {
465 printf("%d:", *(uint32_t *)(fields[i].offset+(uint8_t *)s));
467 printf("%d:", s->reclock.ctdbd.num);
468 printf("%.6f:", s->reclock.ctdbd.min);
469 printf("%.6f:", s->reclock.ctdbd.num?s->reclock.ctdbd.total/s->reclock.ctdbd.num:0.0);
470 printf("%.6f:", s->reclock.ctdbd.max);
472 printf("%d:", s->reclock.recd.num);
473 printf("%.6f:", s->reclock.recd.min);
474 printf("%.6f:", s->reclock.recd.num?s->reclock.recd.total/s->reclock.recd.num:0.0);
475 printf("%.6f:", s->reclock.recd.max);
477 printf("%d:", s->call_latency.num);
478 printf("%.6f:", s->call_latency.min);
479 printf("%.6f:", s->call_latency.num?s->call_latency.total/s->call_latency.num:0.0);
480 printf("%.6f:", s->call_latency.max);
482 printf("%d:", s->childwrite_latency.num);
483 printf("%.6f:", s->childwrite_latency.min);
484 printf("%.6f:", s->childwrite_latency.num?s->childwrite_latency.total/s->childwrite_latency.num:0.0);
485 printf("%.6f:", s->childwrite_latency.max);
486 printf("\n");
487 } else {
488 printf("CTDB version %u\n", CTDB_VERSION);
489 printf("Current time of statistics : %s", ctime(&s->statistics_current_time.tv_sec));
490 printf("Statistics collected since : (%03d %02d:%02d:%02d) %s", days, hours, minutes, seconds, ctime(&s->statistics_start_time.tv_sec));
492 for (i=0;i<ARRAY_SIZE(fields);i++) {
493 if (strchr(fields[i].name, '.')) {
494 preflen = strcspn(fields[i].name, ".")+1;
495 if (!prefix || strncmp(prefix, fields[i].name, preflen) != 0) {
496 prefix = fields[i].name;
497 printf(" %*.*s\n", preflen-1, preflen-1, fields[i].name);
499 } else {
500 preflen = 0;
502 printf(" %*s%-22s%*s%10u\n",
503 preflen?4:0, "",
504 fields[i].name+preflen,
505 preflen?0:4, "",
506 *(uint32_t *)(fields[i].offset+(uint8_t *)s));
508 printf(" hop_count_buckets:");
509 for (i=0;i<MAX_COUNT_BUCKETS;i++) {
510 printf(" %d", s->hop_count_bucket[i]);
512 printf("\n");
513 printf(" lock_buckets:");
514 for (i=0; i<MAX_COUNT_BUCKETS; i++) {
515 printf(" %d", s->locks.buckets[i]);
517 printf("\n");
518 printf(" %-30s %.6f/%.6f/%.6f sec out of %d\n", "locks_latency MIN/AVG/MAX", s->locks.latency.min, s->locks.latency.num?s->locks.latency.total/s->locks.latency.num:0.0, s->locks.latency.max, s->locks.latency.num);
520 printf(" %-30s %.6f/%.6f/%.6f sec out of %d\n", "reclock_ctdbd MIN/AVG/MAX", s->reclock.ctdbd.min, s->reclock.ctdbd.num?s->reclock.ctdbd.total/s->reclock.ctdbd.num:0.0, s->reclock.ctdbd.max, s->reclock.ctdbd.num);
522 printf(" %-30s %.6f/%.6f/%.6f sec out of %d\n", "reclock_recd MIN/AVG/MAX", s->reclock.recd.min, s->reclock.recd.num?s->reclock.recd.total/s->reclock.recd.num:0.0, s->reclock.recd.max, s->reclock.recd.num);
524 printf(" %-30s %.6f/%.6f/%.6f sec out of %d\n", "call_latency MIN/AVG/MAX", s->call_latency.min, s->call_latency.num?s->call_latency.total/s->call_latency.num:0.0, s->call_latency.max, s->call_latency.num);
525 printf(" %-30s %.6f/%.6f/%.6f sec out of %d\n", "childwrite_latency MIN/AVG/MAX", s->childwrite_latency.min, s->childwrite_latency.num?s->childwrite_latency.total/s->childwrite_latency.num:0.0, s->childwrite_latency.max, s->childwrite_latency.num);
528 talloc_free(tmp_ctx);
532 display remote ctdb statistics combined from all nodes
534 static int control_statistics_all(struct ctdb_context *ctdb)
536 int ret, i;
537 struct ctdb_statistics statistics;
538 uint32_t *nodes;
539 uint32_t num_nodes;
541 nodes = ctdb_get_connected_nodes(ctdb, TIMELIMIT(), ctdb, &num_nodes);
542 CTDB_NO_MEMORY(ctdb, nodes);
544 ZERO_STRUCT(statistics);
546 for (i=0;i<num_nodes;i++) {
547 struct ctdb_statistics s1;
548 int j;
549 uint32_t *v1 = (uint32_t *)&s1;
550 uint32_t *v2 = (uint32_t *)&statistics;
551 uint32_t num_ints =
552 offsetof(struct ctdb_statistics, __last_counter) / sizeof(uint32_t);
553 ret = ctdb_ctrl_statistics(ctdb, nodes[i], &s1);
554 if (ret != 0) {
555 DEBUG(DEBUG_ERR, ("Unable to get statistics from node %u\n", nodes[i]));
556 return ret;
558 for (j=0;j<num_ints;j++) {
559 v2[j] += v1[j];
561 statistics.max_hop_count =
562 MAX(statistics.max_hop_count, s1.max_hop_count);
563 statistics.call_latency.max =
564 MAX(statistics.call_latency.max, s1.call_latency.max);
566 talloc_free(nodes);
567 printf("Gathered statistics for %u nodes\n", num_nodes);
568 show_statistics(&statistics, 1);
569 return 0;
573 display remote ctdb statistics
575 static int control_statistics(struct ctdb_context *ctdb, int argc, const char **argv)
577 int ret;
578 struct ctdb_statistics statistics;
580 if (options.pnn == CTDB_BROADCAST_ALL) {
581 return control_statistics_all(ctdb);
584 ret = ctdb_ctrl_statistics(ctdb, options.pnn, &statistics);
585 if (ret != 0) {
586 DEBUG(DEBUG_ERR, ("Unable to get statistics from node %u\n", options.pnn));
587 return ret;
589 show_statistics(&statistics, 1);
590 return 0;
595 reset remote ctdb statistics
597 static int control_statistics_reset(struct ctdb_context *ctdb, int argc, const char **argv)
599 int ret;
601 ret = ctdb_statistics_reset(ctdb, options.pnn);
602 if (ret != 0) {
603 DEBUG(DEBUG_ERR, ("Unable to reset statistics on node %u\n", options.pnn));
604 return ret;
606 return 0;
611 display remote ctdb rolling statistics
613 static int control_stats(struct ctdb_context *ctdb, int argc, const char **argv)
615 int ret;
616 struct ctdb_statistics_wire *stats;
617 int i, num_records = -1;
619 assert_single_node_only();
621 if (argc ==1) {
622 num_records = atoi(argv[0]) - 1;
625 ret = ctdb_ctrl_getstathistory(ctdb, TIMELIMIT(), options.pnn, ctdb, &stats);
626 if (ret != 0) {
627 DEBUG(DEBUG_ERR, ("Unable to get rolling statistics from node %u\n", options.pnn));
628 return ret;
630 for (i=0;i<stats->num;i++) {
631 if (stats->stats[i].statistics_start_time.tv_sec == 0) {
632 continue;
634 show_statistics(&stats->stats[i], i==0);
635 if (i == num_records) {
636 break;
639 return 0;
644 display remote ctdb db statistics
646 static int control_dbstatistics(struct ctdb_context *ctdb, int argc, const char **argv)
648 TALLOC_CTX *tmp_ctx = talloc_new(ctdb);
649 struct ctdb_db_statistics *dbstat;
650 int i;
651 uint32_t db_id;
652 int num_hot_keys;
653 int ret;
655 if (argc < 1) {
656 usage();
659 if (!db_exists(ctdb, argv[0], &db_id, NULL)) {
660 return -1;
663 ret = ctdb_ctrl_dbstatistics(ctdb, options.pnn, db_id, tmp_ctx, &dbstat);
664 if (ret != 0) {
665 DEBUG(DEBUG_ERR,("Failed to read db statistics from node\n"));
666 talloc_free(tmp_ctx);
667 return -1;
670 printf("DB Statistics: %s\n", argv[0]);
671 printf(" %*s%-22s%*s%10u\n", 0, "", "ro_delegations", 4, "",
672 dbstat->db_ro_delegations);
673 printf(" %*s%-22s%*s%10u\n", 0, "", "ro_revokes", 4, "",
674 dbstat->db_ro_delegations);
675 printf(" %s\n", "locks");
676 printf(" %*s%-22s%*s%10u\n", 4, "", "total", 0, "",
677 dbstat->locks.num_calls);
678 printf(" %*s%-22s%*s%10u\n", 4, "", "failed", 0, "",
679 dbstat->locks.num_failed);
680 printf(" %*s%-22s%*s%10u\n", 4, "", "current", 0, "",
681 dbstat->locks.num_current);
682 printf(" %*s%-22s%*s%10u\n", 4, "", "pending", 0, "",
683 dbstat->locks.num_pending);
684 printf(" %s", "hop_count_buckets:");
685 for (i=0; i<MAX_COUNT_BUCKETS; i++) {
686 printf(" %d", dbstat->hop_count_bucket[i]);
688 printf("\n");
689 printf(" %s", "lock_buckets:");
690 for (i=0; i<MAX_COUNT_BUCKETS; i++) {
691 printf(" %d", dbstat->locks.buckets[i]);
693 printf("\n");
694 printf(" %-30s %.6f/%.6f/%.6f sec out of %d\n",
695 "locks_latency MIN/AVG/MAX",
696 dbstat->locks.latency.min,
697 (dbstat->locks.latency.num ?
698 dbstat->locks.latency.total /dbstat->locks.latency.num :
699 0.0),
700 dbstat->locks.latency.max,
701 dbstat->locks.latency.num);
702 num_hot_keys = 0;
703 for (i=0; i<dbstat->num_hot_keys; i++) {
704 if (dbstat->hot_keys[i].count > 0) {
705 num_hot_keys++;
708 dbstat->num_hot_keys = num_hot_keys;
710 printf(" Num Hot Keys: %d\n", dbstat->num_hot_keys);
711 for (i = 0; i < dbstat->num_hot_keys; i++) {
712 int j;
713 printf(" Count:%d Key:", dbstat->hot_keys[i].count);
714 for (j = 0; j < dbstat->hot_keys[i].key.dsize; j++) {
715 printf("%02x", dbstat->hot_keys[i].key.dptr[j]&0xff);
717 printf("\n");
720 talloc_free(tmp_ctx);
721 return 0;
725 display uptime of remote node
727 static int control_uptime(struct ctdb_context *ctdb, int argc, const char **argv)
729 int ret;
730 struct ctdb_uptime *uptime = NULL;
731 int tmp, days, hours, minutes, seconds;
733 ret = ctdb_ctrl_uptime(ctdb, ctdb, TIMELIMIT(), options.pnn, &uptime);
734 if (ret != 0) {
735 DEBUG(DEBUG_ERR, ("Unable to get uptime from node %u\n", options.pnn));
736 return ret;
739 if (options.machinereadable){
740 printf(":Current Node Time:Ctdb Start Time:Last Recovery/Failover Time:Last Recovery/IPFailover Duration:\n");
741 printf(":%u:%u:%u:%lf\n",
742 (unsigned int)uptime->current_time.tv_sec,
743 (unsigned int)uptime->ctdbd_start_time.tv_sec,
744 (unsigned int)uptime->last_recovery_finished.tv_sec,
745 timeval_delta(&uptime->last_recovery_finished,
746 &uptime->last_recovery_started)
748 return 0;
751 printf("Current time of node : %s", ctime(&uptime->current_time.tv_sec));
753 tmp = uptime->current_time.tv_sec - uptime->ctdbd_start_time.tv_sec;
754 seconds = tmp%60;
755 tmp /= 60;
756 minutes = tmp%60;
757 tmp /= 60;
758 hours = tmp%24;
759 tmp /= 24;
760 days = tmp;
761 printf("Ctdbd start time : (%03d %02d:%02d:%02d) %s", days, hours, minutes, seconds, ctime(&uptime->ctdbd_start_time.tv_sec));
763 tmp = uptime->current_time.tv_sec - uptime->last_recovery_finished.tv_sec;
764 seconds = tmp%60;
765 tmp /= 60;
766 minutes = tmp%60;
767 tmp /= 60;
768 hours = tmp%24;
769 tmp /= 24;
770 days = tmp;
771 printf("Time of last recovery/failover: (%03d %02d:%02d:%02d) %s", days, hours, minutes, seconds, ctime(&uptime->last_recovery_finished.tv_sec));
773 printf("Duration of last recovery/failover: %lf seconds\n",
774 timeval_delta(&uptime->last_recovery_finished,
775 &uptime->last_recovery_started));
777 return 0;
781 show the PNN of the current node
783 static int control_pnn(struct ctdb_context *ctdb, int argc, const char **argv)
785 uint32_t mypnn;
787 mypnn = getpnn(ctdb);
789 printf("PNN:%d\n", mypnn);
790 return 0;
794 struct pnn_node {
795 struct pnn_node *next;
796 const char *addr;
797 int pnn;
800 static struct pnn_node *read_nodes_file(TALLOC_CTX *mem_ctx)
802 const char *nodes_list;
803 int nlines;
804 char **lines;
805 int i, pnn;
806 struct pnn_node *pnn_nodes = NULL;
807 struct pnn_node *pnn_node;
808 struct pnn_node *tmp_node;
810 /* read the nodes file */
811 nodes_list = getenv("CTDB_NODES");
812 if (nodes_list == NULL) {
813 nodes_list = talloc_asprintf(mem_ctx, "%s/nodes",
814 getenv("CTDB_BASE"));
815 if (nodes_list == NULL) {
816 DEBUG(DEBUG_ALERT,(__location__ " Out of memory\n"));
817 exit(1);
820 lines = file_lines_load(nodes_list, &nlines, mem_ctx);
821 if (lines == NULL) {
822 return NULL;
824 while (nlines > 0 && strcmp(lines[nlines-1], "") == 0) {
825 nlines--;
827 for (i=0, pnn=0; i<nlines; i++) {
828 char *node;
830 node = lines[i];
831 /* strip leading spaces */
832 while((*node == ' ') || (*node == '\t')) {
833 node++;
835 if (*node == '#') {
836 pnn++;
837 continue;
839 if (strcmp(node, "") == 0) {
840 continue;
842 pnn_node = talloc(mem_ctx, struct pnn_node);
843 pnn_node->pnn = pnn++;
844 pnn_node->addr = talloc_strdup(pnn_node, node);
845 pnn_node->next = pnn_nodes;
846 pnn_nodes = pnn_node;
849 /* swap them around so we return them in incrementing order */
850 pnn_node = pnn_nodes;
851 pnn_nodes = NULL;
852 while (pnn_node) {
853 tmp_node = pnn_node;
854 pnn_node = pnn_node->next;
856 tmp_node->next = pnn_nodes;
857 pnn_nodes = tmp_node;
860 return pnn_nodes;
864 show the PNN of the current node
865 discover the pnn by loading the nodes file and try to bind to all
866 addresses one at a time until the ip address is found.
868 static int control_xpnn(struct ctdb_context *ctdb, int argc, const char **argv)
870 TALLOC_CTX *mem_ctx = talloc_new(NULL);
871 struct pnn_node *pnn_nodes;
872 struct pnn_node *pnn_node;
874 assert_single_node_only();
876 pnn_nodes = read_nodes_file(mem_ctx);
877 if (pnn_nodes == NULL) {
878 DEBUG(DEBUG_ERR,("Failed to read nodes file\n"));
879 talloc_free(mem_ctx);
880 return -1;
883 for(pnn_node=pnn_nodes;pnn_node;pnn_node=pnn_node->next) {
884 ctdb_sock_addr addr;
886 if (parse_ip(pnn_node->addr, NULL, 63999, &addr) == 0) {
887 DEBUG(DEBUG_ERR,("Wrongly formed ip address '%s' in nodes file\n", pnn_node->addr));
888 talloc_free(mem_ctx);
889 return -1;
892 if (ctdb_sys_have_ip(&addr)) {
893 printf("PNN:%d\n", pnn_node->pnn);
894 talloc_free(mem_ctx);
895 return 0;
899 printf("Failed to detect which PNN this node is\n");
900 talloc_free(mem_ctx);
901 return -1;
904 /* Helpers for ctdb status
906 static bool is_partially_online(struct ctdb_context *ctdb, struct ctdb_node_and_flags *node)
908 TALLOC_CTX *tmp_ctx = talloc_new(NULL);
909 int j;
910 bool ret = false;
912 if (node->flags == 0) {
913 struct ctdb_control_get_ifaces *ifaces;
915 if (ctdb_ctrl_get_ifaces(ctdb, TIMELIMIT(), node->pnn,
916 tmp_ctx, &ifaces) == 0) {
917 for (j=0; j < ifaces->num; j++) {
918 if (ifaces->ifaces[j].link_state != 0) {
919 continue;
921 ret = true;
922 break;
926 talloc_free(tmp_ctx);
928 return ret;
931 static void control_status_header_machine(void)
933 printf(":Node:IP:Disconnected:Banned:Disabled:Unhealthy:Stopped"
934 ":Inactive:PartiallyOnline:ThisNode:\n");
937 static int control_status_1_machine(struct ctdb_context *ctdb, int mypnn,
938 struct ctdb_node_and_flags *node)
940 printf(":%d:%s:%d:%d:%d:%d:%d:%d:%d:%c:\n", node->pnn,
941 ctdb_addr_to_str(&node->addr),
942 !!(node->flags&NODE_FLAGS_DISCONNECTED),
943 !!(node->flags&NODE_FLAGS_BANNED),
944 !!(node->flags&NODE_FLAGS_PERMANENTLY_DISABLED),
945 !!(node->flags&NODE_FLAGS_UNHEALTHY),
946 !!(node->flags&NODE_FLAGS_STOPPED),
947 !!(node->flags&NODE_FLAGS_INACTIVE),
948 is_partially_online(ctdb, node) ? 1 : 0,
949 (node->pnn == mypnn)?'Y':'N');
951 return node->flags;
954 static int control_status_1_human(struct ctdb_context *ctdb, int mypnn,
955 struct ctdb_node_and_flags *node)
957 printf("pnn:%d %-16s %s%s\n", node->pnn,
958 ctdb_addr_to_str(&node->addr),
959 is_partially_online(ctdb, node) ? "PARTIALLYONLINE" : pretty_print_flags(node->flags),
960 node->pnn == mypnn?" (THIS NODE)":"");
962 return node->flags;
966 display remote ctdb status
968 static int control_status(struct ctdb_context *ctdb, int argc, const char **argv)
970 TALLOC_CTX *tmp_ctx = talloc_new(ctdb);
971 int i;
972 struct ctdb_vnn_map *vnnmap=NULL;
973 struct ctdb_node_map *nodemap=NULL;
974 uint32_t recmode, recmaster, mypnn;
975 int num_deleted_nodes = 0;
976 int ret;
978 mypnn = getpnn(ctdb);
980 ret = ctdb_ctrl_getnodemap(ctdb, TIMELIMIT(), options.pnn, tmp_ctx, &nodemap);
981 if (ret != 0) {
982 DEBUG(DEBUG_ERR, ("Unable to get nodemap from node %u\n", options.pnn));
983 talloc_free(tmp_ctx);
984 return -1;
987 if (options.machinereadable) {
988 control_status_header_machine();
989 for (i=0;i<nodemap->num;i++) {
990 if (nodemap->nodes[i].flags & NODE_FLAGS_DELETED) {
991 continue;
993 (void) control_status_1_machine(ctdb, mypnn,
994 &nodemap->nodes[i]);
996 talloc_free(tmp_ctx);
997 return 0;
1000 for (i=0; i<nodemap->num; i++) {
1001 if (nodemap->nodes[i].flags & NODE_FLAGS_DELETED) {
1002 num_deleted_nodes++;
1005 if (num_deleted_nodes == 0) {
1006 printf("Number of nodes:%d\n", nodemap->num);
1007 } else {
1008 printf("Number of nodes:%d (including %d deleted nodes)\n",
1009 nodemap->num, num_deleted_nodes);
1011 for(i=0;i<nodemap->num;i++){
1012 if (nodemap->nodes[i].flags & NODE_FLAGS_DELETED) {
1013 continue;
1015 (void) control_status_1_human(ctdb, mypnn, &nodemap->nodes[i]);
1018 ret = ctdb_ctrl_getvnnmap(ctdb, TIMELIMIT(), options.pnn, tmp_ctx, &vnnmap);
1019 if (ret != 0) {
1020 DEBUG(DEBUG_ERR, ("Unable to get vnnmap from node %u\n", options.pnn));
1021 talloc_free(tmp_ctx);
1022 return -1;
1024 if (vnnmap->generation == INVALID_GENERATION) {
1025 printf("Generation:INVALID\n");
1026 } else {
1027 printf("Generation:%d\n",vnnmap->generation);
1029 printf("Size:%d\n",vnnmap->size);
1030 for(i=0;i<vnnmap->size;i++){
1031 printf("hash:%d lmaster:%d\n", i, vnnmap->map[i]);
1034 ret = ctdb_ctrl_getrecmode(ctdb, tmp_ctx, TIMELIMIT(), options.pnn, &recmode);
1035 if (ret != 0) {
1036 DEBUG(DEBUG_ERR, ("Unable to get recmode from node %u\n", options.pnn));
1037 talloc_free(tmp_ctx);
1038 return -1;
1040 printf("Recovery mode:%s (%d)\n",recmode==CTDB_RECOVERY_NORMAL?"NORMAL":"RECOVERY",recmode);
1042 ret = ctdb_ctrl_getrecmaster(ctdb, tmp_ctx, TIMELIMIT(), options.pnn, &recmaster);
1043 if (ret != 0) {
1044 DEBUG(DEBUG_ERR, ("Unable to get recmaster from node %u\n", options.pnn));
1045 talloc_free(tmp_ctx);
1046 return -1;
1048 printf("Recovery master:%d\n",recmaster);
1050 talloc_free(tmp_ctx);
1051 return 0;
1054 static int control_nodestatus(struct ctdb_context *ctdb, int argc, const char **argv)
1056 TALLOC_CTX *tmp_ctx = talloc_new(ctdb);
1057 int i, ret;
1058 struct ctdb_node_map *nodemap=NULL;
1059 uint32_t * nodes;
1060 uint32_t pnn_mode, mypnn;
1062 if (argc > 1) {
1063 usage();
1066 if (!parse_nodestring(ctdb, tmp_ctx, argc == 1 ? argv[0] : NULL,
1067 options.pnn, true, &nodes, &pnn_mode)) {
1068 return -1;
1071 if (options.machinereadable) {
1072 control_status_header_machine();
1073 } else if (pnn_mode == CTDB_BROADCAST_ALL) {
1074 printf("Number of nodes:%d\n", (int) talloc_array_length(nodes));
1077 mypnn = getpnn(ctdb);
1079 ret = ctdb_ctrl_getnodemap(ctdb, TIMELIMIT(), options.pnn, tmp_ctx, &nodemap);
1080 if (ret != 0) {
1081 DEBUG(DEBUG_ERR, ("Unable to get nodemap from node %u\n", options.pnn));
1082 talloc_free(tmp_ctx);
1083 return -1;
1086 ret = 0;
1088 for (i = 0; i < talloc_array_length(nodes); i++) {
1089 if (options.machinereadable) {
1090 ret |= control_status_1_machine(ctdb, mypnn,
1091 &nodemap->nodes[nodes[i]]);
1092 } else {
1093 ret |= control_status_1_human(ctdb, mypnn,
1094 &nodemap->nodes[nodes[i]]);
1098 talloc_free(tmp_ctx);
1099 return ret;
1102 struct natgw_node {
1103 struct natgw_node *next;
1104 const char *addr;
1107 static int find_natgw(struct ctdb_context *ctdb,
1108 struct ctdb_node_map *nodemap, uint32_t flags,
1109 uint32_t *pnn, const char **ip)
1111 int i;
1112 uint32_t capabilities;
1113 int ret;
1115 for (i=0;i<nodemap->num;i++) {
1116 if (!(nodemap->nodes[i].flags & flags)) {
1117 ret = ctdb_ctrl_getcapabilities(ctdb, TIMELIMIT(),
1118 nodemap->nodes[i].pnn,
1119 &capabilities);
1120 if (ret != 0) {
1121 DEBUG(DEBUG_ERR, ("Unable to get capabilities from node %u\n",
1122 nodemap->nodes[i].pnn));
1123 return -1;
1125 if (!(capabilities&CTDB_CAP_NATGW)) {
1126 continue;
1128 *pnn = nodemap->nodes[i].pnn;
1129 *ip = ctdb_addr_to_str(&nodemap->nodes[i].addr);
1130 return 0;
1134 return 2; /* matches ENOENT */
1138 display the list of nodes belonging to this natgw configuration
1140 static int control_natgwlist(struct ctdb_context *ctdb, int argc, const char **argv)
1142 TALLOC_CTX *tmp_ctx = talloc_new(ctdb);
1143 int i, ret;
1144 const char *natgw_list;
1145 int nlines;
1146 char **lines;
1147 struct natgw_node *natgw_nodes = NULL;
1148 struct natgw_node *natgw_node;
1149 struct ctdb_node_map *nodemap=NULL;
1150 uint32_t mypnn, pnn;
1151 const char *ip;
1153 /* When we have some nodes that could be the NATGW, make a
1154 * series of attempts to find the first node that doesn't have
1155 * certain status flags set.
