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/>.
22 #include "system/time.h"
23 #include "system/filesys.h"
24 #include "system/network.h"
25 #include "system/locale.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 "lib/tdb_wrap/tdb_wrap.h"
33 #include "lib/util/dlinklist.h"
35 #define ERR_TIMEOUT 20 /* timed out trying to reach node */
36 #define ERR_NONODE 21 /* node does not exist */
37 #define ERR_DISNODE 22 /* node is disconnected */
39 static void usage(void);
46 const char *machineseparator;
49 int printemptyrecords;
56 #define LONGTIMEOUT options.timelimit*10
58 #define TIMELIMIT() timeval_current_ofs(options.timelimit, 0)
59 #define LONGTIMELIMIT() timeval_current_ofs(LONGTIMEOUT, 0)
61 static double timeval_delta(struct timeval *tv2, struct timeval *tv)
63 return (tv2->tv_sec - tv->tv_sec) +
64 (tv2->tv_usec - tv->tv_usec)*1.0e-6;
67 static int control_version(struct ctdb_context *ctdb, int argc, const char **argv)
69 printf("CTDB version: %s\n", CTDB_VERSION_STRING);
73 /* Like printf(3) but substitute for separator in format */
74 static int printm(const char *format, ...) PRINTF_ATTRIBUTE(1,2);
75 static int printm(const char *format, ...)
79 size_t len = strlen(format);
80 char new_format[len+1];
82 strcpy(new_format, format);
84 if (options.machineseparator[0] != ':') {
85 all_string_sub(new_format,
86 ":", options.machineseparator, len + 1);
90 ret = vprintf(new_format, ap);
96 #define CTDB_NOMEM_ABORT(p) do { if (!(p)) { \
97 DEBUG(DEBUG_ALERT,("ctdb fatal error: %s\n", \
98 "Out of memory in " __location__ )); \
102 static uint32_t getpnn(struct ctdb_context *ctdb)
104 if ((options.pnn == CTDB_BROADCAST_ALL) ||
105 (options.pnn == CTDB_MULTICAST)) {
107 ("Cannot get PNN for node %u\n", options.pnn));
111 if (options.pnn == CTDB_CURRENT_NODE) {
112 return ctdb_get_pnn(ctdb);
118 static void assert_single_node_only(void)
120 if ((options.pnn == CTDB_BROADCAST_ALL) ||
121 (options.pnn == CTDB_MULTICAST)) {
123 ("This control can not be applied to multiple PNNs\n"));
128 static void assert_current_node_only(struct ctdb_context *ctdb)
130 if (options.pnn != ctdb_get_pnn(ctdb)) {
132 ("This control can only be applied to the current node\n"));
137 /* Pretty print the flags to a static buffer in human-readable format.
138 * This never returns NULL!
140 static const char *pretty_print_flags(uint32_t flags)
143 static const struct {
147 { NODE_FLAGS_DISCONNECTED, "DISCONNECTED" },
148 { NODE_FLAGS_PERMANENTLY_DISABLED, "DISABLED" },
149 { NODE_FLAGS_BANNED, "BANNED" },
150 { NODE_FLAGS_UNHEALTHY, "UNHEALTHY" },
151 { NODE_FLAGS_DELETED, "DELETED" },
152 { NODE_FLAGS_STOPPED, "STOPPED" },
153 { NODE_FLAGS_INACTIVE, "INACTIVE" },
155 static char flags_str[512]; /* Big enough to contain all flag names */
158 for (j=0;j<ARRAY_SIZE(flag_names);j++) {
159 if (flags & flag_names[j].flag) {
160 if (flags_str[0] == '\0') {
161 (void) strcpy(flags_str, flag_names[j].name);
163 (void) strncat(flags_str, "|", sizeof(flags_str)-1);
164 (void) strncat(flags_str, flag_names[j].name,
165 sizeof(flags_str)-1);
169 if (flags_str[0] == '\0') {
170 (void) strcpy(flags_str, "OK");
176 static int h2i(char h)
178 if (h >= 'a' && h <= 'f') return h - 'a' + 10;
179 if (h >= 'A' && h <= 'F') return h - 'f' + 10;
183 static TDB_DATA hextodata(TALLOC_CTX *mem_ctx, const char *str)
186 TDB_DATA key = {NULL, 0};
190 DEBUG(DEBUG_ERR,("Key specified with odd number of hexadecimal digits\n"));
195 key.dptr = talloc_size(mem_ctx, key.dsize);
197 for (i=0; i < len/2; i++) {
198 key.dptr[i] = h2i(str[i*2]) << 4 | h2i(str[i*2+1]);
203 /* Parse a nodestring. Parameter dd_ok controls what happens to nodes
204 * that are disconnected or deleted. If dd_ok is true those nodes are
205 * included in the output list of nodes. If dd_ok is false, those
206 * nodes are filtered from the "all" case and cause an error if
207 * explicitly specified.