1157 uint32_t exclude_flags[] = {
1158 /* Look for a nice healthy node */
1159 NODE_FLAGS_DISCONNECTED|NODE_FLAGS_STOPPED|NODE_FLAGS_DELETED|NODE_FLAGS_BANNED|NODE_FLAGS_UNHEALTHY,
1160 /* If not found, an UNHEALTHY/BANNED node will do */
1161 NODE_FLAGS_DISCONNECTED|NODE_FLAGS_STOPPED|NODE_FLAGS_DELETED,
1162 /* If not found, a STOPPED node will do */
1163 NODE_FLAGS_DISCONNECTED|NODE_FLAGS_DELETED,
1167 /* read the natgw nodes file into a linked list */
1168 natgw_list = getenv("CTDB_NATGW_NODES");
1169 if (natgw_list == NULL) {
1170 natgw_list = talloc_asprintf(tmp_ctx, "%s/natgw_nodes",
1171 getenv("CTDB_BASE"));
1172 if (natgw_list == NULL) {
1173 DEBUG(DEBUG_ALERT,(__location__ " Out of memory\n"));
1174 exit(1);
1177 lines = file_lines_load(natgw_list, &nlines, ctdb);
1178 if (lines == NULL) {
1179 ctdb_set_error(ctdb, "Failed to load natgw node list '%s'\n", natgw_list);
1180 talloc_free(tmp_ctx);
1181 return -1;
1183 for (i=0;i<nlines;i++) {
1184 char *node;
1186 node = lines[i];
1187 /* strip leading spaces */
1188 while((*node == ' ') || (*node == '\t')) {
1189 node++;
1191 if (*node == '#') {
1192 continue;
1194 if (strcmp(node, "") == 0) {
1195 continue;
1197 natgw_node = talloc(ctdb, struct natgw_node);
1198 natgw_node->addr = talloc_strdup(natgw_node, node);
1199 CTDB_NO_MEMORY(ctdb, natgw_node->addr);
1200 natgw_node->next = natgw_nodes;
1201 natgw_nodes = natgw_node;
1204 ret = ctdb_ctrl_getnodemap(ctdb, TIMELIMIT(), CTDB_CURRENT_NODE, tmp_ctx, &nodemap);
1205 if (ret != 0) {
1206 DEBUG(DEBUG_ERR, ("Unable to get nodemap from local node.\n"));
1207 talloc_free(tmp_ctx);
1208 return -1;
1211 /* Trim the nodemap so it only includes connected nodes in the
1212 * current natgw group.
1214 i=0;
1215 while(i<nodemap->num) {
1216 for(natgw_node=natgw_nodes;natgw_node;natgw_node=natgw_node->next) {
1217 if (!strcmp(natgw_node->addr, ctdb_addr_to_str(&nodemap->nodes[i].addr))) {
1218 break;
1222 /* this node was not in the natgw so we just remove it from
1223 * the list
1225 if ((natgw_node == NULL)
1226 || (nodemap->nodes[i].flags & NODE_FLAGS_DISCONNECTED) ) {
1227 int j;
1229 for (j=i+1; j<nodemap->num; j++) {
1230 nodemap->nodes[j-1] = nodemap->nodes[j];
1232 nodemap->num--;
1233 continue;
1236 i++;
1239 ret = 2; /* matches ENOENT */
1240 pnn = -1;
1241 ip = "0.0.0.0";
1242 for (i = 0; exclude_flags[i] != 0; i++) {
1243 ret = find_natgw(ctdb, nodemap,
1244 exclude_flags[i],
1245 &pnn, &ip);
1246 if (ret == -1) {
1247 goto done;
1249 if (ret == 0) {
1250 break;
1254 if (options.machinereadable) {
1255 printf(":Node:IP:\n");
1256 printf(":%d:%s:\n", pnn, ip);
1257 } else {
1258 printf("%d %s\n", pnn, ip);
1261 /* print the pruned list of nodes belonging to this natgw list */
1262 mypnn = getpnn(ctdb);
1263 if (options.machinereadable) {
1264 control_status_header_machine();
1265 } else {
1266 printf("Number of nodes:%d\n", nodemap->num);
1268 for(i=0;i<nodemap->num;i++){
1269 if (nodemap->nodes[i].flags & NODE_FLAGS_DELETED) {
1270 continue;
1272 if (options.machinereadable) {
1273 control_status_1_machine(ctdb, mypnn, &(nodemap->nodes[i]));
1274 } else {
1275 control_status_1_human(ctdb, mypnn, &(nodemap->nodes[i]));
1279 done:
1280 talloc_free(tmp_ctx);
1281 return ret;
1285 display the status of the scripts for monitoring (or other events)
1287 static int control_one_scriptstatus(struct ctdb_context *ctdb,
1288 enum ctdb_eventscript_call type)
1290 struct ctdb_scripts_wire *script_status;
1291 int ret, i;
1293 ret = ctdb_ctrl_getscriptstatus(ctdb, TIMELIMIT(), options.pnn, ctdb, type, &script_status);
1294 if (ret != 0) {
1295 DEBUG(DEBUG_ERR, ("Unable to get script status from node %u\n", options.pnn));
1296 return ret;
1299 if (script_status == NULL) {
1300 if (!options.machinereadable) {
1301 printf("%s cycle never run\n",
1302 ctdb_eventscript_call_names[type]);
1304 return 0;
1307 if (!options.machinereadable) {
1308 printf("%d scripts were executed last %s cycle\n",
1309 script_status->num_scripts,
1310 ctdb_eventscript_call_names[type]);
1312 for (i=0; i<script_status->num_scripts; i++) {
1313 const char *status = NULL;
1315 switch (script_status->scripts[i].status) {
1316 case -ETIME:
1317 status = "TIMEDOUT";
1318 break;
1319 case -ENOEXEC:
1320 status = "DISABLED";
1321 break;
1322 case 0:
1323 status = "OK";
1324 break;
1325 default:
1326 if (script_status->scripts[i].status > 0)
1327 status = "ERROR";
1328 break;
1330 if (options.machinereadable) {
1331 printf(":%s:%s:%i:%s:%lu.%06lu:%lu.%06lu:%s:\n",
1332 ctdb_eventscript_call_names[type],
1333 script_status->scripts[i].name,
1334 script_status->scripts[i].status,
1335 status,
1336 (long)script_status->scripts[i].start.tv_sec,
1337 (long)script_status->scripts[i].start.tv_usec,
1338 (long)script_status->scripts[i].finished.tv_sec,
1339 (long)script_status->scripts[i].finished.tv_usec,
1340 script_status->scripts[i].output);
1341 continue;
1343 if (status)
1344 printf("%-20s Status:%s ",
1345 script_status->scripts[i].name, status);
1346 else
1347 /* Some other error, eg from stat. */
1348 printf("%-20s Status:CANNOT RUN (%s)",
1349 script_status->scripts[i].name,
1350 strerror(-script_status->scripts[i].status));
1352 if (script_status->scripts[i].status >= 0) {
1353 printf("Duration:%.3lf ",
1354 timeval_delta(&script_status->scripts[i].finished,
1355 &script_status->scripts[i].start));
1357 if (script_status->scripts[i].status != -ENOEXEC) {
1358 printf("%s",
1359 ctime(&script_status->scripts[i].start.tv_sec));
1360 if (script_status->scripts[i].status != 0) {
1361 printf(" OUTPUT:%s\n",
1362 script_status->scripts[i].output);
1364 } else {
1365 printf("\n");
1368 return 0;
1372 static int control_scriptstatus(struct ctdb_context *ctdb,
1373 int argc, const char **argv)
1375 int ret;
1376 enum ctdb_eventscript_call type, min, max;
1377 const char *arg;
1379 if (argc > 1) {
1380 DEBUG(DEBUG_ERR, ("Unknown arguments to scriptstatus\n"));
1381 return -1;
1384 if (argc == 0)
1385 arg = ctdb_eventscript_call_names[CTDB_EVENT_MONITOR];
1386 else
1387 arg = argv[0];
1389 for (type = 0; type < CTDB_EVENT_MAX; type++) {
1390 if (strcmp(arg, ctdb_eventscript_call_names[type]) == 0) {
1391 min = type;
1392 max = type+1;
1393 break;
1396 if (type == CTDB_EVENT_MAX) {
1397 if (strcmp(arg, "all") == 0) {
1398 min = 0;
1399 max = CTDB_EVENT_MAX;
1400 } else {
1401 DEBUG(DEBUG_ERR, ("Unknown event type %s\n", argv[0]));
1402 return -1;
1406 if (options.machinereadable) {
1407 printf(":Type:Name:Code:Status:Start:End:Error Output...:\n");
1410 for (type = min; type < max; type++) {
1411 ret = control_one_scriptstatus(ctdb, type);
1412 if (ret != 0) {
1413 return ret;
1417 return 0;
1421 enable an eventscript
1423 static int control_enablescript(struct ctdb_context *ctdb, int argc, const char **argv)
1425 int ret;
1427 if (argc < 1) {
1428 usage();
1431 ret = ctdb_ctrl_enablescript(ctdb, TIMELIMIT(), options.pnn, argv[0]);
1432 if (ret != 0) {
1433 DEBUG(DEBUG_ERR, ("Unable to enable script %s on node %u\n", argv[0], options.pnn));
1434 return ret;
1437 return 0;
1441 disable an eventscript
1443 static int control_disablescript(struct ctdb_context *ctdb, int argc, const char **argv)
1445 int ret;
1447 if (argc < 1) {
1448 usage();
1451 ret = ctdb_ctrl_disablescript(ctdb, TIMELIMIT(), options.pnn, argv[0]);
1452 if (ret != 0) {
1453 DEBUG(DEBUG_ERR, ("Unable to disable script %s on node %u\n", argv[0], options.pnn));
1454 return ret;
1457 return 0;
1461 display the pnn of the recovery master
1463 static int control_recmaster(struct ctdb_context *ctdb, int argc, const char **argv)
1465 uint32_t recmaster;
1466 int ret;
1468 ret = ctdb_ctrl_getrecmaster(ctdb, ctdb, TIMELIMIT(), options.pnn, &recmaster);
1469 if (ret != 0) {
1470 DEBUG(DEBUG_ERR, ("Unable to get recmaster from node %u\n", options.pnn));
1471 return -1;
1473 printf("%d\n",recmaster);
1475 return 0;
1479 add a tickle to a public address
1481 static int control_add_tickle(struct ctdb_context *ctdb, int argc, const char **argv)
1483 struct ctdb_tcp_connection t;
1484 TDB_DATA data;
1485 int ret;
1487 assert_single_node_only();
1489 if (argc < 2) {
1490 usage();
1493 if (parse_ip_port(argv[0], &t.src_addr) == 0) {
1494 DEBUG(DEBUG_ERR,("Wrongly formed ip address '%s'\n", argv[0]));
1495 return -1;
1497 if (parse_ip_port(argv[1], &t.dst_addr) == 0) {
1498 DEBUG(DEBUG_ERR,("Wrongly formed ip address '%s'\n", argv[1]));
1499 return -1;
1502 data.dptr = (uint8_t *)&t;
1503 data.dsize = sizeof(t);
1505 /* tell all nodes about this tcp connection */
1506 ret = ctdb_control(ctdb, options.pnn, 0, CTDB_CONTROL_TCP_ADD_DELAYED_UPDATE,
1507 0, data, ctdb, NULL, NULL, NULL, NULL);
1508 if (ret != 0) {
1509 DEBUG(DEBUG_ERR,("Failed to add tickle\n"));
1510 return -1;
1513 return 0;
1518 delete a tickle from a node
1520 static int control_del_tickle(struct ctdb_context *ctdb, int argc, const char **argv)
1522 struct ctdb_tcp_connection t;
1523 TDB_DATA data;
1524 int ret;
1526 assert_single_node_only();
1528 if (argc < 2) {
1529 usage();
1532 if (parse_ip_port(argv[0], &t.src_addr) == 0) {
1533 DEBUG(DEBUG_ERR,("Wrongly formed ip address '%s'\n", argv[0]));
1534 return -1;
1536 if (parse_ip_port(argv[1], &t.dst_addr) == 0) {
1537 DEBUG(DEBUG_ERR,("Wrongly formed ip address '%s'\n", argv[1]));
1538 return -1;
1541 data.dptr = (uint8_t *)&t;
1542 data.dsize = sizeof(t);
1544 /* tell all nodes about this tcp connection */
1545 ret = ctdb_control(ctdb, options.pnn, 0, CTDB_CONTROL_TCP_REMOVE,
1546 0, data, ctdb, NULL, NULL, NULL, NULL);
1547 if (ret != 0) {
1548 DEBUG(DEBUG_ERR,("Failed to remove tickle\n"));
1549 return -1;
1552 return 0;
1557 get a list of all tickles for this pnn
1559 static int control_get_tickles(struct ctdb_context *ctdb, int argc, const char **argv)
1561 struct ctdb_control_tcp_tickle_list *list;
1562 ctdb_sock_addr addr;
1563 int i, ret;
1564 unsigned port = 0;
1566 assert_single_node_only();
1568 if (argc < 1) {
1569 usage();
1572 if (argc == 2) {
1573 port = atoi(argv[1]);
1576 if (parse_ip(argv[0], NULL, 0, &addr) == 0) {
1577 DEBUG(DEBUG_ERR,("Wrongly formed ip address '%s'\n", argv[0]));
1578 return -1;
1581 ret = ctdb_ctrl_get_tcp_tickles(ctdb, TIMELIMIT(), options.pnn, ctdb, &addr, &list);
1582 if (ret == -1) {
1583 DEBUG(DEBUG_ERR, ("Unable to list tickles\n"));
1584 return -1;
1587 if (options.machinereadable){
1588 printf(":source ip:port:destination ip:port:\n");
1589 for (i=0;i<list->tickles.num;i++) {
1590 if (port && port != ntohs(list->tickles.connections[i].dst_addr.ip.sin_port)) {
1591 continue;
1593 printf(":%s:%u", ctdb_addr_to_str(&list->tickles.connections[i].src_addr), ntohs(list->tickles.connections[i].src_addr.ip.sin_port));
1594 printf(":%s:%u:\n", ctdb_addr_to_str(&list->tickles.connections[i].dst_addr), ntohs(list->tickles.connections[i].dst_addr.ip.sin_port));
1596 } else {
1597 printf("Tickles for ip:%s\n", ctdb_addr_to_str(&list->addr));
1598 printf("Num tickles:%u\n", list->tickles.num);
1599 for (i=0;i<list->tickles.num;i++) {
1600 if (port && port != ntohs(list->tickles.connections[i].dst_addr.ip.sin_port)) {
1601 continue;
1603 printf("SRC: %s:%u ", ctdb_addr_to_str(&list->tickles.connections[i].src_addr), ntohs(list->tickles.connections[i].src_addr.ip.sin_port));
1604 printf("DST: %s:%u\n", ctdb_addr_to_str(&list->tickles.connections[i].dst_addr), ntohs(list->tickles.connections[i].dst_addr.ip.sin_port));
1608 talloc_free(list);
1610 return 0;
1614 static int move_ip(struct ctdb_context *ctdb, ctdb_sock_addr *addr, uint32_t pnn)
1616 struct ctdb_all_public_ips *ips;
1617 struct ctdb_public_ip ip;
1618 int i, ret;
1619 uint32_t *nodes;
1620 uint32_t disable_time;
1621 TDB_DATA data;
1622 struct ctdb_node_map *nodemap=NULL;
1623 TALLOC_CTX *tmp_ctx = talloc_new(ctdb);
1625 disable_time = 30;
1626 data.dptr = (uint8_t*)&disable_time;
1627 data.dsize = sizeof(disable_time);
1628 ret = ctdb_client_send_message(ctdb, CTDB_BROADCAST_CONNECTED, CTDB_SRVID_DISABLE_IP_CHECK, data);
1629 if (ret != 0) {
1630 DEBUG(DEBUG_ERR,("Failed to send message to disable ipcheck\n"));
1631 return -1;
1636 /* read the public ip list from the node */
1637 ret = ctdb_ctrl_get_public_ips(ctdb, TIMELIMIT(), pnn, ctdb, &ips);
1638 if (ret != 0) {
1639 DEBUG(DEBUG_ERR, ("Unable to get public ip list from node %u\n", pnn));
1640 talloc_free(tmp_ctx);
1641 return -1;
1644 for (i=0;i<ips->num;i++) {
1645 if (ctdb_same_ip(addr, &ips->ips[i].addr)) {
1646 break;
1649 if (i==ips->num) {
1650 DEBUG(DEBUG_ERR, ("Node %u can not host ip address '%s'\n",
1651 pnn, ctdb_addr_to_str(addr)));
1652 talloc_free(tmp_ctx);
1653 return -1;
1656 ip.pnn = pnn;
1657 ip.addr = *addr;
1659 data.dptr = (uint8_t *)&ip;
1660 data.dsize = sizeof(ip);
1662 ret = ctdb_ctrl_getnodemap(ctdb, TIMELIMIT(), options.pnn, tmp_ctx, &nodemap);
1663 if (ret != 0) {
1664 DEBUG(DEBUG_ERR, ("Unable to get nodemap from node %u\n", options.pnn));
1665 talloc_free(tmp_ctx);
1666 return ret;
1669 nodes = list_of_nodes(ctdb, nodemap, tmp_ctx, NODE_FLAGS_INACTIVE, pnn);
1670 ret = ctdb_client_async_control(ctdb, CTDB_CONTROL_RELEASE_IP,
1671 nodes, 0,
1672 LONGTIMELIMIT(),
1673 false, data,
1674 NULL, NULL,
1675 NULL);
1676 if (ret != 0) {
1677 DEBUG(DEBUG_ERR,("Failed to release IP on nodes\n"));
1678 talloc_free(tmp_ctx);
1679 return -1;
1682 ret = ctdb_ctrl_takeover_ip(ctdb, LONGTIMELIMIT(), pnn, &ip);
1683 if (ret != 0) {
1684 DEBUG(DEBUG_ERR,("Failed to take over IP on node %d\n", pnn));
1685 talloc_free(tmp_ctx);
1686 return -1;
1689 /* update the recovery daemon so it now knows to expect the new
1690 node assignment for this ip.
1692 ret = ctdb_client_send_message(ctdb, CTDB_BROADCAST_CONNECTED, CTDB_SRVID_RECD_UPDATE_IP, data);
1693 if (ret != 0) {
1694 DEBUG(DEBUG_ERR,("Failed to send message to update the ip on the recovery master.\n"));
1695 return -1;
1698 talloc_free(tmp_ctx);
1699 return 0;
1704 * scans all other nodes and returns a pnn for another node that can host this
1705 * ip address or -1
1707 static int
1708 find_other_host_for_public_ip(struct ctdb_context *ctdb, ctdb_sock_addr *addr)
1710 TALLOC_CTX *tmp_ctx = talloc_new(ctdb);
1711 struct ctdb_all_public_ips *ips;
1712 struct ctdb_node_map *nodemap=NULL;
1713 int i, j, ret;
1715 ret = ctdb_ctrl_getnodemap(ctdb, TIMELIMIT(), CTDB_CURRENT_NODE, tmp_ctx, &nodemap);
1716 if (ret != 0) {
1717 DEBUG(DEBUG_ERR, ("Unable to get nodemap from node %u\n", options.pnn));
1718 talloc_free(tmp_ctx);
1719 return ret;
1722 for(i=0;i<nodemap->num;i++){
1723 if (nodemap->nodes[i].flags & NODE_FLAGS_INACTIVE) {
1724 continue;
1726 if (nodemap->nodes[i].pnn == options.pnn) {
1727 continue;
1730 /* read the public ip list from this node */
1731 ret = ctdb_ctrl_get_public_ips(ctdb, TIMELIMIT(), nodemap->nodes[i].pnn, tmp_ctx, &ips);
1732 if (ret != 0) {
1733 DEBUG(DEBUG_ERR, ("Unable to get public ip list from node %u\n", nodemap->nodes[i].pnn));
1734 return -1;
1737 for (j=0;j<ips->num;j++) {
1738 if (ctdb_same_ip(addr, &ips->ips[j].addr)) {
1739 talloc_free(tmp_ctx);
1740 return nodemap->nodes[i].pnn;
1743 talloc_free(ips);
1746 talloc_free(tmp_ctx);
1747 return -1;
1750 /* If pnn is -1 then try to find a node to move IP to... */
1751 static bool try_moveip(struct ctdb_context *ctdb, ctdb_sock_addr *addr, uint32_t pnn)
1753 bool pnn_specified = (pnn == -1 ? false : true);
1754 int retries = 0;
1756 while (retries < 5) {
1757 if (!pnn_specified) {
1758 pnn = find_other_host_for_public_ip(ctdb, addr);
1759 if (pnn == -1) {
1760 return false;
1762 DEBUG(DEBUG_NOTICE,
1763 ("Trying to move public IP to node %u\n", pnn));
1766 if (move_ip(ctdb, addr, pnn) == 0) {
1767 return true;
1770 sleep(3);
1771 retries++;
1774 return false;
1779 move/failover an ip address to a specific node
1781 static int control_moveip(struct ctdb_context *ctdb, int argc, const char **argv)
1783 uint32_t pnn;
1784 ctdb_sock_addr addr;
1786 assert_single_node_only();
1788 if (argc < 2) {
1789 usage();
1790 return -1;
1793 if (parse_ip(argv[0], NULL, 0, &addr) == 0) {
1794 DEBUG(DEBUG_ERR,("Wrongly formed ip address '%s'\n", argv[0]));
1795 return -1;
1799 if (sscanf(argv[1], "%u", &pnn) != 1) {
1800 DEBUG(DEBUG_ERR, ("Badly formed pnn\n"));
1801 return -1;
1804 if (!try_moveip(ctdb, &addr, pnn)) {
1805 DEBUG(DEBUG_ERR,("Failed to move IP to node %d.\n", pnn));
1806 return -1;
1809 return 0;
1812 static int rebalance_node(struct ctdb_context *ctdb, uint32_t pnn)
1814 TDB_DATA data;
1816 data.dptr = (uint8_t *)&pnn;
1817 data.dsize = sizeof(uint32_t);
1818 if (ctdb_client_send_message(ctdb, CTDB_BROADCAST_CONNECTED, CTDB_SRVID_REBALANCE_NODE, data) != 0) {
1819 DEBUG(DEBUG_ERR,
1820 ("Failed to send message to force node %u to be a rebalancing target\n",
1821 pnn));
1822 return -1;
1825 return 0;
1830 rebalance a node by setting it to allow failback and triggering a
1831 takeover run
1833 static int control_rebalancenode(struct ctdb_context *ctdb, int argc, const char **argv)
1835 TALLOC_CTX *tmp_ctx = talloc_new(ctdb);
1836 uint32_t *nodes;
1837 uint32_t pnn_mode;
1838 int i, ret;
1840 assert_single_node_only();
1842 if (argc > 1) {
1843 usage();
1846 /* Determine the nodes where IPs need to be reloaded */
1847 if (!parse_nodestring(ctdb, tmp_ctx, argc == 1 ? argv[0] : NULL,
1848 options.pnn, true, &nodes, &pnn_mode)) {
1849 ret = -1;
1850 goto done;
1853 for (i = 0; i < talloc_array_length(nodes); i++) {
1854 if (!rebalance_node(ctdb, nodes[i])) {
1855 ret = -1;
1859 done:
1860 talloc_free(tmp_ctx);
1861 return ret;
1864 static int rebalance_ip(struct ctdb_context *ctdb, ctdb_sock_addr *addr)
1866 struct ctdb_public_ip ip;
1867 int ret;
1868 uint32_t *nodes;
1869 uint32_t disable_time;
1870 TDB_DATA data;
1871 struct ctdb_node_map *nodemap=NULL;
1872 TALLOC_CTX *tmp_ctx = talloc_new(ctdb);
1874 disable_time = 30;
1875 data.dptr = (uint8_t*)&disable_time;
1876 data.dsize = sizeof(disable_time);
1877 ret = ctdb_client_send_message(ctdb, CTDB_BROADCAST_CONNECTED, CTDB_SRVID_DISABLE_IP_CHECK, data);
1878 if (ret != 0) {
1879 DEBUG(DEBUG_ERR,("Failed to send message to disable ipcheck\n"));
1880 return -1;
1883 ip.pnn = -1;
1884 ip.addr = *addr;
1886 data.dptr = (uint8_t *)&ip;
1887 data.dsize = sizeof(ip);
1889 ret = ctdb_ctrl_getnodemap(ctdb, TIMELIMIT(), options.pnn, tmp_ctx, &nodemap);
1890 if (ret != 0) {
1891 DEBUG(DEBUG_ERR, ("Unable to get nodemap from node %u\n", options.pnn));
1892 talloc_free(tmp_ctx);
1893 return ret;
1896 nodes = list_of_active_nodes(ctdb, nodemap, tmp_ctx, true);
1897 ret = ctdb_client_async_control(ctdb, CTDB_CONTROL_RELEASE_IP,
1898 nodes, 0,
1899 LONGTIMELIMIT(),
1900 false, data,
1901 NULL, NULL,
1902 NULL);
1903 if (ret != 0) {
1904 DEBUG(DEBUG_ERR,("Failed to release IP on nodes\n"));
1905 talloc_free(tmp_ctx);
1906 return -1;
1909 talloc_free(tmp_ctx);
1910 return 0;
1914 release an ip form all nodes and have it re-assigned by recd
1916 static int control_rebalanceip(struct ctdb_context *ctdb, int argc, const char **argv)
1918 ctdb_sock_addr addr;
1920 assert_single_node_only();
1922 if (argc < 1) {
1923 usage();
1924 return -1;
1927 if (parse_ip(argv[0], NULL, 0, &addr) == 0) {
1928 DEBUG(DEBUG_ERR,("Wrongly formed ip address '%s'\n", argv[0]));
1929 return -1;
1932 if (rebalance_ip(ctdb, &addr) != 0) {
1933 DEBUG(DEBUG_ERR,("Error when trying to reassign ip\n"));
1934 return -1;
1937 return 0;
1940 static int getips_store_callback(void *param, void *data)
1942 struct ctdb_public_ip *node_ip = (struct ctdb_public_ip *)data;
1943 struct ctdb_all_public_ips *ips = param;
1944 int i;
1946 i = ips->num++;
1947 ips->ips[i].pnn = node_ip->pnn;
1948 ips->ips[i].addr = node_ip->addr;
1949 return 0;
1952 static int getips_count_callback(void *param, void *data)
1954 uint32_t *count = param;
1956 (*count)++;
1957 return 0;
1960 #define IP_KEYLEN 4
1961 static uint32_t *ip_key(ctdb_sock_addr *ip)
1963 static uint32_t key[IP_KEYLEN];
1965 bzero(key, sizeof(key));
1967 switch (ip->sa.sa_family) {
1968 case AF_INET:
1969 key[0] = ip->ip.sin_addr.s_addr;
1970 break;
1971 case AF_INET6: {
1972 uint32_t *s6_a32 = (uint32_t *)&(ip->ip6.sin6_addr.s6_addr);
1973 key[0] = s6_a32[3];
1974 key[1] = s6_a32[2];
1975 key[2] = s6_a32[1];
1976 key[3] = s6_a32[0];
1977 break;
1979 default:
1980 DEBUG(DEBUG_ERR, (__location__ " ERROR, unknown family passed :%u\n", ip->sa.sa_family));
1981 return key;
1984 return key;
1987 static void *add_ip_callback(void *parm, void *data)
1989 return parm;
1992 static int
1993 control_get_all_public_ips(struct ctdb_context *ctdb, TALLOC_CTX *tmp_ctx, struct ctdb_all_public_ips **ips)
1995 struct ctdb_all_public_ips *tmp_ips;
1996 struct ctdb_node_map *nodemap=NULL;
1997 trbt_tree_t *ip_tree;
1998 int i, j, len, ret;
1999 uint32_t count;
2001 ret = ctdb_ctrl_getnodemap(ctdb, TIMELIMIT(), CTDB_CURRENT_NODE, tmp_ctx, &nodemap);
2002 if (ret != 0) {
2003 DEBUG(DEBUG_ERR, ("Unable to get nodemap from node %u\n", options.pnn));
2004 return ret;
2007 ip_tree = trbt_create(tmp_ctx, 0);
2009 for(i=0;i<nodemap->num;i++){
2010 if (nodemap->nodes[i].flags & NODE_FLAGS_DELETED) {
2011 continue;
2013 if (nodemap->nodes[i].flags & NODE_FLAGS_DISCONNECTED) {
2014 continue;
2017 /* read the public ip list from this node */
2018 ret = ctdb_ctrl_get_public_ips(ctdb, TIMELIMIT(), nodemap->nodes[i].pnn, tmp_ctx, &tmp_ips);
2019 if (ret != 0) {
2020 DEBUG(DEBUG_ERR, ("Unable to get public ip list from node %u\n", nodemap->nodes[i].pnn));
2021 return -1;
2024 for (j=0; j<tmp_ips->num;j++) {
2025 struct ctdb_public_ip *node_ip;
2027 node_ip = talloc(tmp_ctx, struct ctdb_public_ip);
2028 node_ip->pnn = tmp_ips->ips[j].pnn;
2029 node_ip->addr = tmp_ips->ips[j].addr;
2031 trbt_insertarray32_callback(ip_tree,
2032 IP_KEYLEN, ip_key(&tmp_ips->ips[j].addr),
2033 add_ip_callback,
2034 node_ip);
2036 talloc_free(tmp_ips);
2039 /* traverse */
2040 count = 0;
2041 trbt_traversearray32(ip_tree, IP_KEYLEN, getips_count_callback, &count);
2043 len = offsetof(struct ctdb_all_public_ips, ips) +
2044 count*sizeof(struct ctdb_public_ip);
2045 tmp_ips = talloc_zero_size(tmp_ctx, len);
2046 trbt_traversearray32(ip_tree, IP_KEYLEN, getips_store_callback, tmp_ips);
2048 *ips = tmp_ips;
2050 return 0;
2054 static void ctdb_every_second(struct event_context *ev, struct timed_event *te, struct timeval t, void *p)
2056 struct ctdb_context *ctdb = talloc_get_type(p, struct ctdb_context);
2058 event_add_timed(ctdb->ev, ctdb,
2059 timeval_current_ofs(1, 0),
2060 ctdb_every_second, ctdb);
2063 struct srvid_reply_handler_data {
2064 bool done;
2065 bool wait_for_all;
2066 uint32_t *nodes;
2067 const char *srvid_str;
2070 static void srvid_broadcast_reply_handler(struct ctdb_context *ctdb,
2071 uint64_t srvid,
2072 TDB_DATA data,
2073 void *private_data)
2075 struct srvid_reply_handler_data *d =
2076 (struct srvid_reply_handler_data *)private_data;
2077 int i;
2078 int32_t ret;
2080 if (data.dsize != sizeof(ret)) {
2081 DEBUG(DEBUG_ERR, (__location__ " Wrong reply size\n"));
2082 return;
2085 /* ret will be a PNN (i.e. >=0) on success, or negative on error */
2086 ret = *(int32_t *)data.dptr;
2087 if (ret < 0) {
2088 DEBUG(DEBUG_ERR,
2089 ("%s failed with result %d\n", d->srvid_str, ret));
2090 return;
2093 if (!d->wait_for_all) {
2094 d->done = true;
2095 return;
2098 /* Wait for all replies */
2099 d->done = true;
2100 for (i = 0; i < talloc_array_length(d->nodes); i++) {
2101 if (d->nodes[i] == ret) {
2102 DEBUG(DEBUG_INFO,
2103 ("%s reply received from node %u\n",
2104 d->srvid_str, ret));
2105 d->nodes[i] = -1;
2107 if (d->nodes[i] != -1) {
2108 /* Found a node that hasn't yet replied */
2109 d->done = false;
2114 /* Broadcast the given SRVID to all connected nodes. Wait for 1 reply
2115 * or replies from all connected nodes. arg is the data argument to
2116 * pass in the srvid_request structure - pass 0 if this isn't needed.