209 static bool parse_nodestring(struct ctdb_context *ctdb,
211 const char * nodestring,
212 uint32_t current_pnn,
217 TALLOC_CTX *tmp_ctx = talloc_new(mem_ctx);
220 struct ctdb_node_map *nodemap;
225 ret = ctdb_ctrl_getnodemap(ctdb, TIMELIMIT(), CTDB_CURRENT_NODE, tmp_ctx, &nodemap);
227 DEBUG(DEBUG_ERR, ("Unable to get nodemap from local node\n"));
228 talloc_free(tmp_ctx);
232 if (nodestring != NULL) {
233 *nodes = talloc_array(mem_ctx, uint32_t, 0);
234 if (*nodes == NULL) {
240 if (strcmp(nodestring, "all") == 0) {
241 *pnn_mode = CTDB_BROADCAST_ALL;
244 for (i = 0; i < nodemap->num; i++) {
245 if ((nodemap->nodes[i].flags &
246 (NODE_FLAGS_DISCONNECTED |
247 NODE_FLAGS_DELETED)) && !dd_ok) {
250 *nodes = talloc_realloc(mem_ctx, *nodes,
252 if (*nodes == NULL) {
262 ns = talloc_strdup(tmp_ctx, nodestring);
263 tok = strtok(ns, ",");
264 while (tok != NULL) {
267 i = (uint32_t)strtoul(tok, &endptr, 0);
268 if (i == 0 && tok == endptr) {
270 ("Invalid node %s\n", tok));
271 talloc_free(tmp_ctx);
274 if (i >= nodemap->num) {
275 DEBUG(DEBUG_ERR, ("Node %u does not exist\n", i));
276 talloc_free(tmp_ctx);
279 if ((nodemap->nodes[i].flags &
280 (NODE_FLAGS_DISCONNECTED |
281 NODE_FLAGS_DELETED)) && !dd_ok) {
282 DEBUG(DEBUG_ERR, ("Node %u has status %s\n", i, pretty_print_flags(nodemap->nodes[i].flags)));
283 talloc_free(tmp_ctx);
286 if ((pnn = ctdb_ctrl_getpnn(ctdb, TIMELIMIT(), i)) < 0) {
287 DEBUG(DEBUG_ERR, ("Can not access node %u. Node is not operational.\n", i));
288 talloc_free(tmp_ctx);
292 *nodes = talloc_realloc(mem_ctx, *nodes,
294 if (*nodes == NULL) {
301 tok = strtok(NULL, ",");
306 *pnn_mode = (*nodes)[0];
308 *pnn_mode = CTDB_MULTICAST;
312 /* default - no nodes specified */
313 *nodes = talloc_array(mem_ctx, uint32_t, 1);
314 if (*nodes == NULL) {
317 *pnn_mode = CTDB_CURRENT_NODE;
319 if (((*nodes)[0] = ctdb_ctrl_getpnn(ctdb, TIMELIMIT(), current_pnn)) < 0) {
324 talloc_free(tmp_ctx);
328 talloc_free(tmp_ctx);
333 check if a database exists
335 static bool db_exists(struct ctdb_context *ctdb, const char *dbarg,
336 uint32_t *dbid, const char **dbname, uint8_t *flags)
339 struct ctdb_dbid_map *dbmap=NULL;
340 bool dbid_given = false, found = false;
342 TALLOC_CTX *tmp_ctx = talloc_new(ctdb);
343 const char *name = NULL;
345 ret = ctdb_ctrl_getdbmap(ctdb, TIMELIMIT(), options.pnn, tmp_ctx, &dbmap);
347 DEBUG(DEBUG_ERR, ("Unable to get dbids from node %u\n", options.pnn));
351 if (strncmp(dbarg, "0x", 2) == 0) {
352 id = strtoul(dbarg, NULL, 0);
356 for(i=0; i<dbmap->num; i++) {
358 if (id == dbmap->dbs[i].dbid) {
363 ret = ctdb_ctrl_getdbname(ctdb, TIMELIMIT(), options.pnn, dbmap->dbs[i].dbid, tmp_ctx, &name);
365 DEBUG(DEBUG_ERR, ("Unable to get dbname from dbid %u\n", dbmap->dbs[i].dbid));
369 if (strcmp(name, dbarg) == 0) {
370 id = dbmap->dbs[i].dbid;
377 if (found && dbid_given && dbname != NULL) {
378 ret = ctdb_ctrl_getdbname(ctdb, TIMELIMIT(), options.pnn, dbmap->dbs[i].dbid, tmp_ctx, &name);
380 DEBUG(DEBUG_ERR, ("Unable to get dbname from dbid %u\n", dbmap->dbs[i].dbid));
387 if (dbid) *dbid = id;
388 if (dbname) *dbname = talloc_strdup(ctdb, name);