2118 static int srvid_broadcast(struct ctdb_context *ctdb,
2119 uint64_t srvid, uint32_t arg,
2120 const char *srvid_str, bool wait_for_all)
2122 int ret;
2123 TDB_DATA data;
2124 struct srvid_request request;
2125 struct srvid_reply_handler_data reply_data;
2126 struct timeval tv;
2128 ZERO_STRUCT(request);
2130 /* Time ticks to enable timeouts to be processed */
2131 event_add_timed(ctdb->ev, ctdb,
2132 timeval_current_ofs(1, 0),
2133 ctdb_every_second, ctdb);
2135 request.pnn = ctdb_get_pnn(ctdb);
2136 request.srvid = getpid();
2137 request.data = arg;
2139 /* Register message port for reply from recovery master */
2140 ctdb_client_set_message_handler(ctdb, request.srvid,
2141 srvid_broadcast_reply_handler,
2142 &reply_data);
2144 data.dptr = (uint8_t *)&request;
2145 data.dsize = sizeof(request);
2147 reply_data.wait_for_all = wait_for_all;
2148 reply_data.nodes = NULL;
2149 reply_data.srvid_str = srvid_str;
2151 again:
2152 reply_data.done = false;
2154 if (wait_for_all) {
2155 struct ctdb_node_map *nodemap;
2157 ret = ctdb_ctrl_getnodemap(ctdb, TIMELIMIT(),
2158 CTDB_CURRENT_NODE, ctdb, &nodemap);
2159 if (ret != 0) {
2160 DEBUG(DEBUG_ERR,
2161 ("Unable to get nodemap from current node, try again\n"));
2162 sleep(1);
2163 goto again;
2166 if (reply_data.nodes != NULL) {
2167 talloc_free(reply_data.nodes);
2169 reply_data.nodes = list_of_connected_nodes(ctdb, nodemap,
2170 NULL, true);
2172 talloc_free(nodemap);
2175 /* Send to all connected nodes. Only recmaster replies */
2176 ret = ctdb_client_send_message(ctdb, CTDB_BROADCAST_CONNECTED,
2177 srvid, data);
2178 if (ret != 0) {
2179 /* This can only happen if the socket is closed and
2180 * there's no way to recover from that, so don't try
2181 * again.
2183 DEBUG(DEBUG_ERR,
2184 ("Failed to send %s request to connected nodes\n",
2185 srvid_str));
2186 return -1;
2189 tv = timeval_current();
2190 /* This loop terminates the reply is received */
2191 while (timeval_elapsed(&tv) < 5.0 && !reply_data.done) {
2192 event_loop_once(ctdb->ev);
2195 if (!reply_data.done) {
2196 DEBUG(DEBUG_NOTICE,
2197 ("Still waiting for confirmation of %s\n", srvid_str));
2198 sleep(1);
2199 goto again;
2202 ctdb_client_remove_message_handler(ctdb, request.srvid, &reply_data);
2204 talloc_free(reply_data.nodes);
2206 return 0;
2209 static int ipreallocate(struct ctdb_context *ctdb)
2211 return srvid_broadcast(ctdb, CTDB_SRVID_TAKEOVER_RUN, 0,
2212 "IP reallocation", false);
2216 static int control_ipreallocate(struct ctdb_context *ctdb, int argc, const char **argv)
2218 return ipreallocate(ctdb);
2222 add a public ip address to a node
2224 static int control_addip(struct ctdb_context *ctdb, int argc, const char **argv)
2226 int i, ret;
2227 int len, retries = 0;
2228 unsigned mask;
2229 ctdb_sock_addr addr;
2230 struct ctdb_control_ip_iface *pub;
2231 TALLOC_CTX *tmp_ctx = talloc_new(ctdb);
2232 struct ctdb_all_public_ips *ips;
2235 if (argc != 2) {
2236 talloc_free(tmp_ctx);
2237 usage();
2240 if (!parse_ip_mask(argv[0], argv[1], &addr, &mask)) {
2241 DEBUG(DEBUG_ERR, ("Badly formed ip/mask : %s\n", argv[0]));
2242 talloc_free(tmp_ctx);
2243 return -1;
2246 /* read the public ip list from the node */
2247 ret = ctdb_ctrl_get_public_ips(ctdb, TIMELIMIT(), options.pnn, tmp_ctx, &ips);
2248 if (ret != 0) {
2249 DEBUG(DEBUG_ERR, ("Unable to get public ip list from node %u\n", options.pnn));
2250 talloc_free(tmp_ctx);
2251 return -1;
2253 for (i=0;i<ips->num;i++) {
2254 if (ctdb_same_ip(&addr, &ips->ips[i].addr)) {
2255 DEBUG(DEBUG_ERR,("Can not add ip to node. Node already hosts this ip\n"));
2256 return 0;
2262 /* Dont timeout. This command waits for an ip reallocation
2263 which sometimes can take wuite a while if there has
2264 been a recent recovery
2266 alarm(0);
2268 len = offsetof(struct ctdb_control_ip_iface, iface) + strlen(argv[1]) + 1;
2269 pub = talloc_size(tmp_ctx, len);
2270 CTDB_NO_MEMORY(ctdb, pub);
2272 pub->addr = addr;
2273 pub->mask = mask;
2274 pub->len = strlen(argv[1])+1;
2275 memcpy(&pub->iface[0], argv[1], strlen(argv[1])+1);
2277 do {
2278 ret = ctdb_ctrl_add_public_ip(ctdb, TIMELIMIT(), options.pnn, pub);
2279 if (ret != 0) {
2280 DEBUG(DEBUG_ERR, ("Unable to add public ip to node %u. Wait 3 seconds and try again.\n", options.pnn));
2281 sleep(3);
2282 retries++;
2284 } while (retries < 5 && ret != 0);
2285 if (ret != 0) {
2286 DEBUG(DEBUG_ERR, ("Unable to add public ip to node %u. Giving up.\n", options.pnn));
2287 talloc_free(tmp_ctx);
2288 return ret;
2291 if (rebalance_node(ctdb, options.pnn) != 0) {
2292 DEBUG(DEBUG_ERR,("Error when trying to rebalance node\n"));
2293 return ret;
2296 talloc_free(tmp_ctx);
2297 return 0;
2301 add a public ip address to a node
2303 static int control_ipiface(struct ctdb_context *ctdb, int argc, const char **argv)
2305 ctdb_sock_addr addr;
2307 if (argc != 1) {
2308 usage();
2311 if (!parse_ip(argv[0], NULL, 0, &addr)) {
2312 printf("Badly formed ip : %s\n", argv[0]);
2313 return -1;
2316 printf("IP on interface %s\n", ctdb_sys_find_ifname(&addr));
2318 return 0;
2321 static int control_delip(struct ctdb_context *ctdb, int argc, const char **argv);
2323 static int control_delip_all(struct ctdb_context *ctdb, int argc, const char **argv, ctdb_sock_addr *addr)
2325 TALLOC_CTX *tmp_ctx = talloc_new(ctdb);
2326 struct ctdb_node_map *nodemap=NULL;
2327 struct ctdb_all_public_ips *ips;
2328 int ret, i, j;
2330 ret = ctdb_ctrl_getnodemap(ctdb, TIMELIMIT(), CTDB_CURRENT_NODE, tmp_ctx, &nodemap);
2331 if (ret != 0) {
2332 DEBUG(DEBUG_ERR, ("Unable to get nodemap from current node\n"));
2333 return ret;
2336 /* remove it from the nodes that are not hosting the ip currently */
2337 for(i=0;i<nodemap->num;i++){
2338 if (nodemap->nodes[i].flags & NODE_FLAGS_INACTIVE) {
2339 continue;
2341 ret = ctdb_ctrl_get_public_ips(ctdb, TIMELIMIT(), nodemap->nodes[i].pnn, tmp_ctx, &ips);
2342 if (ret != 0) {
2343 DEBUG(DEBUG_ERR, ("Unable to get public ip list from node %d\n", nodemap->nodes[i].pnn));
2344 continue;
2347 for (j=0;j<ips->num;j++) {
2348 if (ctdb_same_ip(addr, &ips->ips[j].addr)) {
2349 break;
2352 if (j==ips->num) {
2353 continue;
2356 if (ips->ips[j].pnn == nodemap->nodes[i].pnn) {
2357 continue;
2360 options.pnn = nodemap->nodes[i].pnn;
2361 control_delip(ctdb, argc, argv);
2365 /* remove it from every node (also the one hosting it) */
2366 for(i=0;i<nodemap->num;i++){
2367 if (nodemap->nodes[i].flags & NODE_FLAGS_INACTIVE) {
2368 continue;
2370 ret = ctdb_ctrl_get_public_ips(ctdb, TIMELIMIT(), nodemap->nodes[i].pnn, tmp_ctx, &ips);
2371 if (ret != 0) {
2372 DEBUG(DEBUG_ERR, ("Unable to get public ip list from node %d\n", nodemap->nodes[i].pnn));
2373 continue;
2376 for (j=0;j<ips->num;j++) {
2377 if (ctdb_same_ip(addr, &ips->ips[j].addr)) {
2378 break;
2381 if (j==ips->num) {
2382 continue;
2385 options.pnn = nodemap->nodes[i].pnn;
2386 control_delip(ctdb, argc, argv);
2389 talloc_free(tmp_ctx);
2390 return 0;
2394 delete a public ip address from a node
2396 static int control_delip(struct ctdb_context *ctdb, int argc, const char **argv)
2398 int i, ret;
2399 ctdb_sock_addr addr;
2400 struct ctdb_control_ip_iface pub;
2401 TALLOC_CTX *tmp_ctx = talloc_new(ctdb);
2402 struct ctdb_all_public_ips *ips;
2404 if (argc != 1) {
2405 talloc_free(tmp_ctx);
2406 usage();
2409 if (parse_ip(argv[0], NULL, 0, &addr) == 0) {
2410 DEBUG(DEBUG_ERR,("Wrongly formed ip address '%s'\n", argv[0]));
2411 return -1;
2414 if (options.pnn == CTDB_BROADCAST_ALL) {
2415 return control_delip_all(ctdb, argc, argv, &addr);
2418 pub.addr = addr;
2419 pub.mask = 0;
2420 pub.len = 0;
2422 ret = ctdb_ctrl_get_public_ips(ctdb, TIMELIMIT(), options.pnn, tmp_ctx, &ips);
2423 if (ret != 0) {
2424 DEBUG(DEBUG_ERR, ("Unable to get public ip list from cluster\n"));
2425 talloc_free(tmp_ctx);
2426 return ret;
2429 for (i=0;i<ips->num;i++) {
2430 if (ctdb_same_ip(&addr, &ips->ips[i].addr)) {
2431 break;
2435 if (i==ips->num) {
2436 DEBUG(DEBUG_ERR, ("This node does not support this public address '%s'\n",
2437 ctdb_addr_to_str(&addr)));
2438 talloc_free(tmp_ctx);
2439 return -1;
2442 /* This is an optimisation. If this node is hosting the IP
2443 * then try to move it somewhere else without invoking a full
2444 * takeover run. We don't care if this doesn't work!
2446 if (ips->ips[i].pnn == options.pnn) {
2447 (void) try_moveip(ctdb, &addr, -1);
2450 ret = ctdb_ctrl_del_public_ip(ctdb, TIMELIMIT(), options.pnn, &pub);
2451 if (ret != 0) {
2452 DEBUG(DEBUG_ERR, ("Unable to del public ip from node %u\n", options.pnn));
2453 talloc_free(tmp_ctx);
2454 return ret;
2457 talloc_free(tmp_ctx);
2458 return 0;
2461 static int kill_tcp_from_file(struct ctdb_context *ctdb,
2462 int argc, const char **argv)
2464 struct ctdb_control_killtcp *killtcp;
2465 int max_entries, current, i;
2466 struct timeval timeout;
2467 char line[128], src[128], dst[128];
2468 int linenum;
2469 TDB_DATA data;
2470 struct client_async_data *async_data;
2471 struct ctdb_client_control_state *state;
2473 if (argc != 0) {
2474 usage();
2477 linenum = 1;
2478 killtcp = NULL;
2479 max_entries = 0;
2480 current = 0;
2481 while (!feof(stdin)) {
2482 if (fgets(line, sizeof(line), stdin) == NULL) {
2483 continue;
2486 /* Silently skip empty lines */
2487 if (line[0] == '\n') {
2488 continue;
2491 if (sscanf(line, "%s %s\n", src, dst) != 2) {
2492 DEBUG(DEBUG_ERR, ("Bad line [%d]: '%s'\n",
2493 linenum, line));
2494 talloc_free(killtcp);
2495 return -1;
2498 if (current >= max_entries) {
2499 max_entries += 1024;
2500 killtcp = talloc_realloc(ctdb, killtcp,
2501 struct ctdb_control_killtcp,
2502 max_entries);
2503 CTDB_NO_MEMORY(ctdb, killtcp);
2506 if (!parse_ip_port(src, &killtcp[current].src_addr)) {
2507 DEBUG(DEBUG_ERR, ("Bad IP:port on line [%d]: '%s'\n",
2508 linenum, src));
2509 talloc_free(killtcp);
2510 return -1;
2513 if (!parse_ip_port(dst, &killtcp[current].dst_addr)) {
2514 DEBUG(DEBUG_ERR, ("Bad IP:port on line [%d]: '%s'\n",
2515 linenum, dst));
2516 talloc_free(killtcp);
2517 return -1;
2520 current++;
2523 async_data = talloc_zero(ctdb, struct client_async_data);
2524 if (async_data == NULL) {
2525 talloc_free(killtcp);
2526 return -1;
2529 for (i = 0; i < current; i++) {
2531 data.dsize = sizeof(struct ctdb_control_killtcp);
2532 data.dptr = (unsigned char *)&killtcp[i];
2534 timeout = TIMELIMIT();
2535 state = ctdb_control_send(ctdb, options.pnn, 0,
2536 CTDB_CONTROL_KILL_TCP, 0, data,
2537 async_data, &timeout, NULL);
2539 if (state == NULL) {
2540 DEBUG(DEBUG_ERR,
2541 ("Failed to call async killtcp control to node %u\n",
2542 options.pnn));
2543 talloc_free(killtcp);
2544 return -1;
2547 ctdb_client_async_add(async_data, state);
2550 if (ctdb_client_async_wait(ctdb, async_data) != 0) {
2551 DEBUG(DEBUG_ERR,("killtcp failed\n"));
2552 talloc_free(killtcp);
2553 return -1;
2556 talloc_free(killtcp);
2557 return 0;
2562 kill a tcp connection
2564 static int kill_tcp(struct ctdb_context *ctdb, int argc, const char **argv)
2566 int ret;
2567 struct ctdb_control_killtcp killtcp;
2569 assert_single_node_only();
2571 if (argc == 0) {
2572 return kill_tcp_from_file(ctdb, argc, argv);
2575 if (argc < 2) {
2576 usage();
2579 if (!parse_ip_port(argv[0], &killtcp.src_addr)) {
2580 DEBUG(DEBUG_ERR, ("Bad IP:port '%s'\n", argv[0]));
2581 return -1;
2584 if (!parse_ip_port(argv[1], &killtcp.dst_addr)) {
2585 DEBUG(DEBUG_ERR, ("Bad IP:port '%s'\n", argv[1]));
2586 return -1;
2589 ret = ctdb_ctrl_killtcp(ctdb, TIMELIMIT(), options.pnn, &killtcp);
2590 if (ret != 0) {
2591 DEBUG(DEBUG_ERR, ("Unable to killtcp from node %u\n", options.pnn));
2592 return ret;
2595 return 0;
2600 send a gratious arp
2602 static int control_gratious_arp(struct ctdb_context *ctdb, int argc, const char **argv)
2604 int ret;
2605 ctdb_sock_addr addr;
2607 assert_single_node_only();
2609 if (argc < 2) {
2610 usage();
2613 if (!parse_ip(argv[0], NULL, 0, &addr)) {
2614 DEBUG(DEBUG_ERR, ("Bad IP '%s'\n", argv[0]));
2615 return -1;
2618 ret = ctdb_ctrl_gratious_arp(ctdb, TIMELIMIT(), options.pnn, &addr, argv[1]);
2619 if (ret != 0) {
2620 DEBUG(DEBUG_ERR, ("Unable to send gratious_arp from node %u\n", options.pnn));
2621 return ret;
2624 return 0;
2628 register a server id
2630 static int regsrvid(struct ctdb_context *ctdb, int argc, const char **argv)
2632 int ret;
2633 struct ctdb_server_id server_id;
2635 if (argc < 3) {
2636 usage();
2639 server_id.pnn = strtoul(argv[0], NULL, 0);
2640 server_id.type = strtoul(argv[1], NULL, 0);
2641 server_id.server_id = strtoul(argv[2], NULL, 0);
2643 ret = ctdb_ctrl_register_server_id(ctdb, TIMELIMIT(), &server_id);
2644 if (ret != 0) {
2645 DEBUG(DEBUG_ERR, ("Unable to register server_id from node %u\n", options.pnn));
2646 return ret;
2648 DEBUG(DEBUG_ERR,("Srvid registered. Sleeping for 999 seconds\n"));
2649 sleep(999);
2650 return -1;
2654 unregister a server id
2656 static int unregsrvid(struct ctdb_context *ctdb, int argc, const char **argv)
2658 int ret;
2659 struct ctdb_server_id server_id;
2661 if (argc < 3) {
2662 usage();
2665 server_id.pnn = strtoul(argv[0], NULL, 0);
2666 server_id.type = strtoul(argv[1], NULL, 0);
2667 server_id.server_id = strtoul(argv[2], NULL, 0);
2669 ret = ctdb_ctrl_unregister_server_id(ctdb, TIMELIMIT(), &server_id);
2670 if (ret != 0) {
2671 DEBUG(DEBUG_ERR, ("Unable to unregister server_id from node %u\n", options.pnn));
2672 return ret;
2674 return -1;
2678 check if a server id exists
2680 static int chksrvid(struct ctdb_context *ctdb, int argc, const char **argv)
2682 uint32_t status;
2683 int ret;
2684 struct ctdb_server_id server_id;
2686 if (argc < 3) {
2687 usage();
2690 server_id.pnn = strtoul(argv[0], NULL, 0);
2691 server_id.type = strtoul(argv[1], NULL, 0);
2692 server_id.server_id = strtoul(argv[2], NULL, 0);
2694 ret = ctdb_ctrl_check_server_id(ctdb, TIMELIMIT(), options.pnn, &server_id, &status);
2695 if (ret != 0) {
2696 DEBUG(DEBUG_ERR, ("Unable to check server_id from node %u\n", options.pnn));
2697 return ret;
2700 if (status) {
2701 printf("Server id %d:%d:%d EXISTS\n", server_id.pnn, server_id.type, server_id.server_id);
2702 } else {
2703 printf("Server id %d:%d:%d does NOT exist\n", server_id.pnn, server_id.type, server_id.server_id);
2705 return 0;
2709 get a list of all server ids that are registered on a node
2711 static int getsrvids(struct ctdb_context *ctdb, int argc, const char **argv)
2713 int i, ret;
2714 struct ctdb_server_id_list *server_ids;
2716 ret = ctdb_ctrl_get_server_id_list(ctdb, ctdb, TIMELIMIT(), options.pnn, &server_ids);
2717 if (ret != 0) {
2718 DEBUG(DEBUG_ERR, ("Unable to get server_id list from node %u\n", options.pnn));
2719 return ret;
2722 for (i=0; i<server_ids->num; i++) {
2723 printf("Server id %d:%d:%d\n",
2724 server_ids->server_ids[i].pnn,
2725 server_ids->server_ids[i].type,
2726 server_ids->server_ids[i].server_id);
2729 return -1;
2733 check if a server id exists
2735 static int check_srvids(struct ctdb_context *ctdb, int argc, const char **argv)
2737 TALLOC_CTX *tmp_ctx = talloc_new(NULL);
2738 uint64_t *ids;
2739 uint8_t *result;
2740 int i;
2742 if (argc < 1) {
2743 talloc_free(tmp_ctx);
2744 usage();
2747 ids = talloc_array(tmp_ctx, uint64_t, argc);
2748 result = talloc_array(tmp_ctx, uint8_t, argc);
2750 for (i = 0; i < argc; i++) {
2751 ids[i] = strtoull(argv[i], NULL, 0);
2754 if (!ctdb_client_check_message_handlers(ctdb, ids, argc, result)) {
2755 DEBUG(DEBUG_ERR, ("Unable to check server_id from node %u\n",
2756 options.pnn));
2757 talloc_free(tmp_ctx);
2758 return -1;
2761 for (i=0; i < argc; i++) {
2762 printf("Server id %d:%llu %s\n", options.pnn, (long long)ids[i],
2763 result[i] ? "exists" : "does not exist");
2766 talloc_free(tmp_ctx);
2767 return 0;
2771 send a tcp tickle ack
2773 static int tickle_tcp(struct ctdb_context *ctdb, int argc, const char **argv)
2775 int ret;
2776 ctdb_sock_addr src, dst;
2778 if (argc < 2) {
2779 usage();
2782 if (!parse_ip_port(argv[0], &src)) {
2783 DEBUG(DEBUG_ERR, ("Bad IP:port '%s'\n", argv[0]));
2784 return -1;
2787 if (!parse_ip_port(argv[1], &dst)) {
2788 DEBUG(DEBUG_ERR, ("Bad IP:port '%s'\n", argv[1]));
2789 return -1;
2792 ret = ctdb_sys_send_tcp(&src, &dst, 0, 0, 0);
2793 if (ret==0) {
2794 return 0;
2796 DEBUG(DEBUG_ERR, ("Error while sending tickle ack\n"));
2798 return -1;
2803 display public ip status
2805 static int control_ip(struct ctdb_context *ctdb, int argc, const char **argv)
2807 int i, ret;
2808 TALLOC_CTX *tmp_ctx = talloc_new(ctdb);
2809 struct ctdb_all_public_ips *ips;
2811 if (options.pnn == CTDB_BROADCAST_ALL) {
2812 /* read the list of public ips from all nodes */
2813 ret = control_get_all_public_ips(ctdb, tmp_ctx, &ips);
2814 } else {
2815 /* read the public ip list from this node */
2816 ret = ctdb_ctrl_get_public_ips(ctdb, TIMELIMIT(), options.pnn, tmp_ctx, &ips);
2818 if (ret != 0) {
2819 DEBUG(DEBUG_ERR, ("Unable to get public ips from node %u\n", options.pnn));
2820 talloc_free(tmp_ctx);
2821 return ret;
2824 if (options.machinereadable){
2825 printf(":Public IP:Node:");
2826 if (options.verbose){
2827 printf("ActiveInterface:AvailableInterfaces:ConfiguredInterfaces:");
2829 printf("\n");
2830 } else {
2831 if (options.pnn == CTDB_BROADCAST_ALL) {
2832 printf("Public IPs on ALL nodes\n");
2833 } else {
2834 printf("Public IPs on node %u\n", options.pnn);
2838 for (i=1;i<=ips->num;i++) {
2839 struct ctdb_control_public_ip_info *info = NULL;
2840 int32_t pnn;
2841 char *aciface = NULL;
2842 char *avifaces = NULL;
2843 char *cifaces = NULL;
2845 if (options.pnn == CTDB_BROADCAST_ALL) {
2846 pnn = ips->ips[ips->num-i].pnn;
2847 } else {
2848 pnn = options.pnn;
2851 if (pnn != -1) {
2852 ret = ctdb_ctrl_get_public_ip_info(ctdb, TIMELIMIT(), pnn, ctdb,
2853 &ips->ips[ips->num-i].addr, &info);
2854 } else {
2855 ret = -1;
2858 if (ret == 0) {
2859 int j;
2860 for (j=0; j < info->num; j++) {
2861 if (cifaces == NULL) {
2862 cifaces = talloc_strdup(info,
2863 info->ifaces[j].name);
2864 } else {
2865 cifaces = talloc_asprintf_append(cifaces,
2866 ",%s",
2867 info->ifaces[j].name);
2870 if (info->active_idx == j) {
2871 aciface = info->ifaces[j].name;
2874 if (info->ifaces[j].link_state == 0) {
2875 continue;
2878 if (avifaces == NULL) {
2879 avifaces = talloc_strdup(info, info->ifaces[j].name);
2880 } else {
2881 avifaces = talloc_asprintf_append(avifaces,
2882 ",%s",
2883 info->ifaces[j].name);
2888 if (options.machinereadable){
2889 printf(":%s:%d:",
2890 ctdb_addr_to_str(&ips->ips[ips->num-i].addr),
2891 ips->ips[ips->num-i].pnn);
2892 if (options.verbose){
2893 printf("%s:%s:%s:",
2894 aciface?aciface:"",
2895 avifaces?avifaces:"",
2896 cifaces?cifaces:"");
2898 printf("\n");
2899 } else {
2900 if (options.verbose) {
2901 printf("%s node[%d] active[%s] available[%s] configured[%s]\n",
2902 ctdb_addr_to_str(&ips->ips[ips->num-i].addr),
2903 ips->ips[ips->num-i].pnn,
2904 aciface?aciface:"",
2905 avifaces?avifaces:"",
2906 cifaces?cifaces:"");
2907 } else {
2908 printf("%s %d\n",
2909 ctdb_addr_to_str(&ips->ips[ips->num-i].addr),
2910 ips->ips[ips->num-i].pnn);
2913 talloc_free(info);
2916 talloc_free(tmp_ctx);
2917 return 0;
2921 public ip info
2923 static int control_ipinfo(struct ctdb_context *ctdb, int argc, const char **argv)
2925 int i, ret;
2926 ctdb_sock_addr addr;
2927 TALLOC_CTX *tmp_ctx = talloc_new(ctdb);
2928 struct ctdb_control_public_ip_info *info;
2930 if (argc != 1) {
2931 talloc_free(tmp_ctx);
2932 usage();
2935 if (parse_ip(argv[0], NULL, 0, &addr) == 0) {
2936 DEBUG(DEBUG_ERR,("Wrongly formed ip address '%s'\n", argv[0]));
2937 return -1;
2940 /* read the public ip info from this node */
2941 ret = ctdb_ctrl_get_public_ip_info(ctdb, TIMELIMIT(), options.pnn,
2942 tmp_ctx, &addr, &info);
2943 if (ret != 0) {
2944 DEBUG(DEBUG_ERR, ("Unable to get public ip[%s]info from node %u\n",
2945 argv[0], options.pnn));
2946 talloc_free(tmp_ctx);
2947 return ret;
2950 printf("Public IP[%s] info on node %u\n",
2951 ctdb_addr_to_str(&info->ip.addr),
2952 options.pnn);
2954 printf("IP:%s\nCurrentNode:%d\nNumInterfaces:%u\n",
2955 ctdb_addr_to_str(&info->ip.addr),
2956 info->ip.pnn, info->num);
2958 for (i=0; i<info->num; i++) {
2959 info->ifaces[i].name[CTDB_IFACE_SIZE] = '\0';
2961 printf("Interface[%u]: Name:%s Link:%s References:%u%s\n",
2962 i+1, info->ifaces[i].name,
2963 info->ifaces[i].link_state?"up":"down",
2964 (unsigned int)info->ifaces[i].references,
2965 (i==info->active_idx)?" (active)":"");
2968 talloc_free(tmp_ctx);
2969 return 0;
2973 display interfaces status
2975 static int control_ifaces(struct ctdb_context *ctdb, int argc, const char **argv)
2977 TALLOC_CTX *tmp_ctx = talloc_new(ctdb);
2978 int i;
2979 struct ctdb_control_get_ifaces *ifaces;
2980 int ret;
2982 /* read the public ip list from this node */
2983 ret = ctdb_ctrl_get_ifaces(ctdb, TIMELIMIT(), options.pnn, tmp_ctx, &ifaces);
2984 if (ret != 0) {
2985 DEBUG(DEBUG_ERR, ("Unable to get interfaces from node %u\n",
2986 options.pnn));
2987 talloc_free(tmp_ctx);
2988 return -1;
2991 if (options.machinereadable){
2992 printf(":Name:LinkStatus:References:\n");
2993 } else {
2994 printf("Interfaces on node %u\n", options.pnn);
2997 for (i=0; i<ifaces->num; i++) {
2998 if (options.machinereadable){
2999 printf(":%s:%s:%u\n",
3000 ifaces->ifaces[i].name,
3001 ifaces->ifaces[i].link_state?"1":"0",
3002 (unsigned int)ifaces->ifaces[i].references);
3003 } else {
3004 printf("name:%s link:%s references:%u\n",
3005 ifaces->ifaces[i].name,
3006 ifaces->ifaces[i].link_state?"up":"down",
3007 (unsigned int)ifaces->ifaces[i].references);
3011 talloc_free(tmp_ctx);
3012 return 0;
3017 set link status of an interface
3019 static int control_setifacelink(struct ctdb_context *ctdb, int argc, const char **argv)
3021 int ret;
3022 TALLOC_CTX *tmp_ctx = talloc_new(ctdb);
3023 struct ctdb_control_iface_info info;
3025 ZERO_STRUCT(info);
3027 if (argc != 2) {
3028 usage();
3031 if (strlen(argv[0]) > CTDB_IFACE_SIZE) {
3032 DEBUG(DEBUG_ERR, ("interfaces name '%s' too long\n",
3033 argv[0]));
3034 talloc_free(tmp_ctx);
3035 return -1;
3037 strcpy(info.name, argv[0]);
3039 if (strcmp(argv[1], "up") == 0) {
3040 info.link_state = 1;
3041 } else if (strcmp(argv[1], "down") == 0) {
3042 info.link_state = 0;
3043 } else {
3044 DEBUG(DEBUG_ERR, ("link state invalid '%s' should be 'up' or 'down'\n",
3045 argv[1]));
3046 talloc_free(tmp_ctx);
3047 return -1;
3050 /* read the public ip list from this node */
3051 ret = ctdb_ctrl_set_iface_link(ctdb, TIMELIMIT(), options.pnn,
3052 tmp_ctx, &info);
3053 if (ret != 0) {
3054 DEBUG(DEBUG_ERR, ("Unable to set link state for interfaces %s node %u\n",
3055 argv[0], options.pnn));
3056 talloc_free(tmp_ctx);
3057 return ret;
3060 talloc_free(tmp_ctx);
3061 return 0;
3065 display pid of a ctdb daemon
3067 static int control_getpid(struct ctdb_context *ctdb, int argc, const char **argv)
3069 uint32_t pid;
3070 int ret;
3072 ret = ctdb_ctrl_getpid(ctdb, TIMELIMIT(), options.pnn, &pid);
3073 if (ret != 0) {
3074 DEBUG(DEBUG_ERR, ("Unable to get daemon pid from node %u\n", options.pnn));
3075 return ret;
3077 printf("Pid:%d\n", pid);
3079 return 0;
3082 typedef bool update_flags_handler_t(struct ctdb_context *ctdb, void *data);
3084 static int update_flags_and_ipreallocate(struct ctdb_context *ctdb,
3085 void *data,
3086 update_flags_handler_t handler,
3087 uint32_t flag,
3088 const char *desc,
3089 bool set_flag)
3091 struct ctdb_node_map *nodemap = NULL;
3092 bool flag_is_set;
3093 int ret;
3095 /* Check if the node is already in the desired state */
3096 ret = ctdb_ctrl_getnodemap(ctdb, TIMELIMIT(), CTDB_CURRENT_NODE, ctdb, &nodemap);
3097 if (ret != 0) {
3098 DEBUG(DEBUG_ERR, ("Unable to get nodemap from local node\n"));
3099 exit(10);
3101 flag_is_set = nodemap->nodes[options.pnn].flags & flag;
3102 if (set_flag == flag_is_set) {
3103 DEBUG(DEBUG_NOTICE, ("Node %d is %s %s\n", options.pnn,
3104 (set_flag ? "already" : "not"), desc));
3105 return 0;
3108 do {
3109 if (!handler(ctdb, data)) {
3110 DEBUG(DEBUG_WARNING,
3111 ("Failed to send control to set state %s on node %u, try again\n",
3112 desc, options.pnn));
3115 sleep(1);
3117 /* Read the nodemap and verify the change took effect.
3118 * Even if the above control/hanlder timed out then it
3119 * could still have worked!
3121 ret = ctdb_ctrl_getnodemap(ctdb, TIMELIMIT(), CTDB_CURRENT_NODE,
3122 ctdb, &nodemap);
3123 if (ret != 0) {
3124 DEBUG(DEBUG_WARNING,
3125 ("Unable to get nodemap from local node, try again\n"));
3127 flag_is_set = nodemap->nodes[options.pnn].flags & flag;
3128 } while (nodemap == NULL || (set_flag != flag_is_set));
3130 return ipreallocate(ctdb);
3133 /* Administratively disable a node */
3134 static bool update_flags_disabled(struct ctdb_context *ctdb, void *data)
3136 int ret;
3138 ret = ctdb_ctrl_modflags(ctdb, TIMELIMIT(), options.pnn,
3139 NODE_FLAGS_PERMANENTLY_DISABLED, 0);
3140 return ret == 0;
3143 static int control_disable(struct ctdb_context *ctdb, int argc, const char **argv)
3145 return update_flags_and_ipreallocate(ctdb, NULL,
3146 update_flags_disabled,
3147 NODE_FLAGS_PERMANENTLY_DISABLED,
3148 "disabled",
3149 true /* set_flag*/);
3152 /* Administratively re-enable a node */
3153 static bool update_flags_not_disabled(struct ctdb_context *ctdb, void *data)
3155 int ret;
3157 ret = ctdb_ctrl_modflags(ctdb, TIMELIMIT(), options.pnn,
3158 0, NODE_FLAGS_PERMANENTLY_DISABLED);
3159 return ret == 0;
3162 static int control_enable(struct ctdb_context *ctdb, int argc, const char **argv)
3164 return update_flags_and_ipreallocate(ctdb, NULL,
3165 update_flags_not_disabled,
3166 NODE_FLAGS_PERMANENTLY_DISABLED,
3167 "disabled",
3168 false /* set_flag*/);
3171 /* Stop a node */
3172 static bool update_flags_stopped(struct ctdb_context *ctdb, void *data)
3174 int ret;
3176 ret = ctdb_ctrl_stop_node(ctdb, TIMELIMIT(), options.pnn);
3178 return ret == 0;
3181 static int control_stop(struct ctdb_context *ctdb, int argc, const char **argv)
3183 return update_flags_and_ipreallocate(ctdb, NULL,
3184 update_flags_stopped,
3185 NODE_FLAGS_STOPPED,
3186 "stopped",
3187 true /* set_flag*/);
3190 /* Continue a stopped node */
3191 static bool update_flags_not_stopped(struct ctdb_context *ctdb, void *data)
3193 int ret;
3195 ret = ctdb_ctrl_continue_node(ctdb, TIMELIMIT(), options.pnn);
3197 return ret == 0;
3200 static int control_continue(struct ctdb_context *ctdb, int argc, const char **argv)
3202 return update_flags_and_ipreallocate(ctdb, NULL,
3203 update_flags_not_stopped,
3204 NODE_FLAGS_STOPPED,
3205 "stopped",
3206 false /* set_flag */);
3209 static uint32_t get_generation(struct ctdb_context *ctdb)
3211 TALLOC_CTX *tmp_ctx = talloc_new(ctdb);
3212 struct ctdb_vnn_map *vnnmap=NULL;
3213 int ret;
3214 uint32_t generation;
3216 /* wait until the recmaster is not in recovery mode */
3217 while (1) {
3218 uint32_t recmode, recmaster;
3220 if (vnnmap != NULL) {
3221 talloc_free(vnnmap);
3222 vnnmap = NULL;
3225 /* get the recmaster */
3226 ret = ctdb_ctrl_getrecmaster(ctdb, tmp_ctx, TIMELIMIT(), CTDB_CURRENT_NODE, &recmaster);
3227 if (ret != 0) {
3228 DEBUG(DEBUG_ERR, ("Unable to get recmaster from node %u\n", options.pnn));
3229 talloc_free(tmp_ctx);
3230 exit(10);
3233 /* get recovery mode */
3234 ret = ctdb_ctrl_getrecmode(ctdb, tmp_ctx, TIMELIMIT(), recmaster, &recmode);
3235 if (ret != 0) {
3236 DEBUG(DEBUG_ERR, ("Unable to get recmode from node %u\n", options.pnn));
3237 talloc_free(tmp_ctx);
3238 exit(10);
3241 /* get the current generation number */
3242 ret = ctdb_ctrl_getvnnmap(ctdb, TIMELIMIT(), recmaster, tmp_ctx, &vnnmap);
3243 if (ret != 0) {
3244 DEBUG(DEBUG_ERR, ("Unable to get vnnmap from recmaster (%u)\n", recmaster));
3245 talloc_free(tmp_ctx);
3246 exit(10);
3249 if ((recmode == CTDB_RECOVERY_NORMAL) && (vnnmap->generation != 1)) {
3250 generation = vnnmap->generation;
3251 talloc_free(tmp_ctx);
3252 return generation;
3254 sleep(1);
3258 /* Ban a node */
3259 static bool update_state_banned(struct ctdb_context *ctdb, void *data)
3261 struct ctdb_ban_time *bantime = (struct ctdb_ban_time *)data;
3262 int ret;
3264 ret = ctdb_ctrl_set_ban(ctdb, TIMELIMIT(), options.pnn, bantime);
3266 return ret == 0;
3269 static int control_ban(struct ctdb_context *ctdb, int argc, const char **argv)
3271 struct ctdb_ban_time bantime;
3273 if (argc < 1) {
3274 usage();
3277 bantime.pnn = options.pnn;
3278 bantime.time = strtoul(argv[0], NULL, 0);
3280 if (bantime.time == 0) {
3281 DEBUG(DEBUG_ERR, ("Invalid ban time specified - must be >0\n"));
3282 return -1;
3285 return update_flags_and_ipreallocate(ctdb, &bantime,
3286 update_state_banned,
3287 NODE_FLAGS_BANNED,
3288 "banned",
3289 true /* set_flag*/);
3293 /* Unban a node */
3294 static int control_unban(struct ctdb_context *ctdb, int argc, const char **argv)
3296 struct ctdb_ban_time bantime;
3298 bantime.pnn = options.pnn;
3299 bantime.time = 0;
3301 return update_flags_and_ipreallocate(ctdb, &bantime,
3302 update_state_banned,
3303 NODE_FLAGS_BANNED,
3304 "banned",
3305 false /* set_flag*/);
3309 show ban information for a node
3311 static int control_showban(struct ctdb_context *ctdb, int argc, const char **argv)
3313 int ret;
3314 struct ctdb_node_map *nodemap=NULL;
3315 struct ctdb_ban_time *bantime;
3317 /* verify the node exists */
3318 ret = ctdb_ctrl_getnodemap(ctdb, TIMELIMIT(), CTDB_CURRENT_NODE, ctdb, &nodemap);
3319 if (ret != 0) {
3320 DEBUG(DEBUG_ERR, ("Unable to get nodemap from local node\n"));
3321 return ret;
3324 ret = ctdb_ctrl_get_ban(ctdb, TIMELIMIT(), options.pnn, ctdb, &bantime);
3325 if (ret != 0) {
3326 DEBUG(DEBUG_ERR,("Showing ban info for node %d failed.\n", options.pnn));
3327 return -1;
3330 if (bantime->time == 0) {
3331 printf("Node %u is not banned\n", bantime->pnn);
3332 } else {
3333 printf("Node %u is banned, %d seconds remaining\n",
3334 bantime->pnn, bantime->time);
3337 return 0;
3341 shutdown a daemon
3343 static int control_shutdown(struct ctdb_context *ctdb, int argc, const char **argv)
3345 int ret;
3347 ret = ctdb_ctrl_shutdown(ctdb, TIMELIMIT(), options.pnn);
3348 if (ret != 0) {
3349 DEBUG(DEBUG_ERR, ("Unable to shutdown node %u\n", options.pnn));
3350 return ret;
3353 return 0;
3357 trigger a recovery
3359 static int control_recover(struct ctdb_context *ctdb, int argc, const char **argv)
3361 int ret;
3362 uint32_t generation, next_generation;
3363 bool force;
3365 /* "force" option ignores freeze failure and forces recovery */
3366 force = (argc == 1) && (strcasecmp(argv[0], "force") == 0);
3368 /* record the current generation number */
3369 generation = get_generation(ctdb);
3371 ret = ctdb_ctrl_freeze_priority(ctdb, TIMELIMIT(), options.pnn, 1);
3372 if (ret != 0) {
3373 if (!force) {
3374 DEBUG(DEBUG_ERR, ("Unable to freeze node\n"));
3375 return ret;
3377 DEBUG(DEBUG_WARNING, ("Unable to freeze node but proceeding because \"force\" option given\n"));
3380 ret = ctdb_ctrl_setrecmode(ctdb, TIMELIMIT(), options.pnn, CTDB_RECOVERY_ACTIVE);
3381 if (ret != 0) {
3382 DEBUG(DEBUG_ERR, ("Unable to set recovery mode\n"));
3383 return ret;
3386 /* wait until we are in a new generation */
3387 while (1) {
3388 next_generation = get_generation(ctdb);
3389 if (next_generation != generation) {
3390 return 0;
3392 sleep(1);
3395 return 0;
3400 display monitoring mode of a remote node
3402 static int control_getmonmode(struct ctdb_context *ctdb, int argc, const char **argv)
3404 uint32_t monmode;
3405 int ret;
3407 ret = ctdb_ctrl_getmonmode(ctdb, TIMELIMIT(), options.pnn, &monmode);
3408 if (ret != 0) {
3409 DEBUG(DEBUG_ERR, ("Unable to get monmode from node %u\n", options.pnn));
3410 return ret;
3412 if (!options.machinereadable){
3413 printf("Monitoring mode:%s (%d)\n",monmode==CTDB_MONITORING_ACTIVE?"ACTIVE":"DISABLED",monmode);
3414 } else {
3415 printf(":mode:\n");
3416 printf(":%d:\n",monmode);
3418 return 0;
3423 display capabilities of a remote node
3425 static int control_getcapabilities(struct ctdb_context *ctdb, int argc, const char **argv)
3427 uint32_t capabilities;
3428 int ret;
3430 ret = ctdb_ctrl_getcapabilities(ctdb, TIMELIMIT(), options.pnn, &capabilities);
3431 if (ret != 0) {
3432 DEBUG(DEBUG_ERR, ("Unable to get capabilities from node %u\n", options.pnn));
3433 return -1;
3436 if (!options.machinereadable){
3437 printf("RECMASTER: %s\n", (capabilities&CTDB_CAP_RECMASTER)?"YES":"NO");