389 if (flags) *flags = dbmap->dbs[i].flags;
391 DEBUG(DEBUG_ERR,("No database matching '%s' found\n", dbarg));
395 talloc_free(tmp_ctx);
400 see if a process exists
402 static int control_process_exists(struct ctdb_context *ctdb, int argc, const char **argv)
410 if (sscanf(argv[0], "%u:%u", &pnn, &pid) != 2) {
411 DEBUG(DEBUG_ERR, ("Badly formed pnn:pid\n"));
415 ret = ctdb_ctrl_process_exists(ctdb, pnn, pid);
417 printf("%u:%u exists\n", pnn, pid);
419 printf("%u:%u does not exist\n", pnn, pid);
425 display statistics structure
427 static void show_statistics(struct ctdb_statistics *s, int show_header)
429 TALLOC_CTX *tmp_ctx = talloc_new(NULL);
431 const char *prefix=NULL;
433 int tmp, days, hours, minutes, seconds;
438 #define STATISTICS_FIELD(n) { #n, offsetof(struct ctdb_statistics, n) }
439 STATISTICS_FIELD(num_clients),
440 STATISTICS_FIELD(frozen),
441 STATISTICS_FIELD(recovering),
442 STATISTICS_FIELD(num_recoveries),
443 STATISTICS_FIELD(client_packets_sent),
444 STATISTICS_FIELD(client_packets_recv),
445 STATISTICS_FIELD(node_packets_sent),
446 STATISTICS_FIELD(node_packets_recv),
447 STATISTICS_FIELD(keepalive_packets_sent),
448 STATISTICS_FIELD(keepalive_packets_recv),
449 STATISTICS_FIELD(node.req_call),
450 STATISTICS_FIELD(node.reply_call),
451 STATISTICS_FIELD(node.req_dmaster),
452 STATISTICS_FIELD(node.reply_dmaster),
453 STATISTICS_FIELD(node.reply_error),
454 STATISTICS_FIELD(node.req_message),
455 STATISTICS_FIELD(node.req_control),
456 STATISTICS_FIELD(node.reply_control),
457 STATISTICS_FIELD(client.req_call),
458 STATISTICS_FIELD(client.req_message),
459 STATISTICS_FIELD(client.req_control),
460 STATISTICS_FIELD(timeouts.call),
461 STATISTICS_FIELD(timeouts.control),
462 STATISTICS_FIELD(timeouts.traverse),
463 STATISTICS_FIELD(locks.num_calls),
464 STATISTICS_FIELD(locks.num_current),
465 STATISTICS_FIELD(locks.num_pending),
466 STATISTICS_FIELD(locks.num_failed),
467 STATISTICS_FIELD(total_calls),
468 STATISTICS_FIELD(pending_calls),
469 STATISTICS_FIELD(childwrite_calls),
470 STATISTICS_FIELD(pending_childwrite_calls),
471 STATISTICS_FIELD(memory_used),
472 STATISTICS_FIELD(max_hop_count),
473 STATISTICS_FIELD(total_ro_delegations),
474 STATISTICS_FIELD(total_ro_revokes),
477 tmp = s->statistics_current_time.tv_sec - s->statistics_start_time.tv_sec;
486 if (options.machinereadable){
488 printm("CTDB version:");
489 printm("Current time of statistics:");
490 printm("Statistics collected since:");
491 for (i=0;i<ARRAY_SIZE(fields);i++) {
492 printm("%s:", fields[i].name);
494 printm("num_reclock_ctdbd_latency:");
495 printm("min_reclock_ctdbd_latency:");
496 printm("avg_reclock_ctdbd_latency:");
497 printm("max_reclock_ctdbd_latency:");
499 printm("num_reclock_recd_latency:");
500 printm("min_reclock_recd_latency:");
501 printm("avg_reclock_recd_latency:");
502 printm("max_reclock_recd_latency:");
504 printm("num_call_latency:");
505 printm("min_call_latency:");
506 printm("avg_call_latency:");
507 printm("max_call_latency:");
509 printm("num_lockwait_latency:");
510 printm("min_lockwait_latency:");
511 printm("avg_lockwait_latency:");
512 printm("max_lockwait_latency:");
514 printm("num_childwrite_latency:");
515 printm("min_childwrite_latency:");
516 printm("avg_childwrite_latency:");
517 printm("max_childwrite_latency:");
520 printm("%d:", CTDB_PROTOCOL);
521 printm("%d:", (int)s->statistics_current_time.tv_sec);
522 printm("%d:", (int)s->statistics_start_time.tv_sec);
523 for (i=0;i<ARRAY_SIZE(fields);i++) {
524 printm("%d:", *(uint32_t *)(fields[i].offset+(uint8_t *)s));
526 printm("%d:", s->reclock.ctdbd.num);
527 printm("%.6f:", s->reclock.ctdbd.min);
528 printm("%.6f:", s->reclock.ctdbd.num?s->reclock.ctdbd.total/s->reclock.ctdbd.num:0.0);
529 printm("%.6f:", s->reclock.ctdbd.max);
531 printm("%d:", s->reclock.recd.num);
532 printm("%.6f:", s->reclock.recd.min);
533 printm("%.6f:", s->reclock.recd.num?s->reclock.recd.total/s->reclock.recd.num:0.0);
534 printm("%.6f:", s->reclock.recd.max);
536 printm("%d:", s->call_latency.num);
537 printm("%.6f:", s->call_latency.min);
538 printm("%.6f:", s->call_latency.num?s->call_latency.total/s->call_latency.num:0.0);
539 printm("%.6f:", s->call_latency.max);
541 printm("%d:", s->childwrite_latency.num);
542 printm("%.6f:", s->childwrite_latency.min);
543 printm("%.6f:", s->childwrite_latency.num?s->childwrite_latency.total/s->childwrite_latency.num:0.0);
544 printm("%.6f:", s->childwrite_latency.max);
547 printf("CTDB version %u\n", CTDB_PROTOCOL);
548 printf("Current time of statistics : %s", ctime(&s->statistics_current_time.tv_sec));
549 printf("Statistics collected since : (%03d %02d:%02d:%02d) %s", days, hours, minutes, seconds, ctime(&s->statistics_start_time.tv_sec));
551 for (i=0;i<ARRAY_SIZE(fields);i++) {
552 if (strchr(fields[i].name, '.')) {
553 preflen = strcspn(fields[i].name, ".")+1;
554 if (!prefix || strncmp(prefix, fields[i].name, preflen) != 0) {
555 prefix = fields[i].name;
556 printf(" %*.*s\n", preflen-1, preflen-1, fields[i].name);
561 printf(" %*s%-22s%*s%10u\n",
563 fields[i].name+preflen,
565 *(uint32_t *)(fields[i].offset+(uint8_t *)s));
567 printf(" hop_count_buckets:");
568 for (i=0;i<MAX_COUNT_BUCKETS;i++) {
569 printf(" %d", s->hop_count_bucket[i]);
572 printf(" lock_buckets:");
573 for (i=0; i<MAX_COUNT_BUCKETS; i++) {
574 printf(" %d", s->locks.buckets[i]);
577 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);
579 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);
581 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);
583 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);
584 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);
587 talloc_free(tmp_ctx);
591 display remote ctdb statistics combined from all nodes
593 static int control_statistics_all(struct ctdb_context *ctdb)
596 struct ctdb_statistics statistics;
600 nodes = ctdb_get_connected_nodes(ctdb, TIMELIMIT(), ctdb, &num_nodes);
601 CTDB_NO_MEMORY(ctdb, nodes);
603 ZERO_STRUCT(statistics);
605 for (i=0;i<num_nodes;i++) {
606 struct ctdb_statistics s1;
608 uint32_t *v1 = (uint32_t *)&s1;
609 uint32_t *v2 = (uint32_t *)&statistics;
611 offsetof(struct ctdb_statistics, __last_counter) / sizeof(uint32_t);
612 ret = ctdb_ctrl_statistics(ctdb, nodes[i], &s1);