3438 printf("LMASTER: %s\n", (capabilities&CTDB_CAP_LMASTER)?"YES":"NO");
3439 printf("LVS: %s\n", (capabilities&CTDB_CAP_LVS)?"YES":"NO");
3440 printf("NATGW: %s\n", (capabilities&CTDB_CAP_NATGW)?"YES":"NO");
3441 } else {
3442 printf(":RECMASTER:LMASTER:LVS:NATGW:\n");
3443 printf(":%d:%d:%d:%d:\n",
3444 !!(capabilities&CTDB_CAP_RECMASTER),
3445 !!(capabilities&CTDB_CAP_LMASTER),
3446 !!(capabilities&CTDB_CAP_LVS),
3447 !!(capabilities&CTDB_CAP_NATGW));
3449 return 0;
3453 display lvs configuration
3455 static int control_lvs(struct ctdb_context *ctdb, int argc, const char **argv)
3457 TALLOC_CTX *tmp_ctx = talloc_new(ctdb);
3458 uint32_t *capabilities;
3459 struct ctdb_node_map *nodemap=NULL;
3460 int i, ret;
3461 int healthy_count = 0;
3463 ret = ctdb_ctrl_getnodemap(ctdb, TIMELIMIT(), options.pnn, tmp_ctx, &nodemap);
3464 if (ret != 0) {
3465 DEBUG(DEBUG_ERR, ("Unable to get nodemap from node %u\n", options.pnn));
3466 talloc_free(tmp_ctx);
3467 return -1;
3470 capabilities = talloc_array(ctdb, uint32_t, nodemap->num);
3471 CTDB_NO_MEMORY(ctdb, capabilities);
3473 ret = 0;
3475 /* collect capabilities for all connected nodes */
3476 for (i=0; i<nodemap->num; i++) {
3477 if (nodemap->nodes[i].flags & NODE_FLAGS_INACTIVE) {
3478 continue;
3480 if (nodemap->nodes[i].flags & NODE_FLAGS_PERMANENTLY_DISABLED) {
3481 continue;
3484 ret = ctdb_ctrl_getcapabilities(ctdb, TIMELIMIT(), i, &capabilities[i]);
3485 if (ret != 0) {
3486 DEBUG(DEBUG_ERR, ("Unable to get capabilities from node %u\n", i));
3487 ret = -1;
3488 goto done;
3491 if (!(capabilities[i] & CTDB_CAP_LVS)) {
3492 continue;
3495 if (!(nodemap->nodes[i].flags & NODE_FLAGS_UNHEALTHY)) {
3496 healthy_count++;
3500 /* Print all LVS nodes */
3501 for (i=0; i<nodemap->num; i++) {
3502 if (nodemap->nodes[i].flags & NODE_FLAGS_INACTIVE) {
3503 continue;
3505 if (nodemap->nodes[i].flags & NODE_FLAGS_PERMANENTLY_DISABLED) {
3506 continue;
3508 if (!(capabilities[i] & CTDB_CAP_LVS)) {
3509 continue;
3512 if (healthy_count != 0) {
3513 if (nodemap->nodes[i].flags & NODE_FLAGS_UNHEALTHY) {
3514 continue;
3518 printf("%d:%s\n", i,
3519 ctdb_addr_to_str(&nodemap->nodes[i].addr));
3522 done:
3523 talloc_free(tmp_ctx);
3524 return ret;
3528 display who is the lvs master
3530 static int control_lvsmaster(struct ctdb_context *ctdb, int argc, const char **argv)
3532 TALLOC_CTX *tmp_ctx = talloc_new(ctdb);
3533 uint32_t *capabilities;
3534 struct ctdb_node_map *nodemap=NULL;
3535 int i, ret;
3536 int healthy_count = 0;
3538 ret = ctdb_ctrl_getnodemap(ctdb, TIMELIMIT(), options.pnn, tmp_ctx, &nodemap);
3539 if (ret != 0) {
3540 DEBUG(DEBUG_ERR, ("Unable to get nodemap from node %u\n", options.pnn));
3541 talloc_free(tmp_ctx);
3542 return -1;
3545 capabilities = talloc_array(tmp_ctx, uint32_t, nodemap->num);
3546 if (capabilities == NULL) {
3547 talloc_free(tmp_ctx);
3548 CTDB_NO_MEMORY(ctdb, capabilities);
3551 /* collect capabilities for all connected nodes */
3552 for (i=0; i<nodemap->num; i++) {
3553 if (nodemap->nodes[i].flags & NODE_FLAGS_INACTIVE) {
3554 continue;
3556 if (nodemap->nodes[i].flags & NODE_FLAGS_PERMANENTLY_DISABLED) {
3557 continue;
3560 ret = ctdb_ctrl_getcapabilities(ctdb, TIMELIMIT(), i, &capabilities[i]);
3561 if (ret != 0) {
3562 DEBUG(DEBUG_ERR, ("Unable to get capabilities from node %u\n", i));
3563 ret = -1;
3564 goto done;
3567 if (!(capabilities[i] & CTDB_CAP_LVS)) {
3568 continue;
3571 if (!(nodemap->nodes[i].flags & NODE_FLAGS_UNHEALTHY)) {
3572 healthy_count++;
3576 ret = -1;
3578 /* find and show the lvsmaster */
3579 for (i=0; i<nodemap->num; i++) {
3580 if (nodemap->nodes[i].flags & NODE_FLAGS_INACTIVE) {
3581 continue;
3583 if (nodemap->nodes[i].flags & NODE_FLAGS_PERMANENTLY_DISABLED) {
3584 continue;
3586 if (!(capabilities[i] & CTDB_CAP_LVS)) {
3587 continue;
3590 if (healthy_count != 0) {
3591 if (nodemap->nodes[i].flags & NODE_FLAGS_UNHEALTHY) {
3592 continue;
3596 if (options.machinereadable){
3597 printf("%d\n", i);
3598 } else {
3599 printf("Node %d is LVS master\n", i);
3601 ret = 0;
3602 goto done;
3605 printf("There is no LVS master\n");
3606 done:
3607 talloc_free(tmp_ctx);
3608 return ret;
3612 disable monitoring on a node
3614 static int control_disable_monmode(struct ctdb_context *ctdb, int argc, const char **argv)
3617 int ret;
3619 ret = ctdb_ctrl_disable_monmode(ctdb, TIMELIMIT(), options.pnn);
3620 if (ret != 0) {
3621 DEBUG(DEBUG_ERR, ("Unable to disable monmode on node %u\n", options.pnn));
3622 return ret;
3624 printf("Monitoring mode:%s\n","DISABLED");
3626 return 0;
3630 enable monitoring on a node
3632 static int control_enable_monmode(struct ctdb_context *ctdb, int argc, const char **argv)
3635 int ret;
3637 ret = ctdb_ctrl_enable_monmode(ctdb, TIMELIMIT(), options.pnn);
3638 if (ret != 0) {
3639 DEBUG(DEBUG_ERR, ("Unable to enable monmode on node %u\n", options.pnn));
3640 return ret;
3642 printf("Monitoring mode:%s\n","ACTIVE");
3644 return 0;
3648 display remote list of keys/data for a db
3650 static int control_catdb(struct ctdb_context *ctdb, int argc, const char **argv)
3652 const char *db_name;
3653 struct ctdb_db_context *ctdb_db;
3654 int ret;
3655 struct ctdb_dump_db_context c;
3656 uint8_t flags;
3658 if (argc < 1) {
3659 usage();
3662 db_name = argv[0];
3664 if (!db_exists(ctdb, db_name, NULL, &flags)) {
3665 return -1;
3668 ctdb_db = ctdb_attach(ctdb, TIMELIMIT(), db_name, flags & CTDB_DB_FLAGS_PERSISTENT, 0);
3669 if (ctdb_db == NULL) {
3670 DEBUG(DEBUG_ERR,("Unable to attach to database '%s'\n", db_name));
3671 return -1;
3674 if (options.printlmaster) {
3675 ret = ctdb_ctrl_getvnnmap(ctdb, TIMELIMIT(), options.pnn,
3676 ctdb, &ctdb->vnn_map);
3677 if (ret != 0) {
3678 DEBUG(DEBUG_ERR, ("Unable to get vnnmap from node %u\n",
3679 options.pnn));
3680 return ret;
3684 ZERO_STRUCT(c);
3685 c.f = stdout;
3686 c.printemptyrecords = (bool)options.printemptyrecords;
3687 c.printdatasize = (bool)options.printdatasize;
3688 c.printlmaster = (bool)options.printlmaster;
3689 c.printhash = (bool)options.printhash;
3690 c.printrecordflags = (bool)options.printrecordflags;
3692 /* traverse and dump the cluster tdb */
3693 ret = ctdb_dump_db(ctdb_db, &c);
3694 if (ret == -1) {
3695 DEBUG(DEBUG_ERR, ("Unable to dump database\n"));
3696 DEBUG(DEBUG_ERR, ("Maybe try 'ctdb getdbstatus %s'"
3697 " and 'ctdb getvar AllowUnhealthyDBRead'\n",
3698 db_name));
3699 return -1;
3701 talloc_free(ctdb_db);
3703 printf("Dumped %d records\n", ret);
3704 return 0;
3707 struct cattdb_data {
3708 struct ctdb_context *ctdb;
3709 uint32_t count;
3712 static int cattdb_traverse(struct tdb_context *tdb, TDB_DATA key, TDB_DATA data, void *private_data)
3714 struct cattdb_data *d = private_data;
3715 struct ctdb_dump_db_context c;
3717 d->count++;
3719 ZERO_STRUCT(c);
3720 c.f = stdout;
3721 c.printemptyrecords = (bool)options.printemptyrecords;
3722 c.printdatasize = (bool)options.printdatasize;
3723 c.printlmaster = false;
3724 c.printhash = (bool)options.printhash;
3725 c.printrecordflags = true;
3727 return ctdb_dumpdb_record(d->ctdb, key, data, &c);
3731 cat the local tdb database using same format as catdb
3733 static int control_cattdb(struct ctdb_context *ctdb, int argc, const char **argv)
3735 const char *db_name;
3736 struct ctdb_db_context *ctdb_db;
3737 struct cattdb_data d;
3738 uint8_t flags;
3740 if (argc < 1) {
3741 usage();
3744 db_name = argv[0];
3746 if (!db_exists(ctdb, db_name, NULL, &flags)) {
3747 return -1;
3750 ctdb_db = ctdb_attach(ctdb, TIMELIMIT(), db_name, flags & CTDB_DB_FLAGS_PERSISTENT, 0);
3751 if (ctdb_db == NULL) {
3752 DEBUG(DEBUG_ERR,("Unable to attach to database '%s'\n", db_name));
3753 return -1;
3756 /* traverse the local tdb */
3757 d.count = 0;
3758 d.ctdb = ctdb;
3759 if (tdb_traverse_read(ctdb_db->ltdb->tdb, cattdb_traverse, &d) == -1) {
3760 printf("Failed to cattdb data\n");
3761 exit(10);
3763 talloc_free(ctdb_db);
3765 printf("Dumped %d records\n", d.count);
3766 return 0;
3770 display the content of a database key
3772 static int control_readkey(struct ctdb_context *ctdb, int argc, const char **argv)
3774 const char *db_name;
3775 struct ctdb_db_context *ctdb_db;
3776 struct ctdb_record_handle *h;
3777 TALLOC_CTX *tmp_ctx = talloc_new(ctdb);
3778 TDB_DATA key, data;
3779 uint8_t flags;
3781 if (argc < 2) {
3782 usage();
3785 db_name = argv[0];
3787 if (!db_exists(ctdb, db_name, NULL, &flags)) {
3788 return -1;
3791 ctdb_db = ctdb_attach(ctdb, TIMELIMIT(), db_name, flags & CTDB_DB_FLAGS_PERSISTENT, 0);
3792 if (ctdb_db == NULL) {
3793 DEBUG(DEBUG_ERR,("Unable to attach to database '%s'\n", db_name));
3794 return -1;
3797 key.dptr = discard_const(argv[1]);
3798 key.dsize = strlen((char *)key.dptr);
3800 h = ctdb_fetch_lock(ctdb_db, tmp_ctx, key, &data);
3801 if (h == NULL) {
3802 printf("Failed to fetch record '%s' on node %d\n",
3803 (const char *)key.dptr, ctdb_get_pnn(ctdb));
3804 talloc_free(tmp_ctx);
3805 exit(10);
3808 printf("Data: size:%d ptr:[%.*s]\n", (int)data.dsize, (int)data.dsize, data.dptr);
3810 talloc_free(ctdb_db);
3811 talloc_free(tmp_ctx);
3812 return 0;
3816 display the content of a database key
3818 static int control_writekey(struct ctdb_context *ctdb, int argc, const char **argv)
3820 const char *db_name;
3821 struct ctdb_db_context *ctdb_db;
3822 struct ctdb_record_handle *h;
3823 TALLOC_CTX *tmp_ctx = talloc_new(ctdb);
3824 TDB_DATA key, data;
3825 uint8_t flags;
3827 if (argc < 3) {
3828 usage();
3831 db_name = argv[0];
3833 if (!db_exists(ctdb, db_name, NULL, &flags)) {
3834 return -1;
3837 ctdb_db = ctdb_attach(ctdb, TIMELIMIT(), db_name, flags & CTDB_DB_FLAGS_PERSISTENT, 0);
3838 if (ctdb_db == NULL) {
3839 DEBUG(DEBUG_ERR,("Unable to attach to database '%s'\n", db_name));
3840 return -1;
3843 key.dptr = discard_const(argv[1]);
3844 key.dsize = strlen((char *)key.dptr);
3846 h = ctdb_fetch_lock(ctdb_db, tmp_ctx, key, &data);
3847 if (h == NULL) {
3848 printf("Failed to fetch record '%s' on node %d\n",
3849 (const char *)key.dptr, ctdb_get_pnn(ctdb));
3850 talloc_free(tmp_ctx);
3851 exit(10);
3854 data.dptr = discard_const(argv[2]);
3855 data.dsize = strlen((char *)data.dptr);
3857 if (ctdb_record_store(h, data) != 0) {
3858 printf("Failed to store record\n");
3861 talloc_free(h);
3862 talloc_free(ctdb_db);
3863 talloc_free(tmp_ctx);
3864 return 0;
3868 fetch a record from a persistent database
3870 static int control_pfetch(struct ctdb_context *ctdb, int argc, const char **argv)
3872 const char *db_name;
3873 struct ctdb_db_context *ctdb_db;
3874 TALLOC_CTX *tmp_ctx = talloc_new(ctdb);
3875 struct ctdb_transaction_handle *h;
3876 TDB_DATA key, data;
3877 int fd, ret;
3878 bool persistent;
3879 uint8_t flags;
3881 if (argc < 2) {
3882 talloc_free(tmp_ctx);
3883 usage();
3886 db_name = argv[0];
3888 if (!db_exists(ctdb, db_name, NULL, &flags)) {
3889 talloc_free(tmp_ctx);
3890 return -1;
3893 persistent = flags & CTDB_DB_FLAGS_PERSISTENT;
3894 if (!persistent) {
3895 DEBUG(DEBUG_ERR,("Database '%s' is not persistent\n", db_name));
3896 talloc_free(tmp_ctx);
3897 return -1;
3900 ctdb_db = ctdb_attach(ctdb, TIMELIMIT(), db_name, persistent, 0);
3901 if (ctdb_db == NULL) {
3902 DEBUG(DEBUG_ERR,("Unable to attach to database '%s'\n", db_name));
3903 talloc_free(tmp_ctx);
3904 return -1;
3907 h = ctdb_transaction_start(ctdb_db, tmp_ctx);
3908 if (h == NULL) {
3909 DEBUG(DEBUG_ERR,("Failed to start transaction on database %s\n", db_name));
3910 talloc_free(tmp_ctx);
3911 return -1;
3914 key.dptr = discard_const(argv[1]);
3915 key.dsize = strlen(argv[1]);
3916 ret = ctdb_transaction_fetch(h, tmp_ctx, key, &data);
3917 if (ret != 0) {
3918 DEBUG(DEBUG_ERR,("Failed to fetch record\n"));
3919 talloc_free(tmp_ctx);
3920 return -1;
3923 if (data.dsize == 0 || data.dptr == NULL) {
3924 DEBUG(DEBUG_ERR,("Record is empty\n"));
3925 talloc_free(tmp_ctx);
3926 return -1;
3929 if (argc == 3) {
3930 fd = open(argv[2], O_WRONLY|O_CREAT|O_TRUNC, 0600);
3931 if (fd == -1) {
3932 DEBUG(DEBUG_ERR,("Failed to open output file %s\n", argv[2]));
3933 talloc_free(tmp_ctx);
3934 return -1;
3936 write(fd, data.dptr, data.dsize);
3937 close(fd);
3938 } else {
3939 write(1, data.dptr, data.dsize);
3942 /* abort the transaction */
3943 talloc_free(h);
3946 talloc_free(tmp_ctx);
3947 return 0;
3951 fetch a record from a tdb-file
3953 static int control_tfetch(struct ctdb_context *ctdb, int argc, const char **argv)
3955 const char *tdb_file;
3956 TDB_CONTEXT *tdb;
3957 TDB_DATA key, data;
3958 TALLOC_CTX *tmp_ctx = talloc_new(NULL);
3959 int fd;
3961 if (argc < 2) {
3962 usage();
3965 tdb_file = argv[0];
3967 tdb = tdb_open(tdb_file, 0, 0, O_RDONLY, 0);
3968 if (tdb == NULL) {
3969 printf("Failed to open TDB file %s\n", tdb_file);
3970 return -1;
3973 if (!strncmp(argv[1], "0x", 2)) {
3974 key = hextodata(tmp_ctx, argv[1] + 2);
3975 if (key.dsize == 0) {
3976 printf("Failed to convert \"%s\" into a TDB_DATA\n", argv[1]);
3977 return -1;
3979 } else {
3980 key.dptr = discard_const(argv[1]);
3981 key.dsize = strlen(argv[1]);
3984 data = tdb_fetch(tdb, key);
3985 if (data.dptr == NULL || data.dsize < sizeof(struct ctdb_ltdb_header)) {
3986 printf("Failed to read record %s from tdb %s\n", argv[1], tdb_file);
3987 tdb_close(tdb);
3988 return -1;
3991 tdb_close(tdb);
3993 if (argc == 3) {
3994 fd = open(argv[2], O_WRONLY|O_CREAT|O_TRUNC, 0600);
3995 if (fd == -1) {
3996 printf("Failed to open output file %s\n", argv[2]);
3997 return -1;
3999 if (options.verbose){
4000 write(fd, data.dptr, data.dsize);
4001 } else {
4002 write(fd, data.dptr+sizeof(struct ctdb_ltdb_header), data.dsize-sizeof(struct ctdb_ltdb_header));
4004 close(fd);
4005 } else {
4006 if (options.verbose){
4007 write(1, data.dptr, data.dsize);
4008 } else {
4009 write(1, data.dptr+sizeof(struct ctdb_ltdb_header), data.dsize-sizeof(struct ctdb_ltdb_header));
4013 talloc_free(tmp_ctx);
4014 return 0;
4018 store a record and header to a tdb-file
4020 static int control_tstore(struct ctdb_context *ctdb, int argc, const char **argv)
4022 const char *tdb_file;
4023 TDB_CONTEXT *tdb;
4024 TDB_DATA key, data;
4025 TALLOC_CTX *tmp_ctx = talloc_new(NULL);
4027 if (argc < 3) {
4028 usage();
4031 tdb_file = argv[0];
4033 tdb = tdb_open(tdb_file, 0, 0, O_RDWR, 0);
4034 if (tdb == NULL) {
4035 printf("Failed to open TDB file %s\n", tdb_file);
4036 return -1;
4039 if (!strncmp(argv[1], "0x", 2)) {
4040 key = hextodata(tmp_ctx, argv[1] + 2);
4041 if (key.dsize == 0) {
4042 printf("Failed to convert \"%s\" into a TDB_DATA\n", argv[1]);
4043 return -1;
4045 } else {
4046 key.dptr = discard_const(argv[1]);
4047 key.dsize = strlen(argv[1]);
4050 if (!strncmp(argv[2], "0x", 2)) {
4051 data = hextodata(tmp_ctx, argv[2] + 2);
4052 if (data.dsize == 0) {
4053 printf("Failed to convert \"%s\" into a TDB_DATA\n", argv[2]);
4054 return -1;
4056 } else {
4057 data.dptr = discard_const(argv[2]);
4058 data.dsize = strlen(argv[2]);
4061 if (data.dsize < sizeof(struct ctdb_ltdb_header)) {
4062 printf("Not enough data. You must specify the full ctdb_ltdb_header too when storing\n");
4063 return -1;
4065 if (tdb_store(tdb, key, data, TDB_REPLACE) != 0) {
4066 printf("Failed to write record %s to tdb %s\n", argv[1], tdb_file);
4067 tdb_close(tdb);
4068 return -1;
4071 tdb_close(tdb);
4073 talloc_free(tmp_ctx);
4074 return 0;
4078 write a record to a persistent database
4080 static int control_pstore(struct ctdb_context *ctdb, int argc, const char **argv)
4082 const char *db_name;
4083 struct ctdb_db_context *ctdb_db;
4084 TALLOC_CTX *tmp_ctx = talloc_new(ctdb);
4085 struct ctdb_transaction_handle *h;
4086 struct stat st;
4087 TDB_DATA key, data;
4088 int fd, ret;
4090 if (argc < 3) {
4091 talloc_free(tmp_ctx);
4092 usage();
4095 fd = open(argv[2], O_RDONLY);
4096 if (fd == -1) {
4097 DEBUG(DEBUG_ERR,("Failed to open file containing record data : %s %s\n", argv[2], strerror(errno)));
4098 talloc_free(tmp_ctx);
4099 return -1;
4102 ret = fstat(fd, &st);
4103 if (ret == -1) {
4104 DEBUG(DEBUG_ERR,("fstat of file %s failed: %s\n", argv[2], strerror(errno)));
4105 close(fd);
4106 talloc_free(tmp_ctx);
4107 return -1;
4110 if (!S_ISREG(st.st_mode)) {
4111 DEBUG(DEBUG_ERR,("Not a regular file %s\n", argv[2]));
4112 close(fd);
4113 talloc_free(tmp_ctx);
4114 return -1;
4117 data.dsize = st.st_size;
4118 if (data.dsize == 0) {
4119 data.dptr = NULL;
4120 } else {
4121 data.dptr = talloc_size(tmp_ctx, data.dsize);
4122 if (data.dptr == NULL) {
4123 DEBUG(DEBUG_ERR,("Failed to talloc %d of memory to store record data\n", (int)data.dsize));
4124 close(fd);
4125 talloc_free(tmp_ctx);
4126 return -1;
4128 ret = read(fd, data.dptr, data.dsize);
4129 if (ret != data.dsize) {
4130 DEBUG(DEBUG_ERR,("Failed to read %d bytes of record data\n", (int)data.dsize));
4131 close(fd);
4132 talloc_free(tmp_ctx);
4133 return -1;
4136 close(fd);
4139 db_name = argv[0];
4141 ctdb_db = ctdb_attach(ctdb, TIMELIMIT(), db_name, true, 0);
4142 if (ctdb_db == NULL) {
4143 DEBUG(DEBUG_ERR,("Unable to attach to database '%s'\n", db_name));
4144 talloc_free(tmp_ctx);
4145 return -1;
4148 h = ctdb_transaction_start(ctdb_db, tmp_ctx);
4149 if (h == NULL) {
4150 DEBUG(DEBUG_ERR,("Failed to start transaction on database %s\n", db_name));
4151 talloc_free(tmp_ctx);
4152 return -1;
4155 key.dptr = discard_const(argv[1]);
4156 key.dsize = strlen(argv[1]);
4157 ret = ctdb_transaction_store(h, key, data);
4158 if (ret != 0) {
4159 DEBUG(DEBUG_ERR,("Failed to store record\n"));
4160 talloc_free(tmp_ctx);
4161 return -1;
4164 ret = ctdb_transaction_commit(h);
4165 if (ret != 0) {
4166 DEBUG(DEBUG_ERR,("Failed to commit transaction\n"));
4167 talloc_free(tmp_ctx);
4168 return -1;
4172 talloc_free(tmp_ctx);
4173 return 0;
4177 * delete a record from a persistent database
4179 static int control_pdelete(struct ctdb_context *ctdb, int argc, const char **argv)
4181 const char *db_name;
4182 struct ctdb_db_context *ctdb_db;
4183 TALLOC_CTX *tmp_ctx = talloc_new(ctdb);
4184 struct ctdb_transaction_handle *h;
4185 TDB_DATA key;
4186 int ret;
4187 bool persistent;
4188 uint8_t flags;
4190 if (argc < 2) {
4191 talloc_free(tmp_ctx);
4192 usage();
4195 db_name = argv[0];
4197 if (!db_exists(ctdb, db_name, NULL, &flags)) {
4198 talloc_free(tmp_ctx);
4199 return -1;
4202 persistent = flags & CTDB_DB_FLAGS_PERSISTENT;
4203 if (!persistent) {
4204 DEBUG(DEBUG_ERR, ("Database '%s' is not persistent\n", db_name));
4205 talloc_free(tmp_ctx);
4206 return -1;
4209 ctdb_db = ctdb_attach(ctdb, TIMELIMIT(), db_name, persistent, 0);
4210 if (ctdb_db == NULL) {
4211 DEBUG(DEBUG_ERR, ("Unable to attach to database '%s'\n", db_name));
4212 talloc_free(tmp_ctx);
4213 return -1;
4216 h = ctdb_transaction_start(ctdb_db, tmp_ctx);
4217 if (h == NULL) {
4218 DEBUG(DEBUG_ERR, ("Failed to start transaction on database %s\n", db_name));
4219 talloc_free(tmp_ctx);
4220 return -1;
4223 key.dptr = discard_const(argv[1]);
4224 key.dsize = strlen(argv[1]);
4225 ret = ctdb_transaction_store(h, key, tdb_null);
4226 if (ret != 0) {
4227 DEBUG(DEBUG_ERR, ("Failed to delete record\n"));
4228 talloc_free(tmp_ctx);
4229 return -1;
4232 ret = ctdb_transaction_commit(h);
4233 if (ret != 0) {
4234 DEBUG(DEBUG_ERR, ("Failed to commit transaction\n"));
4235 talloc_free(tmp_ctx);
4236 return -1;
4239 talloc_free(tmp_ctx);
4240 return 0;
4244 check if a service is bound to a port or not
4246 static int control_chktcpport(struct ctdb_context *ctdb, int argc, const char **argv)
4248 int s, ret;
4249 unsigned v;
4250 int port;
4251 struct sockaddr_in sin;
4253 if (argc != 1) {
4254 printf("Use: ctdb chktcport <port>\n");
4255 return EINVAL;
4258 port = atoi(argv[0]);
4260 s = socket(PF_INET, SOCK_STREAM, IPPROTO_TCP);
4261 if (s == -1) {
4262 printf("Failed to open local socket\n");
4263 return errno;
4266 v = fcntl(s, F_GETFL, 0);
4267 fcntl(s, F_SETFL, v | O_NONBLOCK);
4269 bzero(&sin, sizeof(sin));
4270 sin.sin_family = PF_INET;
4271 sin.sin_port = htons(port);
4272 ret = bind(s, (struct sockaddr *)&sin, sizeof(sin));
4273 close(s);
4274 if (ret == -1) {
4275 printf("Failed to bind to local socket: %d %s\n", errno, strerror(errno));
4276 return errno;
4279 return 0;
4284 static void log_handler(struct ctdb_context *ctdb, uint64_t srvid,
4285 TDB_DATA data, void *private_data)
4287 DEBUG(DEBUG_ERR,("Log data received\n"));
4288 if (data.dsize > 0) {
4289 printf("%s", data.dptr);
4292 exit(0);
4296 display a list of log messages from the in memory ringbuffer
4298 static int control_getlog(struct ctdb_context *ctdb, int argc, const char **argv)
4300 int ret, i;
4301 bool main_daemon;
4302 struct ctdb_get_log_addr log_addr;
4303 TDB_DATA data;
4304 struct timeval tv;
4306 /* Process options */
4307 main_daemon = true;
4308 log_addr.pnn = ctdb_get_pnn(ctdb);
4309 log_addr.level = DEBUG_NOTICE;
4310 for (i = 0; i < argc; i++) {
4311 if (strcmp(argv[i], "recoverd") == 0) {
4312 main_daemon = false;
4313 } else {
4314 if (isalpha(argv[i][0]) || argv[i][0] == '-') {
4315 log_addr.level = get_debug_by_desc(argv[i]);
4316 } else {
4317 log_addr.level = strtol(argv[i], NULL, 0);
4322 /* Our message port is our PID */
4323 log_addr.srvid = getpid();
4325 data.dptr = (unsigned char *)&log_addr;
4326 data.dsize = sizeof(log_addr);
4328 DEBUG(DEBUG_ERR, ("Pulling logs from node %u\n", options.pnn));
4330 ctdb_client_set_message_handler(ctdb, log_addr.srvid, log_handler, NULL);
4331 sleep(1);
4333 DEBUG(DEBUG_ERR,("Listen for response on %d\n", (int)log_addr.srvid));
4335 if (main_daemon) {
4336 int32_t res;
4337 char *errmsg;
4338 TALLOC_CTX *tmp_ctx = talloc_new(ctdb);
4340 ret = ctdb_control(ctdb, options.pnn, 0, CTDB_CONTROL_GET_LOG,
4341 0, data, tmp_ctx, NULL, &res, NULL, &errmsg);
4342 if (ret != 0 || res != 0) {
4343 DEBUG(DEBUG_ERR,("Failed to get logs - %s\n", errmsg));
4344 talloc_free(tmp_ctx);
4345 return -1;
4347 talloc_free(tmp_ctx);
4348 } else {
4349 ret = ctdb_client_send_message(ctdb, options.pnn,
4350 CTDB_SRVID_GETLOG, data);
4351 if (ret != 0) {
4352 DEBUG(DEBUG_ERR,("Failed to send getlog request message to %u\n", options.pnn));
4353 return -1;
4357 tv = timeval_current();
4358 /* this loop will terminate when we have received the reply */
4359 while (timeval_elapsed(&tv) < (double)options.timelimit) {
4360 event_loop_once(ctdb->ev);
4363 DEBUG(DEBUG_INFO,("Timed out waiting for log data.\n"));
4365 return 0;
4369 clear the in memory log area
4371 static int control_clearlog(struct ctdb_context *ctdb, int argc, const char **argv)
4373 int ret;
4375 if (argc == 0 || (argc >= 1 && strcmp(argv[0], "recoverd") != 0)) {
4376 /* "recoverd" not given - get logs from main daemon */
4377 int32_t res;
4378 char *errmsg;
4379 TALLOC_CTX *tmp_ctx = talloc_new(ctdb);
4381 ret = ctdb_control(ctdb, options.pnn, 0, CTDB_CONTROL_CLEAR_LOG,
4382 0, tdb_null, tmp_ctx, NULL, &res, NULL, &errmsg);
4383 if (ret != 0 || res != 0) {
4384 DEBUG(DEBUG_ERR,("Failed to clear logs\n"));
4385 talloc_free(tmp_ctx);
4386 return -1;
4389 talloc_free(tmp_ctx);
4390 } else {
4391 TDB_DATA data; /* unused in recoverd... */
4392 data.dsize = 0;
4394 ret = ctdb_client_send_message(ctdb, options.pnn, CTDB_SRVID_CLEARLOG, data);
4395 if (ret != 0) {
4396 DEBUG(DEBUG_ERR,("Failed to send clearlog request message to %u\n", options.pnn));
4397 return -1;
4401 return 0;
4404 /* Reload public IPs on a specified nodes */
4405 static int control_reloadips(struct ctdb_context *ctdb, int argc, const char **argv)
4407 TALLOC_CTX *tmp_ctx = talloc_new(ctdb);
4408 uint32_t *nodes;
4409 uint32_t pnn_mode;
4410 int ret;
4412 assert_single_node_only();
4414 if (argc > 1) {
4415 usage();
4418 /* Determine the nodes where IPs need to be reloaded */
4419 if (!parse_nodestring(ctdb, tmp_ctx, argc == 1 ? argv[0] : NULL,
4420 options.pnn, true, &nodes, &pnn_mode)) {
4421 ret = -1;
4422 goto done;
4425 again:
4426 /* Disable takeover runs on all connected nodes. A reply
4427 * indicating success is needed from each node so all nodes
4428 * will need to be active. This will retry until maxruntime
4429 * is exceeded, hence no error handling.
4431 * A check could be added to not allow reloading of IPs when
4432 * there are disconnected nodes. However, this should
4433 * probably be left up to the administrator.
4435 srvid_broadcast(ctdb, CTDB_SRVID_DISABLE_TAKEOVER_RUNS, LONGTIMEOUT,
4436 "Disable takeover runs", true);
4438 /* Now tell all the desired nodes to reload their public IPs.
4439 * Keep trying this until it succeeds. This assumes all
4440 * failures are transient, which might not be true...
4442 if (ctdb_client_async_control(ctdb, CTDB_CONTROL_RELOAD_PUBLIC_IPS,
4443 nodes, 0, LONGTIMELIMIT(),
4444 false, tdb_null,
4445 NULL, NULL, NULL) != 0) {
4446 DEBUG(DEBUG_ERR,
4447 ("Unable to reload IPs on some nodes, try again.\n"));
4448 goto again;
4451 /* It isn't strictly necessary to wait until takeover runs are
4452 * re-enabled but doing so can't hurt.
4454 srvid_broadcast(ctdb, CTDB_SRVID_DISABLE_TAKEOVER_RUNS, 0,
4455 "Enable takeover runs", true);
4457 ipreallocate(ctdb);
4459 ret = 0;
4460 done:
4461 talloc_free(tmp_ctx);
4462 return ret;
4466 display a list of the databases on a remote ctdb
4468 static int control_getdbmap(struct ctdb_context *ctdb, int argc, const char **argv)
4470 int i, ret;
4471 struct ctdb_dbid_map *dbmap=NULL;
4473 ret = ctdb_ctrl_getdbmap(ctdb, TIMELIMIT(), options.pnn, ctdb, &dbmap);
4474 if (ret != 0) {
4475 DEBUG(DEBUG_ERR, ("Unable to get dbids from node %u\n", options.pnn));
4476 return ret;
4479 if(options.machinereadable){
4480 printf(":ID:Name:Path:Persistent:Sticky:Unhealthy:ReadOnly:\n");
4481 for(i=0;i<dbmap->num;i++){
4482 const char *path;
4483 const char *name;
4484 const char *health;
4485 bool persistent;
4486 bool readonly;
4487 bool sticky;
4489 ctdb_ctrl_getdbpath(ctdb, TIMELIMIT(), options.pnn,
4490 dbmap->dbs[i].dbid, ctdb, &path);
4491 ctdb_ctrl_getdbname(ctdb, TIMELIMIT(), options.pnn,
4492 dbmap->dbs[i].dbid, ctdb, &name);
4493 ctdb_ctrl_getdbhealth(ctdb, TIMELIMIT(), options.pnn,
4494 dbmap->dbs[i].dbid, ctdb, &health);
4495 persistent = dbmap->dbs[i].flags & CTDB_DB_FLAGS_PERSISTENT;
4496 readonly = dbmap->dbs[i].flags & CTDB_DB_FLAGS_READONLY;
4497 sticky = dbmap->dbs[i].flags & CTDB_DB_FLAGS_STICKY;
4498 printf(":0x%08X:%s:%s:%d:%d:%d:%d:\n",
4499 dbmap->dbs[i].dbid, name, path,
4500 !!(persistent), !!(sticky),
4501 !!(health), !!(readonly));
4503 return 0;
4506 printf("Number of databases:%d\n", dbmap->num);
4507 for(i=0;i<dbmap->num;i++){
4508 const char *path;
4509 const char *name;
4510 const char *health;
4511 bool persistent;
4512 bool readonly;
4513 bool sticky;
4515 ctdb_ctrl_getdbpath(ctdb, TIMELIMIT(), options.pnn, dbmap->dbs[i].dbid, ctdb, &path);
4516 ctdb_ctrl_getdbname(ctdb, TIMELIMIT(), options.pnn, dbmap->dbs[i].dbid, ctdb, &name);
4517 ctdb_ctrl_getdbhealth(ctdb, TIMELIMIT(), options.pnn, dbmap->dbs[i].dbid, ctdb, &health);
4518 persistent = dbmap->dbs[i].flags & CTDB_DB_FLAGS_PERSISTENT;
4519 readonly = dbmap->dbs[i].flags & CTDB_DB_FLAGS_READONLY;
4520 sticky = dbmap->dbs[i].flags & CTDB_DB_FLAGS_STICKY;
4521 printf("dbid:0x%08x name:%s path:%s%s%s%s%s\n",
4522 dbmap->dbs[i].dbid, name, path,
4523 persistent?" PERSISTENT":"",
4524 sticky?" STICKY":"",
4525 readonly?" READONLY":"",
4526 health?" UNHEALTHY":"");
4529 return 0;
4533 display the status of a database on a remote ctdb
4535 static int control_getdbstatus(struct ctdb_context *ctdb, int argc, const char **argv)
4537 const char *db_name;
4538 uint32_t db_id;
4539 uint8_t flags;
4540 const char *path;
4541 const char *health;
4543 if (argc < 1) {
4544 usage();
4547 db_name = argv[0];
4549 if (!db_exists(ctdb, db_name, &db_id, &flags)) {
4550 return -1;
4553 ctdb_ctrl_getdbpath(ctdb, TIMELIMIT(), options.pnn, db_id, ctdb, &path);
4554 ctdb_ctrl_getdbhealth(ctdb, TIMELIMIT(), options.pnn, db_id, ctdb, &health);
4555 printf("dbid: 0x%08x\nname: %s\npath: %s\nPERSISTENT: %s\nSTICKY: %s\nREADONLY: %s\nHEALTH: %s\n",
4556 db_id, db_name, path,
4557 (flags & CTDB_DB_FLAGS_PERSISTENT ? "yes" : "no"),
4558 (flags & CTDB_DB_FLAGS_STICKY ? "yes" : "no"),
4559 (flags & CTDB_DB_FLAGS_READONLY ? "yes" : "no"),
4560 (health ? health : "OK"));
4562 return 0;
4566 check if the local node is recmaster or not
4567 it will return 1 if this node is the recmaster and 0 if it is not
4568 or if the local ctdb daemon could not be contacted
4570 static int control_isnotrecmaster(struct ctdb_context *ctdb, int argc, const char **argv)
4572 uint32_t mypnn, recmaster;
4573 int ret;
4575 assert_single_node_only();
4577 mypnn = getpnn(ctdb);
4579 ret = ctdb_ctrl_getrecmaster(ctdb, ctdb, TIMELIMIT(), options.pnn, &recmaster);
4580 if (ret != 0) {
4581 printf("Failed to get the recmaster\n");
4582 return 1;
4585 if (recmaster != mypnn) {
4586 printf("this node is not the recmaster\n");
4587 return 1;
4590 printf("this node is the recmaster\n");
4591 return 0;
4595 ping a node
4597 static int control_ping(struct ctdb_context *ctdb, int argc, const char **argv)
4599 int ret;
4600 struct timeval tv = timeval_current();
4601 ret = ctdb_ctrl_ping(ctdb, options.pnn);
4602 if (ret == -1) {
4603 printf("Unable to get ping response from node %u\n", options.pnn);
4604 return -1;
4605 } else {
4606 printf("response from %u time=%.6f sec (%d clients)\n",
4607 options.pnn, timeval_elapsed(&tv), ret);
4609 return 0;
4614 get a node's runstate
4616 static int control_runstate(struct ctdb_context *ctdb, int argc, const char **argv)
4618 int ret;
4619 enum ctdb_runstate runstate;
4621 ret = ctdb_ctrl_get_runstate(ctdb, TIMELIMIT(), options.pnn, &runstate);
4622 if (ret == -1) {
4623 printf("Unable to get runstate response from node %u\n",
4624 options.pnn);
4625 return -1;
4626 } else {
4627 bool found = true;
4628 enum ctdb_runstate t;
4629 int i;
4630 for (i=0; i<argc; i++) {
4631 found = false;
4632 t = runstate_from_string(argv[i]);
4633 if (t == CTDB_RUNSTATE_UNKNOWN) {
4634 printf("Invalid run state (%s)\n", argv[i]);
4635 return -1;
4638 if (t == runstate) {
4639 found = true;
4640 break;
4644 if (!found) {
4645 printf("CTDB not in required run state (got %s)\n",
4646 runstate_to_string((enum ctdb_runstate)runstate));
4647 return -1;
4651 printf("%s\n", runstate_to_string(runstate));
4652 return 0;
4657 get a tunable
4659 static int control_getvar(struct ctdb_context *ctdb, int argc, const char **argv)
4661 const char *name;
4662 uint32_t value;
4663 int ret;
4665 if (argc < 1) {
4666 usage();
4669 name = argv[0];
4670 ret = ctdb_ctrl_get_tunable(ctdb, TIMELIMIT(), options.pnn, name, &value);
4671 if (ret != 0) {
4672 DEBUG(DEBUG_ERR, ("Unable to get tunable variable '%s'\n", name));
4673 return -1;
4676 printf("%-23s = %u\n", name, value);
4677 return 0;
4681 set a tunable
4683 static int control_setvar(struct ctdb_context *ctdb, int argc, const char **argv)
4685 const char *name;
4686 uint32_t value;
4687 int ret;
4689 if (argc < 2) {
4690 usage();
4693 name = argv[0];
4694 value = strtoul(argv[1], NULL, 0);
4696 ret = ctdb_ctrl_set_tunable(ctdb, TIMELIMIT(), options.pnn, name, value);
4697 if (ret == -1) {
4698 DEBUG(DEBUG_ERR, ("Unable to set tunable variable '%s'\n", name));
4699 return -1;
4701 return 0;
4705 list all tunables
4707 static int control_listvars(struct ctdb_context *ctdb, int argc, const char **argv)
4709 uint32_t count;
4710 const char **list;
4711 int ret, i;
4713 ret = ctdb_ctrl_list_tunables(ctdb, TIMELIMIT(), options.pnn, ctdb, &list, &count);
4714 if (ret == -1) {
4715 DEBUG(DEBUG_ERR, ("Unable to list tunable variables\n"));
4716 return -1;
4719 for (i=0;i<count;i++) {
4720 control_getvar(ctdb, 1, &list[i]);
4723 talloc_free(list);
4725 return 0;
4729 display debug level on a node
4731 static int control_getdebug(struct ctdb_context *ctdb, int argc, const char **argv)
4733 int ret;
4734 int32_t level;
4736 ret = ctdb_ctrl_get_debuglevel(ctdb, options.pnn, &level);
4737 if (ret != 0) {
4738 DEBUG(DEBUG_ERR, ("Unable to get debuglevel response from node %u\n", options.pnn));
4739 return ret;
4740 } else {
4741 if (options.machinereadable){
4742 printf(":Name:Level:\n");
4743 printf(":%s:%d:\n",get_debug_by_level(level),level);
4744 } else {
4745 printf("Node %u is at debug level %s (%d)\n", options.pnn, get_debug_by_level(level), level);
4748 return 0;
4752 display reclock file of a node
4754 static int control_getreclock(struct ctdb_context *ctdb, int argc, const char **argv)
4756 int ret;
4757 const char *reclock;
4759 ret = ctdb_ctrl_getreclock(ctdb, TIMELIMIT(), options.pnn, ctdb, &reclock);
4760 if (ret != 0) {
4761 DEBUG(DEBUG_ERR, ("Unable to get reclock file from node %u\n", options.pnn));
4762 return ret;
4763 } else {
4764 if (options.machinereadable){
4765 if (reclock != NULL) {
4766 printf("%s", reclock);
4768 } else {
4769 if (reclock == NULL) {
4770 printf("No reclock file used.\n");
4771 } else {
4772 printf("Reclock file:%s\n", reclock);
4776 return 0;
4780 set the reclock file of a node
4782 static int control_setreclock(struct ctdb_context *ctdb, int argc, const char **argv)
4784 int ret;
4785 const char *reclock;
4787 if (argc == 0) {
4788 reclock = NULL;
4789 } else if (argc == 1) {
4790 reclock = argv[0];
4791 } else {
4792 usage();
4795 ret = ctdb_ctrl_setreclock(ctdb, TIMELIMIT(), options.pnn, reclock);
4796 if (ret != 0) {
4797 DEBUG(DEBUG_ERR, ("Unable to get reclock file from node %u\n", options.pnn));
4798 return ret;
4800 return 0;
4804 set the natgw state on/off
4806 static int control_setnatgwstate(struct ctdb_context *ctdb, int argc, const char **argv)
4808 int ret;
4809 uint32_t natgwstate;
4811 if (argc == 0) {
4812 usage();
4815 if (!strcmp(argv[0], "on")) {
4816 natgwstate = 1;
4817 } else if (!strcmp(argv[0], "off")) {
4818 natgwstate = 0;
4819 } else {
4820 usage();
4823 ret = ctdb_ctrl_setnatgwstate(ctdb, TIMELIMIT(), options.pnn, natgwstate);
4824 if (ret != 0) {
4825 DEBUG(DEBUG_ERR, ("Unable to set the natgw state for node %u\n", options.pnn));
4826 return ret;
4829 return 0;
4833 set the lmaster role on/off
4835 static int control_setlmasterrole(struct ctdb_context *ctdb, int argc, const char **argv)
4837 int ret;
4838 uint32_t lmasterrole;
4840 if (argc == 0) {
4841 usage();
4844 if (!strcmp(argv[0], "on")) {
4845 lmasterrole = 1;
4846 } else if (!strcmp(argv[0], "off")) {
4847 lmasterrole = 0;
4848 } else {
4849 usage();
4852 ret = ctdb_ctrl_setlmasterrole(ctdb, TIMELIMIT(), options.pnn, lmasterrole);
4853 if (ret != 0) {
4854 DEBUG(DEBUG_ERR, ("Unable to set the lmaster role for node %u\n", options.pnn));
4855 return ret;
4858 return 0;
4862 set the recmaster role on/off
4864 static int control_setrecmasterrole(struct ctdb_context *ctdb, int argc, const char **argv)
4866 int ret;
4867 uint32_t recmasterrole;
4869 if (argc == 0) {
4870 usage();
4873 if (!strcmp(argv[0], "on")) {
4874 recmasterrole = 1;
4875 } else if (!strcmp(argv[0], "off")) {
4876 recmasterrole = 0;
4877 } else {
4878 usage();
4881 ret = ctdb_ctrl_setrecmasterrole(ctdb, TIMELIMIT(), options.pnn, recmasterrole);
4882 if (ret != 0) {
4883 DEBUG(DEBUG_ERR, ("Unable to set the recmaster role for node %u\n", options.pnn));
4884 return ret;
4887 return 0;
4891 set debug level on a node or all nodes
4893 static int control_setdebug(struct ctdb_context *ctdb, int argc, const char **argv)
4895 int i, ret;
4896 int32_t level;
4898 if (argc == 0) {
4899 printf("You must specify the debug level. Valid levels are:\n");
4900 for (i=0; debug_levels[i].description != NULL; i++) {
4901 printf("%s (%d)\n", debug_levels[i].description, debug_levels[i].level);
4904 return 0;
4907 if (isalpha(argv[0][0]) || argv[0][0] == '-') {
4908 level = get_debug_by_desc(argv[0]);
4909 } else {
4910 level = strtol(argv[0], NULL, 0);
4913 for (i=0; debug_levels[i].description != NULL; i++) {
4914 if (level == debug_levels[i].level) {
4915 break;
4918 if (debug_levels[i].description == NULL) {
4919 printf("Invalid debug level, must be one of\n");
4920 for (i=0; debug_levels[i].description != NULL; i++) {
4921 printf("%s (%d)\n", debug_levels[i].description, debug_levels[i].level);
4923 return -1;
4926 ret = ctdb_ctrl_set_debuglevel(ctdb, options.pnn, level);
4927 if (ret != 0) {
4928 DEBUG(DEBUG_ERR, ("Unable to set debug level on node %u\n", options.pnn));
4930 return 0;
4935 thaw a node
4937 static int control_thaw(struct ctdb_context *ctdb, int argc, const char **argv)
4939 int ret;
4940 uint32_t priority;
4942 if (argc == 1) {
4943 priority = strtol(argv[0], NULL, 0);
4944 } else {
4945 priority = 0;
4947 DEBUG(DEBUG_ERR,("Thaw by priority %u\n", priority));
4949 ret = ctdb_ctrl_thaw_priority(ctdb, TIMELIMIT(), options.pnn, priority);
4950 if (ret != 0) {
4951 DEBUG(DEBUG_ERR, ("Unable to thaw node %u\n", options.pnn));
4953 return 0;
4958 attach to a database
4960 static int control_attach(struct ctdb_context *ctdb, int argc, const char **argv)
4962 const char *db_name;
4963 struct ctdb_db_context *ctdb_db;
4964 bool persistent = false;
4966 if (argc < 1) {
4967 usage();
4969 db_name = argv[0];
4970 if (argc > 2) {
4971 usage();
4973 if (argc == 2) {
4974 if (strcmp(argv[1], "persistent") != 0) {
4975 usage();
4977 persistent = true;
4980 ctdb_db = ctdb_attach(ctdb, TIMELIMIT(), db_name, persistent, 0);
4981 if (ctdb_db == NULL) {
4982 DEBUG(DEBUG_ERR,("Unable to attach to database '%s'\n", db_name));
4983 return -1;
4986 return 0;
4990 set db priority
4992 static int control_setdbprio(struct ctdb_context *ctdb, int argc, const char **argv)
4994 struct ctdb_db_priority db_prio;
4995 int ret;
4997 if (argc < 2) {
4998 usage();
5001 db_prio.db_id = strtoul(argv[0], NULL, 0);
5002 db_prio.priority = strtoul(argv[1], NULL, 0);
5004 ret = ctdb_ctrl_set_db_priority(ctdb, TIMELIMIT(), options.pnn, &db_prio);
5005 if (ret != 0) {
5006 DEBUG(DEBUG_ERR,("Unable to set db prio\n"));
5007 return -1;
5010 return 0;
5014 get db priority
5016 static int control_getdbprio(struct ctdb_context *ctdb, int argc, const char **argv)
5018 uint32_t db_id, priority;
5019 int ret;
5021 if (argc < 1) {
5022 usage();
5025 if (!db_exists(ctdb, argv[0], &db_id, NULL)) {
5026 return -1;
5029 ret = ctdb_ctrl_get_db_priority(ctdb, TIMELIMIT(), options.pnn, db_id, &priority);
5030 if (ret != 0) {
5031 DEBUG(DEBUG_ERR,("Unable to get db prio\n"));
5032 return -1;
5035 DEBUG(DEBUG_ERR,("Priority:%u\n", priority));
5037 return 0;
5041 set the sticky records capability for a database
5043 static int control_setdbsticky(struct ctdb_context *ctdb, int argc, const char **argv)
5045 TALLOC_CTX *tmp_ctx = talloc_new(ctdb);
5046 uint32_t db_id;
5047 int ret;
5049 if (argc < 1) {
5050 usage();
5053 if (!db_exists(ctdb, argv[0], &db_id, NULL)) {
5054 return -1;
5057 ret = ctdb_ctrl_set_db_sticky(ctdb, options.pnn, db_id);
5058 if (ret != 0) {
5059 DEBUG(DEBUG_ERR,("Unable to set db to support sticky records\n"));
5060 talloc_free(tmp_ctx);
5061 return -1;
5064 talloc_free(tmp_ctx);
5065 return 0;
5069 set the readonly capability for a database
5071 static int control_setdbreadonly(struct ctdb_context *ctdb, int argc, const char **argv)
5073 TALLOC_CTX *tmp_ctx = talloc_new(ctdb);
5074 uint32_t db_id;
5075 int ret;
5077 if (argc < 1) {
5078 usage();
5081 if (!db_exists(ctdb, argv[0], &db_id, NULL)) {
5082 return -1;
5085 ret = ctdb_ctrl_set_db_readonly(ctdb, options.pnn, db_id);
5086 if (ret != 0) {
5087 DEBUG(DEBUG_ERR,("Unable to set db to support readonly\n"));
5088 talloc_free(tmp_ctx);
5089 return -1;
5092 talloc_free(tmp_ctx);
5093 return 0;
5097 get db seqnum
5099 static int control_getdbseqnum(struct ctdb_context *ctdb, int argc, const char **argv)
5101 uint32_t db_id;
5102 uint64_t seqnum;
5103 int ret;
5105 if (argc < 1) {
5106 usage();
5109 if (!db_exists(ctdb, argv[0], &db_id, NULL)) {
5110 return -1;
5113 ret = ctdb_ctrl_getdbseqnum(ctdb, TIMELIMIT(), options.pnn, db_id, &seqnum);
5114 if (ret != 0) {
5115 DEBUG(DEBUG_ERR, ("Unable to get seqnum from node."));
5116 return -1;
5119 printf("Sequence number:%lld\n", (long long)seqnum);
5121 return 0;
5125 run an eventscript on a node
5127 static int control_eventscript(struct ctdb_context *ctdb, int argc, const char **argv)
5129 TDB_DATA data;
5130 int ret;
5131 int32_t res;
5132 char *errmsg;
5133 TALLOC_CTX *tmp_ctx = talloc_new(ctdb);
5135 if (argc != 1) {
5136 DEBUG(DEBUG_ERR,("Invalid arguments\n"));
5137 return -1;
5140 data.dptr = (unsigned char *)discard_const(argv[0]);
5141 data.dsize = strlen((char *)data.dptr) + 1;
5143 DEBUG(DEBUG_ERR, ("Running eventscripts with arguments \"%s\" on node %u\n", data.dptr, options.pnn));
5145 ret = ctdb_control(ctdb, options.pnn, 0, CTDB_CONTROL_RUN_EVENTSCRIPTS,
5146 0, data, tmp_ctx, NULL, &res, NULL, &errmsg);
5147 if (ret != 0 || res != 0) {
5148 DEBUG(DEBUG_ERR,("Failed to run eventscripts - %s\n", errmsg));
5149 talloc_free(tmp_ctx);
5150 return -1;
5152 talloc_free(tmp_ctx);
5153 return 0;
5156 #define DB_VERSION 1
5157 #define MAX_DB_NAME 64
5158 struct db_file_header {
5159 unsigned long version;
5160 time_t timestamp;
5161 unsigned long persistent;
5162 unsigned long size;
5163 const char name[MAX_DB_NAME];
5166 struct backup_data {
5167 struct ctdb_marshall_buffer *records;
5168 uint32_t len;
5169 uint32_t total;
5170 bool traverse_error;
5173 static int backup_traverse(struct tdb_context *tdb, TDB_DATA key, TDB_DATA data, void *private)
5175 struct backup_data *bd = talloc_get_type(private, struct backup_data);
5176 struct ctdb_rec_data *rec;
5178 /* add the record */
5179 rec = ctdb_marshall_record(bd->records, 0, key, NULL, data);
5180 if (rec == NULL) {
5181 bd->traverse_error = true;
5182 DEBUG(DEBUG_ERR,("Failed to marshall record\n"));
5183 return -1;
5185 bd->records = talloc_realloc_size(NULL, bd->records, rec->length + bd->len);
5186 if (bd->records == NULL) {
5187 DEBUG(DEBUG_ERR,("Failed to expand marshalling buffer\n"));
5188 bd->traverse_error = true;
5189 return -1;
5191 bd->records->count++;
5192 memcpy(bd->len+(uint8_t *)bd->records, rec, rec->length);
5193 bd->len += rec->length;
5194 talloc_free(rec);
5196 bd->total++;
5197 return 0;
5201 * backup a database to a file
5203 static int control_backupdb(struct ctdb_context *ctdb, int argc, const char **argv)
5205 int ret;
5206 TALLOC_CTX *tmp_ctx = talloc_new(ctdb);
5207 struct db_file_header dbhdr;