614 DEBUG(DEBUG_ERR, ("Unable to get statistics from node %u\n", nodes[i]));
617 for (j=0;j<num_ints;j++) {
620 statistics.max_hop_count =
621 MAX(statistics.max_hop_count, s1.max_hop_count);
622 statistics.call_latency.max =
623 MAX(statistics.call_latency.max, s1.call_latency.max);
626 printf("Gathered statistics for %u nodes\n", num_nodes);
627 show_statistics(&statistics, 1);
632 display remote ctdb statistics
634 static int control_statistics(struct ctdb_context *ctdb, int argc, const char **argv)
637 struct ctdb_statistics statistics;
639 if (options.pnn == CTDB_BROADCAST_ALL) {
640 return control_statistics_all(ctdb);
643 ret = ctdb_ctrl_statistics(ctdb, options.pnn, &statistics);
645 DEBUG(DEBUG_ERR, ("Unable to get statistics from node %u\n", options.pnn));
648 show_statistics(&statistics, 1);
654 reset remote ctdb statistics
656 static int control_statistics_reset(struct ctdb_context *ctdb, int argc, const char **argv)
660 ret = ctdb_statistics_reset(ctdb, options.pnn);
662 DEBUG(DEBUG_ERR, ("Unable to reset statistics on node %u\n", options.pnn));
670 display remote ctdb rolling statistics
672 static int control_stats(struct ctdb_context *ctdb, int argc, const char **argv)
675 struct ctdb_statistics_wire *stats;
676 int i, num_records = -1;
678 assert_single_node_only();
681 num_records = atoi(argv[0]) - 1;
684 ret = ctdb_ctrl_getstathistory(ctdb, TIMELIMIT(), options.pnn, ctdb, &stats);
686 DEBUG(DEBUG_ERR, ("Unable to get rolling statistics from node %u\n", options.pnn));
689 for (i=0;i<stats->num;i++) {
690 if (stats->stats[i].statistics_start_time.tv_sec == 0) {
693 show_statistics(&stats->stats[i], i==0);
694 if (i == num_records) {
703 display remote ctdb db statistics
705 static int control_dbstatistics(struct ctdb_context *ctdb, int argc, const char **argv)
707 TALLOC_CTX *tmp_ctx = talloc_new(ctdb);
708 struct ctdb_db_statistics *dbstat;
718 if (!db_exists(ctdb, argv[0], &db_id, NULL, NULL)) {
722 ret = ctdb_ctrl_dbstatistics(ctdb, options.pnn, db_id, tmp_ctx, &dbstat);
724 DEBUG(DEBUG_ERR,("Failed to read db statistics from node\n"));
725 talloc_free(tmp_ctx);
729 printf("DB Statistics: %s\n", argv[0]);
730 printf(" %*s%-22s%*s%10u\n", 0, "", "ro_delegations", 4, "",
731 dbstat->db_ro_delegations);
732 printf(" %*s%-22s%*s%10u\n", 0, "", "ro_revokes", 4, "",
733 dbstat->db_ro_delegations);
734 printf(" %s\n", "locks");
735 printf(" %*s%-22s%*s%10u\n", 4, "", "total", 0, "",
736 dbstat->locks.num_calls);
737 printf(" %*s%-22s%*s%10u\n", 4, "", "failed", 0, "",
738 dbstat->locks.num_failed);
739 printf(" %*s%-22s%*s%10u\n", 4, "", "current", 0, "",
740 dbstat->locks.num_current);
741 printf(" %*s%-22s%*s%10u\n", 4, "", "pending", 0, "",
742 dbstat->locks.num_pending);
743 printf(" %s", "hop_count_buckets:");
744 for (i=0; i<MAX_COUNT_BUCKETS; i++) {
745 printf(" %d", dbstat->hop_count_bucket[i]);
748 printf(" %s", "lock_buckets:");
749 for (i=0; i<MAX_COUNT_BUCKETS; i++) {
750 printf(" %d", dbstat->locks.buckets[i]);
753 printf(" %-30s %.6f/%.6f/%.6f sec out of %d\n",
754 "locks_latency MIN/AVG/MAX",
755 dbstat->locks.latency.min,
756 (dbstat->locks.latency.num ?
757 dbstat->locks.latency.total /dbstat->locks.latency.num :
759 dbstat->locks.latency.max,
760 dbstat->locks.latency.num);
761 printf(" %-30s %.6f/%.6f/%.6f sec out of %d\n",
762 "vacuum_latency MIN/AVG/MAX",
763 dbstat->vacuum.latency.min,
764 (dbstat->vacuum.latency.num ?
765 dbstat->vacuum.latency.total /dbstat->vacuum.latency.num :
767 dbstat->vacuum.latency.max,
768 dbstat->vacuum.latency.num);
770 for (i=0; i<dbstat->num_hot_keys; i++) {
771 if (dbstat->hot_keys[i].count > 0) {
775 dbstat->num_hot_keys = num_hot_keys;
777 printf(" Num Hot Keys: %d\n", dbstat->num_hot_keys);
778 for (i = 0; i < dbstat->num_hot_keys; i++) {
780 printf(" Count:%d Key:", dbstat->hot_keys[i].count);
781 for (j = 0; j < dbstat->hot_keys[i].key.dsize; j++) {
782 printf("%02x", dbstat->hot_keys[i].key.dptr[j]&0xff);
787 talloc_free(tmp_ctx);
792 display uptime of remote node
794 static int control_uptime(struct ctdb_context *ctdb, int argc, const char **argv)
797 struct ctdb_uptime *uptime = NULL;
798 int tmp, days, hours, minutes, seconds;
800 ret = ctdb_ctrl_uptime(ctdb, ctdb, TIMELIMIT(), options.pnn, &uptime);
802 DEBUG(DEBUG_ERR, ("Unable to get uptime from node %u\n", options.pnn));
806 if (options.machinereadable){
807 printm(":Current Node Time:Ctdb Start Time:Last Recovery/Failover Time:Last Recovery/IPFailover Duration:\n");
808 printm(":%u:%u:%u:%lf\n",
809 (unsigned int)uptime->current_time.tv_sec,
810 (unsigned int)uptime->ctdbd_start_time.tv_sec,
811 (unsigned int)uptime->last_recovery_finished.tv_sec,
812 timeval_delta(&uptime->last_recovery_finished,
813 &uptime->last_recovery_started)
818 printf("Current time of node : %s", ctime(&uptime->current_time.tv_sec));
820 tmp = uptime->current_time.tv_sec - uptime->ctdbd_start_time.tv_sec;
828 printf("Ctdbd start time : (%03d %02d:%02d:%02d) %s", days, hours, minutes, seconds, ctime(&uptime->ctdbd_start_time.tv_sec));
830 tmp = uptime->current_time.tv_sec - uptime->last_recovery_finished.tv_sec;
838 printf("Time of last recovery/failover: (%03d %02d:%02d:%02d) %s", days, hours, minutes, seconds, ctime(&uptime->last_recovery_finished.tv_sec));
840 printf("Duration of last recovery/failover: %lf seconds\n",
841 timeval_delta(&uptime->last_recovery_finished,
842 &uptime->last_recovery_started));
848 show the PNN of the current node
850 static int control_pnn(struct ctdb_context *ctdb, int argc, const char **argv)
854 mypnn = getpnn(ctdb);
856 printf("PNN:%d\n", mypnn);
861 static struct ctdb_node_map *read_nodes_file(TALLOC_CTX *mem_ctx)
863 const char *nodes_list;
865 /* read the nodes file */
866 nodes_list = getenv("CTDB_NODES");
867 if (nodes_list == NULL) {
868 nodes_list = talloc_asprintf(mem_ctx, "%s/nodes",
869 getenv("CTDB_BASE"));
870 if (nodes_list == NULL) {
871 DEBUG(DEBUG_ALERT,(__location__ " Out of memory\n"));
876 return ctdb_read_nodes_file(mem_ctx, nodes_list);
880 show the PNN of the current node
881 discover the pnn by loading the nodes file and try to bind to all
882 addresses one at a time until the ip address is found.