5208 struct ctdb_db_context *ctdb_db;
5209 struct backup_data *bd;
5210 int fh = -1;
5211 int status = -1;
5212 const char *reason = NULL;
5213 uint32_t db_id;
5214 uint8_t flags;
5216 assert_single_node_only();
5218 if (argc != 2) {
5219 DEBUG(DEBUG_ERR,("Invalid arguments\n"));
5220 return -1;
5223 if (!db_exists(ctdb, argv[0], &db_id, &flags)) {
5224 return -1;
5227 ret = ctdb_ctrl_getdbhealth(ctdb, TIMELIMIT(), options.pnn,
5228 db_id, tmp_ctx, &reason);
5229 if (ret != 0) {
5230 DEBUG(DEBUG_ERR,("Unable to get dbhealth for database '%s'\n",
5231 argv[0]));
5232 talloc_free(tmp_ctx);
5233 return -1;
5235 if (reason) {
5236 uint32_t allow_unhealthy = 0;
5238 ctdb_ctrl_get_tunable(ctdb, TIMELIMIT(), options.pnn,
5239 "AllowUnhealthyDBRead",
5240 &allow_unhealthy);
5242 if (allow_unhealthy != 1) {
5243 DEBUG(DEBUG_ERR,("database '%s' is unhealthy: %s\n",
5244 argv[0], reason));
5246 DEBUG(DEBUG_ERR,("disallow backup : tunable AllowUnhealthyDBRead = %u\n",
5247 allow_unhealthy));
5248 talloc_free(tmp_ctx);
5249 return -1;
5252 DEBUG(DEBUG_WARNING,("WARNING database '%s' is unhealthy - see 'ctdb getdbstatus %s'\n",
5253 argv[0], argv[0]));
5254 DEBUG(DEBUG_WARNING,("WARNING! allow backup of unhealthy database: "
5255 "tunnable AllowUnhealthyDBRead = %u\n",
5256 allow_unhealthy));
5259 ctdb_db = ctdb_attach(ctdb, TIMELIMIT(), argv[0], flags & CTDB_DB_FLAGS_PERSISTENT, 0);
5260 if (ctdb_db == NULL) {
5261 DEBUG(DEBUG_ERR,("Unable to attach to database '%s'\n", argv[0]));
5262 talloc_free(tmp_ctx);
5263 return -1;
5267 ret = tdb_transaction_start(ctdb_db->ltdb->tdb);
5268 if (ret == -1) {
5269 DEBUG(DEBUG_ERR,("Failed to start transaction\n"));
5270 talloc_free(tmp_ctx);
5271 return -1;
5275 bd = talloc_zero(tmp_ctx, struct backup_data);
5276 if (bd == NULL) {
5277 DEBUG(DEBUG_ERR,("Failed to allocate backup_data\n"));
5278 talloc_free(tmp_ctx);
5279 return -1;
5282 bd->records = talloc_zero(bd, struct ctdb_marshall_buffer);
5283 if (bd->records == NULL) {
5284 DEBUG(DEBUG_ERR,("Failed to allocate ctdb_marshall_buffer\n"));
5285 talloc_free(tmp_ctx);
5286 return -1;
5289 bd->len = offsetof(struct ctdb_marshall_buffer, data);
5290 bd->records->db_id = ctdb_db->db_id;
5291 /* traverse the database collecting all records */
5292 if (tdb_traverse_read(ctdb_db->ltdb->tdb, backup_traverse, bd) == -1 ||
5293 bd->traverse_error) {
5294 DEBUG(DEBUG_ERR,("Traverse error\n"));
5295 talloc_free(tmp_ctx);
5296 return -1;
5299 tdb_transaction_cancel(ctdb_db->ltdb->tdb);
5302 fh = open(argv[1], O_RDWR|O_CREAT, 0600);
5303 if (fh == -1) {
5304 DEBUG(DEBUG_ERR,("Failed to open file '%s'\n", argv[1]));
5305 talloc_free(tmp_ctx);
5306 return -1;
5309 dbhdr.version = DB_VERSION;
5310 dbhdr.timestamp = time(NULL);
5311 dbhdr.persistent = flags & CTDB_DB_FLAGS_PERSISTENT;
5312 dbhdr.size = bd->len;
5313 if (strlen(argv[0]) >= MAX_DB_NAME) {
5314 DEBUG(DEBUG_ERR,("Too long dbname\n"));
5315 goto done;
5317 strncpy(discard_const(dbhdr.name), argv[0], MAX_DB_NAME);
5318 ret = write(fh, &dbhdr, sizeof(dbhdr));
5319 if (ret == -1) {
5320 DEBUG(DEBUG_ERR,("write failed: %s\n", strerror(errno)));
5321 goto done;
5323 ret = write(fh, bd->records, bd->len);
5324 if (ret == -1) {
5325 DEBUG(DEBUG_ERR,("write failed: %s\n", strerror(errno)));
5326 goto done;
5329 status = 0;
5330 done:
5331 if (fh != -1) {
5332 ret = close(fh);
5333 if (ret == -1) {
5334 DEBUG(DEBUG_ERR,("close failed: %s\n", strerror(errno)));
5338 DEBUG(DEBUG_ERR,("Database backed up to %s\n", argv[1]));
5340 talloc_free(tmp_ctx);
5341 return status;
5345 * restore a database from a file
5347 static int control_restoredb(struct ctdb_context *ctdb, int argc, const char **argv)
5349 int ret;
5350 TALLOC_CTX *tmp_ctx = talloc_new(ctdb);
5351 TDB_DATA outdata;
5352 TDB_DATA data;
5353 struct db_file_header dbhdr;
5354 struct ctdb_db_context *ctdb_db;
5355 struct ctdb_node_map *nodemap=NULL;
5356 struct ctdb_vnn_map *vnnmap=NULL;
5357 int i, fh;
5358 struct ctdb_control_wipe_database w;
5359 uint32_t *nodes;
5360 uint32_t generation;
5361 struct tm *tm;
5362 char tbuf[100];
5363 char *dbname;
5365 assert_single_node_only();
5367 if (argc < 1 || argc > 2) {
5368 DEBUG(DEBUG_ERR,("Invalid arguments\n"));
5369 return -1;
5372 fh = open(argv[0], O_RDONLY);
5373 if (fh == -1) {
5374 DEBUG(DEBUG_ERR,("Failed to open file '%s'\n", argv[0]));
5375 talloc_free(tmp_ctx);
5376 return -1;
5379 read(fh, &dbhdr, sizeof(dbhdr));
5380 if (dbhdr.version != DB_VERSION) {
5381 DEBUG(DEBUG_ERR,("Invalid version of database dump. File is version %lu but expected version was %u\n", dbhdr.version, DB_VERSION));
5382 talloc_free(tmp_ctx);
5383 return -1;
5386 dbname = discard_const(dbhdr.name);
5387 if (argc == 2) {
5388 dbname = discard_const(argv[1]);
5391 outdata.dsize = dbhdr.size;
5392 outdata.dptr = talloc_size(tmp_ctx, outdata.dsize);
5393 if (outdata.dptr == NULL) {
5394 DEBUG(DEBUG_ERR,("Failed to allocate data of size '%lu'\n", dbhdr.size));
5395 close(fh);
5396 talloc_free(tmp_ctx);
5397 return -1;
5399 read(fh, outdata.dptr, outdata.dsize);
5400 close(fh);
5402 tm = localtime(&dbhdr.timestamp);
5403 strftime(tbuf,sizeof(tbuf)-1,"%Y/%m/%d %H:%M:%S", tm);
5404 printf("Restoring database '%s' from backup @ %s\n",
5405 dbname, tbuf);
5408 ctdb_db = ctdb_attach(ctdb, TIMELIMIT(), dbname, dbhdr.persistent, 0);
5409 if (ctdb_db == NULL) {
5410 DEBUG(DEBUG_ERR,("Unable to attach to database '%s'\n", dbname));
5411 talloc_free(tmp_ctx);
5412 return -1;
5415 ret = ctdb_ctrl_getnodemap(ctdb, TIMELIMIT(), options.pnn, ctdb, &nodemap);
5416 if (ret != 0) {
5417 DEBUG(DEBUG_ERR, ("Unable to get nodemap from node %u\n", options.pnn));
5418 talloc_free(tmp_ctx);
5419 return ret;
5423 ret = ctdb_ctrl_getvnnmap(ctdb, TIMELIMIT(), options.pnn, tmp_ctx, &vnnmap);
5424 if (ret != 0) {
5425 DEBUG(DEBUG_ERR, ("Unable to get vnnmap from node %u\n", options.pnn));
5426 talloc_free(tmp_ctx);
5427 return ret;
5430 /* freeze all nodes */
5431 nodes = list_of_active_nodes(ctdb, nodemap, tmp_ctx, true);
5432 for (i=1; i<=NUM_DB_PRIORITIES; i++) {
5433 if (ctdb_client_async_control(ctdb, CTDB_CONTROL_FREEZE,
5434 nodes, i,
5435 TIMELIMIT(),
5436 false, tdb_null,
5437 NULL, NULL,
5438 NULL) != 0) {
5439 DEBUG(DEBUG_ERR, ("Unable to freeze nodes.\n"));
5440 ctdb_ctrl_setrecmode(ctdb, TIMELIMIT(), options.pnn, CTDB_RECOVERY_ACTIVE);
5441 talloc_free(tmp_ctx);
5442 return -1;
5446 generation = vnnmap->generation;
5447 data.dptr = (void *)&generation;
5448 data.dsize = sizeof(generation);
5450 /* start a cluster wide transaction */
5451 nodes = list_of_active_nodes(ctdb, nodemap, tmp_ctx, true);
5452 if (ctdb_client_async_control(ctdb, CTDB_CONTROL_TRANSACTION_START,
5453 nodes, 0,
5454 TIMELIMIT(), false, data,
5455 NULL, NULL,
5456 NULL) != 0) {
5457 DEBUG(DEBUG_ERR, ("Unable to start cluster wide transactions.\n"));
5458 return -1;
5462 w.db_id = ctdb_db->db_id;
5463 w.transaction_id = generation;
5465 data.dptr = (void *)&w;
5466 data.dsize = sizeof(w);
5468 /* wipe all the remote databases. */
5469 nodes = list_of_active_nodes(ctdb, nodemap, tmp_ctx, true);
5470 if (ctdb_client_async_control(ctdb, CTDB_CONTROL_WIPE_DATABASE,
5471 nodes, 0,
5472 TIMELIMIT(), false, data,
5473 NULL, NULL,
5474 NULL) != 0) {
5475 DEBUG(DEBUG_ERR, ("Unable to wipe database.\n"));
5476 ctdb_ctrl_setrecmode(ctdb, TIMELIMIT(), options.pnn, CTDB_RECOVERY_ACTIVE);
5477 talloc_free(tmp_ctx);
5478 return -1;
5481 /* push the database */
5482 nodes = list_of_active_nodes(ctdb, nodemap, tmp_ctx, true);
5483 if (ctdb_client_async_control(ctdb, CTDB_CONTROL_PUSH_DB,
5484 nodes, 0,
5485 TIMELIMIT(), false, outdata,
5486 NULL, NULL,
5487 NULL) != 0) {
5488 DEBUG(DEBUG_ERR, ("Failed to push database.\n"));
5489 ctdb_ctrl_setrecmode(ctdb, TIMELIMIT(), options.pnn, CTDB_RECOVERY_ACTIVE);
5490 talloc_free(tmp_ctx);
5491 return -1;
5494 data.dptr = (void *)&ctdb_db->db_id;
5495 data.dsize = sizeof(ctdb_db->db_id);
5497 /* mark the database as healthy */
5498 nodes = list_of_active_nodes(ctdb, nodemap, tmp_ctx, true);
5499 if (ctdb_client_async_control(ctdb, CTDB_CONTROL_DB_SET_HEALTHY,
5500 nodes, 0,
5501 TIMELIMIT(), false, data,
5502 NULL, NULL,
5503 NULL) != 0) {
5504 DEBUG(DEBUG_ERR, ("Failed to mark database as healthy.\n"));
5505 ctdb_ctrl_setrecmode(ctdb, TIMELIMIT(), options.pnn, CTDB_RECOVERY_ACTIVE);
5506 talloc_free(tmp_ctx);
5507 return -1;
5510 data.dptr = (void *)&generation;
5511 data.dsize = sizeof(generation);
5513 /* commit all the changes */
5514 if (ctdb_client_async_control(ctdb, CTDB_CONTROL_TRANSACTION_COMMIT,
5515 nodes, 0,
5516 TIMELIMIT(), false, data,
5517 NULL, NULL,
5518 NULL) != 0) {
5519 DEBUG(DEBUG_ERR, ("Unable to commit databases.\n"));
5520 ctdb_ctrl_setrecmode(ctdb, TIMELIMIT(), options.pnn, CTDB_RECOVERY_ACTIVE);
5521 talloc_free(tmp_ctx);
5522 return -1;
5526 /* thaw all nodes */
5527 nodes = list_of_active_nodes(ctdb, nodemap, tmp_ctx, true);
5528 if (ctdb_client_async_control(ctdb, CTDB_CONTROL_THAW,
5529 nodes, 0,
5530 TIMELIMIT(),
5531 false, tdb_null,
5532 NULL, NULL,
5533 NULL) != 0) {
5534 DEBUG(DEBUG_ERR, ("Unable to thaw nodes.\n"));
5535 ctdb_ctrl_setrecmode(ctdb, TIMELIMIT(), options.pnn, CTDB_RECOVERY_ACTIVE);
5536 talloc_free(tmp_ctx);
5537 return -1;
5541 talloc_free(tmp_ctx);
5542 return 0;
5546 * dump a database backup from a file
5548 static int control_dumpdbbackup(struct ctdb_context *ctdb, int argc, const char **argv)
5550 TALLOC_CTX *tmp_ctx = talloc_new(ctdb);
5551 TDB_DATA outdata;
5552 struct db_file_header dbhdr;
5553 int i, fh;
5554 struct tm *tm;
5555 char tbuf[100];
5556 struct ctdb_rec_data *rec = NULL;
5557 struct ctdb_marshall_buffer *m;
5558 struct ctdb_dump_db_context c;
5560 assert_single_node_only();
5562 if (argc != 1) {
5563 DEBUG(DEBUG_ERR,("Invalid arguments\n"));
5564 return -1;
5567 fh = open(argv[0], O_RDONLY);
5568 if (fh == -1) {
5569 DEBUG(DEBUG_ERR,("Failed to open file '%s'\n", argv[0]));
5570 talloc_free(tmp_ctx);
5571 return -1;
5574 read(fh, &dbhdr, sizeof(dbhdr));
5575 if (dbhdr.version != DB_VERSION) {
5576 DEBUG(DEBUG_ERR,("Invalid version of database dump. File is version %lu but expected version was %u\n", dbhdr.version, DB_VERSION));
5577 talloc_free(tmp_ctx);
5578 return -1;
5581 outdata.dsize = dbhdr.size;
5582 outdata.dptr = talloc_size(tmp_ctx, outdata.dsize);
5583 if (outdata.dptr == NULL) {
5584 DEBUG(DEBUG_ERR,("Failed to allocate data of size '%lu'\n", dbhdr.size));
5585 close(fh);
5586 talloc_free(tmp_ctx);
5587 return -1;
5589 read(fh, outdata.dptr, outdata.dsize);
5590 close(fh);
5591 m = (struct ctdb_marshall_buffer *)outdata.dptr;
5593 tm = localtime(&dbhdr.timestamp);
5594 strftime(tbuf,sizeof(tbuf)-1,"%Y/%m/%d %H:%M:%S", tm);
5595 printf("Backup of database name:'%s' dbid:0x%x08x from @ %s\n",
5596 dbhdr.name, m->db_id, tbuf);
5598 ZERO_STRUCT(c);
5599 c.f = stdout;
5600 c.printemptyrecords = (bool)options.printemptyrecords;
5601 c.printdatasize = (bool)options.printdatasize;
5602 c.printlmaster = false;
5603 c.printhash = (bool)options.printhash;
5604 c.printrecordflags = (bool)options.printrecordflags;
5606 for (i=0; i < m->count; i++) {
5607 uint32_t reqid = 0;
5608 TDB_DATA key, data;
5610 /* we do not want the header splitted, so we pass NULL*/
5611 rec = ctdb_marshall_loop_next(m, rec, &reqid,
5612 NULL, &key, &data);
5614 ctdb_dumpdb_record(ctdb, key, data, &c);
5617 printf("Dumped %d records\n", i);
5618 talloc_free(tmp_ctx);
5619 return 0;
5623 * wipe a database from a file
5625 static int control_wipedb(struct ctdb_context *ctdb, int argc,
5626 const char **argv)
5628 int ret;
5629 TALLOC_CTX *tmp_ctx = talloc_new(ctdb);
5630 TDB_DATA data;
5631 struct ctdb_db_context *ctdb_db;
5632 struct ctdb_node_map *nodemap = NULL;
5633 struct ctdb_vnn_map *vnnmap = NULL;
5634 int i;
5635 struct ctdb_control_wipe_database w;
5636 uint32_t *nodes;
5637 uint32_t generation;
5638 uint8_t flags;
5640 assert_single_node_only();
5642 if (argc != 1) {
5643 DEBUG(DEBUG_ERR,("Invalid arguments\n"));
5644 return -1;
5647 if (!db_exists(ctdb, argv[0], NULL, &flags)) {
5648 return -1;
5651 ctdb_db = ctdb_attach(ctdb, TIMELIMIT(), argv[0], flags & CTDB_DB_FLAGS_PERSISTENT, 0);
5652 if (ctdb_db == NULL) {
5653 DEBUG(DEBUG_ERR, ("Unable to attach to database '%s'\n",
5654 argv[0]));
5655 talloc_free(tmp_ctx);
5656 return -1;
5659 ret = ctdb_ctrl_getnodemap(ctdb, TIMELIMIT(), options.pnn, ctdb,
5660 &nodemap);
5661 if (ret != 0) {
5662 DEBUG(DEBUG_ERR, ("Unable to get nodemap from node %u\n",
5663 options.pnn));
5664 talloc_free(tmp_ctx);
5665 return ret;
5668 ret = ctdb_ctrl_getvnnmap(ctdb, TIMELIMIT(), options.pnn, tmp_ctx,
5669 &vnnmap);
5670 if (ret != 0) {
5671 DEBUG(DEBUG_ERR, ("Unable to get vnnmap from node %u\n",
5672 options.pnn));
5673 talloc_free(tmp_ctx);
5674 return ret;
5677 /* freeze all nodes */
5678 nodes = list_of_active_nodes(ctdb, nodemap, tmp_ctx, true);
5679 for (i=1; i<=NUM_DB_PRIORITIES; i++) {
5680 ret = ctdb_client_async_control(ctdb, CTDB_CONTROL_FREEZE,
5681 nodes, i,
5682 TIMELIMIT(),
5683 false, tdb_null,
5684 NULL, NULL,
5685 NULL);
5686 if (ret != 0) {
5687 DEBUG(DEBUG_ERR, ("Unable to freeze nodes.\n"));
5688 ctdb_ctrl_setrecmode(ctdb, TIMELIMIT(), options.pnn,
5689 CTDB_RECOVERY_ACTIVE);
5690 talloc_free(tmp_ctx);
5691 return -1;
5695 generation = vnnmap->generation;
5696 data.dptr = (void *)&generation;
5697 data.dsize = sizeof(generation);
5699 /* start a cluster wide transaction */
5700 nodes = list_of_active_nodes(ctdb, nodemap, tmp_ctx, true);
5701 ret = ctdb_client_async_control(ctdb, CTDB_CONTROL_TRANSACTION_START,
5702 nodes, 0,
5703 TIMELIMIT(), false, data,
5704 NULL, NULL,
5705 NULL);
5706 if (ret!= 0) {
5707 DEBUG(DEBUG_ERR, ("Unable to start cluster wide "
5708 "transactions.\n"));
5709 return -1;
5712 w.db_id = ctdb_db->db_id;
5713 w.transaction_id = generation;
5715 data.dptr = (void *)&w;
5716 data.dsize = sizeof(w);
5718 /* wipe all the remote databases. */
5719 nodes = list_of_active_nodes(ctdb, nodemap, tmp_ctx, true);
5720 if (ctdb_client_async_control(ctdb, CTDB_CONTROL_WIPE_DATABASE,
5721 nodes, 0,
5722 TIMELIMIT(), false, data,
5723 NULL, NULL,
5724 NULL) != 0) {
5725 DEBUG(DEBUG_ERR, ("Unable to wipe database.\n"));
5726 ctdb_ctrl_setrecmode(ctdb, TIMELIMIT(), options.pnn, CTDB_RECOVERY_ACTIVE);
5727 talloc_free(tmp_ctx);
5728 return -1;
5731 data.dptr = (void *)&ctdb_db->db_id;
5732 data.dsize = sizeof(ctdb_db->db_id);
5734 /* mark the database as healthy */
5735 nodes = list_of_active_nodes(ctdb, nodemap, tmp_ctx, true);
5736 if (ctdb_client_async_control(ctdb, CTDB_CONTROL_DB_SET_HEALTHY,
5737 nodes, 0,
5738 TIMELIMIT(), false, data,
5739 NULL, NULL,
5740 NULL) != 0) {
5741 DEBUG(DEBUG_ERR, ("Failed to mark database as healthy.\n"));
5742 ctdb_ctrl_setrecmode(ctdb, TIMELIMIT(), options.pnn, CTDB_RECOVERY_ACTIVE);
5743 talloc_free(tmp_ctx);
5744 return -1;
5747 data.dptr = (void *)&generation;
5748 data.dsize = sizeof(generation);
5750 /* commit all the changes */
5751 if (ctdb_client_async_control(ctdb, CTDB_CONTROL_TRANSACTION_COMMIT,
5752 nodes, 0,
5753 TIMELIMIT(), false, data,
5754 NULL, NULL,
5755 NULL) != 0) {
5756 DEBUG(DEBUG_ERR, ("Unable to commit databases.\n"));
5757 ctdb_ctrl_setrecmode(ctdb, TIMELIMIT(), options.pnn, CTDB_RECOVERY_ACTIVE);
5758 talloc_free(tmp_ctx);
5759 return -1;
5762 /* thaw all nodes */
5763 nodes = list_of_active_nodes(ctdb, nodemap, tmp_ctx, true);
5764 if (ctdb_client_async_control(ctdb, CTDB_CONTROL_THAW,
5765 nodes, 0,
5766 TIMELIMIT(),
5767 false, tdb_null,
5768 NULL, NULL,
5769 NULL) != 0) {
5770 DEBUG(DEBUG_ERR, ("Unable to thaw nodes.\n"));
5771 ctdb_ctrl_setrecmode(ctdb, TIMELIMIT(), options.pnn, CTDB_RECOVERY_ACTIVE);
5772 talloc_free(tmp_ctx);
5773 return -1;
5776 DEBUG(DEBUG_ERR, ("Database wiped.\n"));
5778 talloc_free(tmp_ctx);
5779 return 0;
5783 dump memory usage
5785 static int control_dumpmemory(struct ctdb_context *ctdb, int argc, const char **argv)
5787 TDB_DATA data;
5788 int ret;
5789 int32_t res;
5790 char *errmsg;
5791 TALLOC_CTX *tmp_ctx = talloc_new(ctdb);
5792 ret = ctdb_control(ctdb, options.pnn, 0, CTDB_CONTROL_DUMP_MEMORY,
5793 0, tdb_null, tmp_ctx, &data, &res, NULL, &errmsg);
5794 if (ret != 0 || res != 0) {
5795 DEBUG(DEBUG_ERR,("Failed to dump memory - %s\n", errmsg));
5796 talloc_free(tmp_ctx);
5797 return -1;
5799 write(1, data.dptr, data.dsize);
5800 talloc_free(tmp_ctx);
5801 return 0;
5805 handler for memory dumps
5807 static void mem_dump_handler(struct ctdb_context *ctdb, uint64_t srvid,
5808 TDB_DATA data, void *private_data)
5810 write(1, data.dptr, data.dsize);
5811 exit(0);
5815 dump memory usage on the recovery daemon
5817 static int control_rddumpmemory(struct ctdb_context *ctdb, int argc, const char **argv)
5819 int ret;
5820 TDB_DATA data;
5821 struct srvid_request rd;
5823 rd.pnn = ctdb_get_pnn(ctdb);
5824 rd.srvid = getpid();
5826 /* register a message port for receiveing the reply so that we
5827 can receive the reply
5829 ctdb_client_set_message_handler(ctdb, rd.srvid, mem_dump_handler, NULL);
5832 data.dptr = (uint8_t *)&rd;
5833 data.dsize = sizeof(rd);
5835 ret = ctdb_client_send_message(ctdb, options.pnn, CTDB_SRVID_MEM_DUMP, data);
5836 if (ret != 0) {
5837 DEBUG(DEBUG_ERR,("Failed to send memdump request message to %u\n", options.pnn));
5838 return -1;
5841 /* this loop will terminate when we have received the reply */
5842 while (1) {
5843 event_loop_once(ctdb->ev);
5846 return 0;
5850 send a message to a srvid
5852 static int control_msgsend(struct ctdb_context *ctdb, int argc, const char **argv)
5854 unsigned long srvid;
5855 int ret;
5856 TDB_DATA data;
5858 if (argc < 2) {
5859 usage();
5862 srvid = strtoul(argv[0], NULL, 0);
5864 data.dptr = (uint8_t *)discard_const(argv[1]);
5865 data.dsize= strlen(argv[1]);
5867 ret = ctdb_client_send_message(ctdb, CTDB_BROADCAST_CONNECTED, srvid, data);
5868 if (ret != 0) {
5869 DEBUG(DEBUG_ERR,("Failed to send memdump request message to %u\n", options.pnn));
5870 return -1;
5873 return 0;
5877 handler for msglisten
5879 static void msglisten_handler(struct ctdb_context *ctdb, uint64_t srvid,
5880 TDB_DATA data, void *private_data)
5882 int i;
5884 printf("Message received: ");
5885 for (i=0;i<data.dsize;i++) {