884 static int find_node_xpnn(void)
886 TALLOC_CTX *mem_ctx = talloc_new(NULL);
887 struct ctdb_node_map *node_map;
890 node_map = read_nodes_file(mem_ctx);
891 if (node_map == NULL) {
892 talloc_free(mem_ctx);
896 for (i = 0; i < node_map->num; i++) {
897 if (node_map->nodes[i].flags & NODE_FLAGS_DELETED) {
900 if (ctdb_sys_have_ip(&node_map->nodes[i].addr)) {
901 pnn = node_map->nodes[i].pnn;
902 talloc_free(mem_ctx);
907 printf("Failed to detect which PNN this node is\n");
908 talloc_free(mem_ctx);
912 static int control_xpnn(struct ctdb_context *ctdb, int argc, const char **argv)
916 assert_single_node_only();
918 pnn = find_node_xpnn();
923 printf("PNN:%d\n", pnn);
927 /* Helpers for ctdb status
929 static bool is_partially_online(struct ctdb_context *ctdb, struct ctdb_node_and_flags *node)
931 TALLOC_CTX *tmp_ctx = talloc_new(NULL);
935 if (node->flags == 0) {
936 struct ctdb_control_get_ifaces *ifaces;
938 if (ctdb_ctrl_get_ifaces(ctdb, TIMELIMIT(), node->pnn,
939 tmp_ctx, &ifaces) == 0) {
940 for (j=0; j < ifaces->num; j++) {
941 if (ifaces->ifaces[j].link_state != 0) {
949 talloc_free(tmp_ctx);
954 static void control_status_header_machine(void)
956 printm(":Node:IP:Disconnected:Banned:Disabled:Unhealthy:Stopped"
957 ":Inactive:PartiallyOnline:ThisNode:\n");
960 static int control_status_1_machine(struct ctdb_context *ctdb, int mypnn,
961 struct ctdb_node_and_flags *node)
963 printm(":%d:%s:%d:%d:%d:%d:%d:%d:%d:%c:\n", node->pnn,
964 ctdb_addr_to_str(&node->addr),
965 !!(node->flags&NODE_FLAGS_DISCONNECTED),
966 !!(node->flags&NODE_FLAGS_BANNED),
967 !!(node->flags&NODE_FLAGS_PERMANENTLY_DISABLED),
968 !!(node->flags&NODE_FLAGS_UNHEALTHY),
969 !!(node->flags&NODE_FLAGS_STOPPED),
970 !!(node->flags&NODE_FLAGS_INACTIVE),
971 is_partially_online(ctdb, node) ? 1 : 0,
972 (node->pnn == mypnn)?'Y':'N');
977 static int control_status_1_human(struct ctdb_context *ctdb, int mypnn,
978 struct ctdb_node_and_flags *node)
980 printf("pnn:%d %-16s %s%s\n", node->pnn,
981 ctdb_addr_to_str(&node->addr),
982 is_partially_online(ctdb, node) ? "PARTIALLYONLINE" : pretty_print_flags(node->flags),
983 node->pnn == mypnn?" (THIS NODE)":"");
989 display remote ctdb status
991 static int control_status(struct ctdb_context *ctdb, int argc, const char **argv)
993 TALLOC_CTX *tmp_ctx = talloc_new(ctdb);
995 struct ctdb_vnn_map *vnnmap=NULL;
996 struct ctdb_node_map *nodemap=NULL;
997 uint32_t recmode, recmaster, mypnn;
998 int num_deleted_nodes = 0;
1001 mypnn = getpnn(ctdb);
1003 ret = ctdb_ctrl_getnodemap(ctdb, TIMELIMIT(), options.pnn, tmp_ctx, &nodemap);
1005 DEBUG(DEBUG_ERR, ("Unable to get nodemap from node %u\n", options.pnn));
1006 talloc_free(tmp_ctx);
1010 if (options.machinereadable) {
1011 control_status_header_machine();
1012 for (i=0;i<nodemap->num;i++) {
1013 if (nodemap->nodes[i].flags & NODE_FLAGS_DELETED) {
1016 (void) control_status_1_machine(ctdb, mypnn,
1017 &nodemap->nodes[i]);
1019 talloc_free(tmp_ctx);
1023 for (i=0; i<nodemap->num; i++) {
1024 if (nodemap->nodes[i].flags & NODE_FLAGS_DELETED) {
1025 num_deleted_nodes++;
1028 if (num_deleted_nodes == 0) {