5886 printf("%c", data.dptr[i]);
5888 printf("\n");
5892 listen for messages on a messageport
5894 static int control_msglisten(struct ctdb_context *ctdb, int argc, const char **argv)
5896 uint64_t srvid;
5898 srvid = getpid();
5900 /* register a message port and listen for messages
5902 ctdb_client_set_message_handler(ctdb, srvid, msglisten_handler, NULL);
5903 printf("Listening for messages on srvid:%d\n", (int)srvid);
5905 while (1) {
5906 event_loop_once(ctdb->ev);
5909 return 0;
5913 list all nodes in the cluster
5914 we parse the nodes file directly
5916 static int control_listnodes(struct ctdb_context *ctdb, int argc, const char **argv)
5918 TALLOC_CTX *mem_ctx = talloc_new(NULL);
5919 struct pnn_node *pnn_nodes;
5920 struct pnn_node *pnn_node;
5922 assert_single_node_only();
5924 pnn_nodes = read_nodes_file(mem_ctx);
5925 if (pnn_nodes == NULL) {
5926 DEBUG(DEBUG_ERR,("Failed to read nodes file\n"));
5927 talloc_free(mem_ctx);
5928 return -1;
5931 for(pnn_node=pnn_nodes;pnn_node;pnn_node=pnn_node->next) {
5932 ctdb_sock_addr addr;
5933 if (parse_ip(pnn_node->addr, NULL, 63999, &addr) == 0) {
5934 DEBUG(DEBUG_ERR,("Wrongly formed ip address '%s' in nodes file\n", pnn_node->addr));
5935 talloc_free(mem_ctx);
5936 return -1;
5938 if (options.machinereadable){
5939 printf(":%d:%s:\n", pnn_node->pnn, pnn_node->addr);
5940 } else {
5941 printf("%s\n", pnn_node->addr);
5944 talloc_free(mem_ctx);
5946 return 0;
5950 reload the nodes file on the local node
5952 static int control_reload_nodes_file(struct ctdb_context *ctdb, int argc, const char **argv)
5954 int i, ret;
5955 int mypnn;
5956 struct ctdb_node_map *nodemap=NULL;
5958 assert_single_node_only();
5960 mypnn = ctdb_get_pnn(ctdb);
5962 ret = ctdb_ctrl_getnodemap(ctdb, TIMELIMIT(), CTDB_CURRENT_NODE, ctdb, &nodemap);
5963 if (ret != 0) {
5964 DEBUG(DEBUG_ERR, ("Unable to get nodemap from local node\n"));
5965 return ret;
5968 /* reload the nodes file on all remote nodes */
5969 for (i=0;i<nodemap->num;i++) {
5970 if (nodemap->nodes[i].pnn == mypnn) {
5971 continue;
5973 DEBUG(DEBUG_NOTICE, ("Reloading nodes file on node %u\n", nodemap->nodes[i].pnn));
5974 ret = ctdb_ctrl_reload_nodes_file(ctdb, TIMELIMIT(),
5975 nodemap->nodes[i].pnn);
5976 if (ret != 0) {
5977 DEBUG(DEBUG_ERR, ("ERROR: Failed to reload nodes file on node %u. You MUST fix that node manually!\n", nodemap->nodes[i].pnn));
5981 /* reload the nodes file on the local node */
5982 DEBUG(DEBUG_NOTICE, ("Reloading nodes file on node %u\n", mypnn));
5983 ret = ctdb_ctrl_reload_nodes_file(ctdb, TIMELIMIT(), mypnn);
5984 if (ret != 0) {
5985 DEBUG(DEBUG_ERR, ("ERROR: Failed to reload nodes file on node %u. You MUST fix that node manually!\n", mypnn));
5988 /* initiate a recovery */
5989 control_recover(ctdb, argc, argv);
5991 return 0;
5995 static const struct {
5996 const char *name;
5997 int (*fn)(struct ctdb_context *, int, const char **);
5998 bool auto_all;
5999 bool without_daemon; /* can be run without daemon running ? */
6000 const char *msg;
6001 const char *args;
6002 } ctdb_commands[] = {
6003 { "version", control_version, true, true, "show version of ctdb" },
6004 { "status", control_status, true, false, "show node status" },
6005 { "uptime", control_uptime, true, false, "show node uptime" },
6006 { "ping", control_ping, true, false, "ping all nodes" },
6007 { "runstate", control_runstate, true, false, "get/check runstate of a node", "[setup|first_recovery|startup|running]" },
6008 { "getvar", control_getvar, true, false, "get a tunable variable", "<name>"},
6009 { "setvar", control_setvar, true, false, "set a tunable variable", "<name> <value>"},
6010 { "listvars", control_listvars, true, false, "list tunable variables"},
6011 { "statistics", control_statistics, false, false, "show statistics" },
6012 { "statisticsreset", control_statistics_reset, true, false, "reset statistics"},
6013 { "stats", control_stats, false, false, "show rolling statistics", "[number of history records]" },
6014 { "ip", control_ip, false, false, "show which public ip's that ctdb manages" },
6015 { "ipinfo", control_ipinfo, true, false, "show details about a public ip that ctdb manages", "<ip>" },
6016 { "ifaces", control_ifaces, true, false, "show which interfaces that ctdb manages" },
6017 { "setifacelink", control_setifacelink, true, false, "set interface link status", "<iface> <status>" },
6018 { "process-exists", control_process_exists, true, false, "check if a process exists on a node", "<pid>"},
6019 { "getdbmap", control_getdbmap, true, false, "show the database map" },
6020 { "getdbstatus", control_getdbstatus, true, false, "show the status of a database", "<dbname|dbid>" },
6021 { "catdb", control_catdb, true, false, "dump a ctdb database" , "<dbname|dbid>"},
6022 { "cattdb", control_cattdb, true, false, "dump a local tdb database" , "<dbname|dbid>"},
6023 { "getmonmode", control_getmonmode, true, false, "show monitoring mode" },
6024 { "getcapabilities", control_getcapabilities, true, false, "show node capabilities" },
6025 { "pnn", control_pnn, true, false, "show the pnn of the currnet node" },
6026 { "lvs", control_lvs, true, false, "show lvs configuration" },
6027 { "lvsmaster", control_lvsmaster, true, false, "show which node is the lvs master" },
6028 { "disablemonitor", control_disable_monmode,true, false, "set monitoring mode to DISABLE" },
6029 { "enablemonitor", control_enable_monmode, true, false, "set monitoring mode to ACTIVE" },
6030 { "setdebug", control_setdebug, true, false, "set debug level", "<EMERG|ALERT|CRIT|ERR|WARNING|NOTICE|INFO|DEBUG>" },
6031 { "getdebug", control_getdebug, true, false, "get debug level" },
6032 { "getlog", control_getlog, true, false, "get the log data from the in memory ringbuffer", "[<level>] [recoverd]" },
6033 { "clearlog", control_clearlog, true, false, "clear the log data from the in memory ringbuffer", "[recoverd]" },
6034 { "attach", control_attach, true, false, "attach to a database", "<dbname> [persistent]" },
6035 { "dumpmemory", control_dumpmemory, true, false, "dump memory map to stdout" },
6036 { "rddumpmemory", control_rddumpmemory, true, false, "dump memory map from the recovery daemon to stdout" },
6037 { "getpid", control_getpid, true, false, "get ctdbd process ID" },
6038 { "disable", control_disable, true, false, "disable a nodes public IP" },
6039 { "enable", control_enable, true, false, "enable a nodes public IP" },
6040 { "stop", control_stop, true, false, "stop a node" },
6041 { "continue", control_continue, true, false, "re-start a stopped node" },
6042 { "ban", control_ban, true, false, "ban a node from the cluster", "<bantime>"},
6043 { "unban", control_unban, true, false, "unban a node" },
6044 { "showban", control_showban, true, false, "show ban information"},
6045 { "shutdown", control_shutdown, true, false, "shutdown ctdbd" },
6046 { "recover", control_recover, true, false, "force recovery" },
6047 { "sync", control_ipreallocate, false, false, "wait until ctdbd has synced all state changes" },
6048 { "ipreallocate", control_ipreallocate, false, false, "force the recovery daemon to perform a ip reallocation procedure" },
6049 { "thaw", control_thaw, true, false, "thaw databases", "[priority:1-3]" },
6050 { "isnotrecmaster", control_isnotrecmaster, false, false, "check if the local node is recmaster or not" },
6051 { "killtcp", kill_tcp, false, false, "kill a tcp connection.", "[<srcip:port> <dstip:port>]" },
6052 { "gratiousarp", control_gratious_arp, false, false, "send a gratious arp", "<ip> <interface>" },
6053 { "tickle", tickle_tcp, false, false, "send a tcp tickle ack", "<srcip:port> <dstip:port>" },
6054 { "gettickles", control_get_tickles, false, false, "get the list of tickles registered for this ip", "<ip> [<port>]" },
6055 { "addtickle", control_add_tickle, false, false, "add a tickle for this ip", "<ip>:<port> <ip>:<port>" },
6057 { "deltickle", control_del_tickle, false, false, "delete a tickle from this ip", "<ip>:<port> <ip>:<port>" },
6059 { "regsrvid", regsrvid, false, false, "register a server id", "<pnn> <type> <id>" },
6060 { "unregsrvid", unregsrvid, false, false, "unregister a server id", "<pnn> <type> <id>" },
6061 { "chksrvid", chksrvid, false, false, "check if a server id exists", "<pnn> <type> <id>" },
6062 { "getsrvids", getsrvids, false, false, "get a list of all server ids"},
6063 { "check_srvids", check_srvids, false, false, "check if a srvid exists", "<id>+" },
6064 { "repack", ctdb_repack, false, false, "repack all databases", "[max_freelist]"},
6065 { "listnodes", control_listnodes, false, true, "list all nodes in the cluster"},
6066 { "reloadnodes", control_reload_nodes_file, false, false, "reload the nodes file and restart the transport on all nodes"},
6067 { "moveip", control_moveip, false, false, "move/failover an ip address to another node", "<ip> <node>"},
6068 { "rebalanceip", control_rebalanceip, false, false, "release an ip from the node and let recd rebalance it", "<ip>"},
6069 { "addip", control_addip, true, false, "add a ip address to a node", "<ip/mask> <iface>"},
6070 { "delip", control_delip, false, false, "delete an ip address from a node", "<ip>"},
6071 { "eventscript", control_eventscript, true, false, "run the eventscript with the given parameters on a node", "<arguments>"},
6072 { "backupdb", control_backupdb, false, false, "backup the database into a file.", "<dbname|dbid> <file>"},
6073 { "restoredb", control_restoredb, false, false, "restore the database from a file.", "<file> [dbname]"},
6074 { "dumpdbbackup", control_dumpdbbackup, false, true, "dump database backup from a file.", "<file>"},
6075 { "wipedb", control_wipedb, false, false, "wipe the contents of a database.", "<dbname|dbid>"},
6076 { "recmaster", control_recmaster, true, false, "show the pnn for the recovery master."},
6077 { "scriptstatus", control_scriptstatus, true, false, "show the status of the monitoring scripts (or all scripts)", "[all]"},
6078 { "enablescript", control_enablescript, true, false, "enable an eventscript", "<script>"},
6079 { "disablescript", control_disablescript, true, false, "disable an eventscript", "<script>"},
6080 { "natgwlist", control_natgwlist, true, false, "show the nodes belonging to this natgw configuration"},
6081 { "xpnn", control_xpnn, false, true, "find the pnn of the local node without talking to the daemon (unreliable)" },
6082 { "getreclock", control_getreclock, true, false, "Show the reclock file of a node"},
6083 { "setreclock", control_setreclock, true, false, "Set/clear the reclock file of a node", "[filename]"},
6084 { "setnatgwstate", control_setnatgwstate, false, false, "Set NATGW state to on/off", "{on|off}"},
6085 { "setlmasterrole", control_setlmasterrole, false, false, "Set LMASTER role to on/off", "{on|off}"},
6086 { "setrecmasterrole", control_setrecmasterrole, false, false, "Set RECMASTER role to on/off", "{on|off}"},
6087 { "setdbprio", control_setdbprio, false, false, "Set DB priority", "<dbname|dbid> <prio:1-3>"},
6088 { "getdbprio", control_getdbprio, false, false, "Get DB priority", "<dbname|dbid>"},
6089 { "setdbreadonly", control_setdbreadonly, false, false, "Set DB readonly capable", "<dbname|dbid>"},
6090 { "setdbsticky", control_setdbsticky, false, false, "Set DB sticky-records capable", "<dbname|dbid>"},
6091 { "msglisten", control_msglisten, false, false, "Listen on a srvid port for messages", "<msg srvid>"},
6092 { "msgsend", control_msgsend, false, false, "Send a message to srvid", "<srvid> <message>"},
6093 { "pfetch", control_pfetch, false, false, "fetch a record from a persistent database", "<dbname|dbid> <key> [<file>]" },
6094 { "pstore", control_pstore, false, false, "write a record to a persistent database", "<dbname|dbid> <key> <file containing record>" },
6095 { "pdelete", control_pdelete, false, false, "delete a record from a persistent database", "<dbname|dbid> <key>" },
6096 { "tfetch", control_tfetch, false, true, "fetch a record from a [c]tdb-file [-v]", "<tdb-file> <key> [<file>]" },
6097 { "tstore", control_tstore, false, true, "store a record (including ltdb header)", "<tdb-file> <key> <data+header>" },
6098 { "readkey", control_readkey, true, false, "read the content off a database key", "<tdb-file> <key>" },
6099 { "writekey", control_writekey, true, false, "write to a database key", "<tdb-file> <key> <value>" },
6100 { "checktcpport", control_chktcpport, false, true, "check if a service is bound to a specific tcp port or not", "<port>" },
6101 { "rebalancenode", control_rebalancenode, false, false, "mark nodes as forced IP rebalancing targets", "[<pnn-list>]"},
6102 { "getdbseqnum", control_getdbseqnum, false, false, "get the sequence number off a database", "<dbname|dbid>" },
6103 { "nodestatus", control_nodestatus, true, false, "show and return node status", "[<pnn-list>]" },
6104 { "dbstatistics", control_dbstatistics, false, false, "show db statistics", "<dbname|dbid>" },
6105 { "reloadips", control_reloadips, false, false, "reload the public addresses file on specified nodes" , "[<pnn-list>]" },
6106 { "ipiface", control_ipiface, false, true, "Find which interface an ip address is hosted on", "<ip>" },
6110 show usage message
6112 static void usage(void)
6114 int i;
6115 printf(
6116 "Usage: ctdb [options] <control>\n" \
6117 "Options:\n" \
6118 " -n <node> choose node number, or 'all' (defaults to local node)\n"
6119 " -Y generate machinereadable output\n"
6120 " -v generate verbose output\n"
6121 " -t <timelimit> set timelimit for control in seconds (default %u)\n", options.timelimit);
6122 printf("Controls:\n");
6123 for (i=0;i<ARRAY_SIZE(ctdb_commands);i++) {
6124 printf(" %-15s %-27s %s\n",
6125 ctdb_commands[i].name,
6126 ctdb_commands[i].args?ctdb_commands[i].args:"",
6127 ctdb_commands[i].msg);
6129 exit(1);
6133 static void ctdb_alarm(int sig)
6135 printf("Maximum runtime exceeded - exiting\n");
6136 _exit(ERR_TIMEOUT);
6140 main program
6142 int main(int argc, const char *argv[])
6144 struct ctdb_context *ctdb;
6145 char *nodestring = NULL;
6146 struct poptOption popt_options[] = {
6147 POPT_AUTOHELP
6148 POPT_CTDB_CMDLINE
6149 { "timelimit", 't', POPT_ARG_INT, &options.timelimit, 0, "timelimit", "integer" },
6150 { "node", 'n', POPT_ARG_STRING, &nodestring, 0, "node", "integer|all" },
6151 { "machinereadable", 'Y', POPT_ARG_NONE, &options.machinereadable, 0, "enable machinereadable output", NULL },
6152 { "verbose", 'v', POPT_ARG_NONE, &options.verbose, 0, "enable verbose output", NULL },
6153 { "maxruntime", 'T', POPT_ARG_INT, &options.maxruntime, 0, "die if runtime exceeds this limit (in seconds)", "integer" },
6154 { "print-emptyrecords", 0, POPT_ARG_NONE, &options.printemptyrecords, 0, "print the empty records when dumping databases (catdb, cattdb, dumpdbbackup)", NULL },
6155 { "print-datasize", 0, POPT_ARG_NONE, &options.printdatasize, 0, "do not print record data when dumping databases, only the data size", NULL },
6156 { "print-lmaster", 0, POPT_ARG_NONE, &options.printlmaster, 0, "print the record's lmaster in catdb", NULL },
6157 { "print-hash", 0, POPT_ARG_NONE, &options.printhash, 0, "print the record's hash when dumping databases", NULL },
6158 { "print-recordflags", 0, POPT_ARG_NONE, &options.printrecordflags, 0, "print the record flags in catdb and dumpdbbackup", NULL },
6159 POPT_TABLEEND
6161 int opt;
6162 const char **extra_argv;
6163 int extra_argc = 0;
6164 int ret=-1, i;
6165 poptContext pc;
6166 struct event_context *ev;
6167 const char *control;
6169 setlinebuf(stdout);
6171 /* set some defaults */
6172 options.maxruntime = 0;
6173 options.timelimit = 10;
6174 options.pnn = CTDB_CURRENT_NODE;
6176 pc = poptGetContext(argv[0], argc, argv, popt_options, POPT_CONTEXT_KEEP_FIRST);
6178 while ((opt = poptGetNextOpt(pc)) != -1) {
6179 switch (opt) {
6180 default:
6181 DEBUG(DEBUG_ERR, ("Invalid option %s: %s\n",
6182 poptBadOption(pc, 0), poptStrerror(opt)));
6183 exit(1);
6187 /* setup the remaining options for the main program to use */
6188 extra_argv = poptGetArgs(pc);
6189 if (extra_argv) {
6190 extra_argv++;
6191 while (extra_argv[extra_argc]) extra_argc++;
6194 if (extra_argc < 1) {
6195 usage();
6198 if (options.maxruntime == 0) {
6199 const char *ctdb_timeout;
6200 ctdb_timeout = getenv("CTDB_TIMEOUT");
6201 if (ctdb_timeout != NULL) {
6202 options.maxruntime = strtoul(ctdb_timeout, NULL, 0);
6203 } else {
6204 /* default timeout is 120 seconds */
6205 options.maxruntime = 120;
6209 signal(SIGALRM, ctdb_alarm);
6210 alarm(options.maxruntime);
6212 control = extra_argv[0];
6214 /* Default value for CTDB_BASE - don't override */
6215 setenv("CTDB_BASE", ETCDIR "/ctdb", 0);
6217 ev = event_context_init(NULL);
6218 if (!ev) {
6219 DEBUG(DEBUG_ERR, ("Failed to initialize event system\n"));
6220 exit(1);
6223 for (i=0;i<ARRAY_SIZE(ctdb_commands);i++) {
6224 if (strcmp(control, ctdb_commands[i].name) == 0) {
6225 break;
6229 if (i == ARRAY_SIZE(ctdb_commands)) {
6230 DEBUG(DEBUG_ERR, ("Unknown control '%s'\n", control));
6231 exit(1);
6234 if (ctdb_commands[i].without_daemon == true) {
6235 if (nodestring != NULL) {
6236 DEBUG(DEBUG_ERR, ("Can't specify node(s) with \"ctdb %s\"\n", control));
6237 exit(1);
6239 close(2);
6240 return ctdb_commands[i].fn(NULL, extra_argc-1, extra_argv+1);
6243 /* initialise ctdb */
6244 ctdb = ctdb_cmdline_client(ev, TIMELIMIT());
6246 if (ctdb == NULL) {
6247 DEBUG(DEBUG_ERR, ("Failed to init ctdb\n"));
6248 exit(1);
6251 /* setup the node number(s) to contact */
6252 if (!parse_nodestring(ctdb, ctdb, nodestring, CTDB_CURRENT_NODE, false,
6253 &options.nodes, &options.pnn)) {
6254 usage();
6257 if (options.pnn == CTDB_CURRENT_NODE) {
6258 options.pnn = options.nodes[0];
6261 if (ctdb_commands[i].auto_all &&
6262 ((options.pnn == CTDB_BROADCAST_ALL) ||
6263 (options.pnn == CTDB_MULTICAST))) {
6264 int j;
6266 ret = 0;
6267 for (j = 0; j < talloc_array_length(options.nodes); j++) {
6268 options.pnn = options.nodes[j];
6269 ret |= ctdb_commands[i].fn(ctdb, extra_argc-1, extra_argv+1);
6271 } else {
6272 ret = ctdb_commands[i].fn(ctdb, extra_argc-1, extra_argv+1);
6275 talloc_free(ctdb);
6276 talloc_free(ev);
6277 (void)poptFreeContext(pc);
6279 return ret;