1029 printf("Number of nodes:%d\n", nodemap->num);
1031 printf("Number of nodes:%d (including %d deleted nodes)\n",
1032 nodemap->num, num_deleted_nodes);
1034 for(i=0;i<nodemap->num;i++){
1035 if (nodemap->nodes[i].flags & NODE_FLAGS_DELETED) {
1038 (void) control_status_1_human(ctdb, mypnn, &nodemap->nodes[i]);
1041 ret = ctdb_ctrl_getvnnmap(ctdb, TIMELIMIT(), options.pnn, tmp_ctx, &vnnmap);
1043 DEBUG(DEBUG_ERR, ("Unable to get vnnmap from node %u\n", options.pnn));
1044 talloc_free(tmp_ctx);
1047 if (vnnmap->generation == INVALID_GENERATION) {
1048 printf("Generation:INVALID\n");
1050 printf("Generation:%d\n",vnnmap->generation);
1052 printf("Size:%d\n",vnnmap->size);
1053 for(i=0;i<vnnmap->size;i++){
1054 printf("hash:%d lmaster:%d\n", i, vnnmap->map[i]);
1057 ret = ctdb_ctrl_getrecmode(ctdb, tmp_ctx, TIMELIMIT(), options.pnn, &recmode);
1059 DEBUG(DEBUG_ERR, ("Unable to get recmode from node %u\n", options.pnn));
1060 talloc_free(tmp_ctx);
1063 printf("Recovery mode:%s (%d)\n",recmode==CTDB_RECOVERY_NORMAL?"NORMAL":"RECOVERY",recmode);
1065 ret = ctdb_ctrl_getrecmaster(ctdb, tmp_ctx, TIMELIMIT(), options.pnn, &recmaster);
1067 DEBUG(DEBUG_ERR, ("Unable to get recmaster from node %u\n", options.pnn));
1068 talloc_free(tmp_ctx);
1071 printf("Recovery master:%d\n",recmaster);
1073 talloc_free(tmp_ctx);
1077 static int control_nodestatus(struct ctdb_context *ctdb, int argc, const char **argv)
1079 TALLOC_CTX *tmp_ctx = talloc_new(ctdb);
1081 struct ctdb_node_map *nodemap=NULL;
1083 uint32_t pnn_mode, mypnn;
1089 if (!parse_nodestring(ctdb, tmp_ctx, argc == 1 ? argv[0] : NULL,
1090 options.pnn, true, &nodes, &pnn_mode)) {
1094 if (options.machinereadable) {
1095 control_status_header_machine();
1096 } else if (pnn_mode == CTDB_BROADCAST_ALL) {
1097 printf("Number of nodes:%d\n", (int) talloc_array_length(nodes));
1100 mypnn = getpnn(ctdb);
1102 ret = ctdb_ctrl_getnodemap(ctdb, TIMELIMIT(), options.pnn, tmp_ctx, &nodemap);
1104 DEBUG(DEBUG_ERR, ("Unable to get nodemap from node %u\n", options.pnn));
1105 talloc_free(tmp_ctx);
1111 for (i = 0; i < talloc_array_length(nodes); i++) {
1112 if (options.machinereadable) {
1113 ret |= control_status_1_machine(ctdb, mypnn,
1114 &nodemap->nodes[nodes[i]]);
1116 ret |= control_status_1_human(ctdb, mypnn,
1117 &nodemap->nodes[nodes[i]]);
1121 talloc_free(tmp_ctx);
1125 static struct ctdb_node_map *read_natgw_nodes_file(struct ctdb_context *ctdb,
1126 TALLOC_CTX *mem_ctx)
1128 const char *natgw_list;
1129 struct ctdb_node_map *natgw_nodes = NULL;
1131 natgw_list = getenv("CTDB_NATGW_NODES");
1132 if (natgw_list == NULL) {
1133 natgw_list = talloc_asprintf(mem_ctx, "%s/natgw_nodes",
1134 getenv("CTDB_BASE"));
1135 if (natgw_list == NULL) {
1136 DEBUG(DEBUG_ALERT,(__location__ " Out of memory\n"));
1140 /* The PNNs/flags will be junk but they're not used */
1141 natgw_nodes = ctdb_read_nodes_file(mem_ctx, natgw_list);
1142 if (natgw_nodes == NULL) {
1144 ("Failed to load natgw node list '%s'\n", natgw_list));
1150 /* talloc off the existing nodemap... */
1151 static struct ctdb_node_map *talloc_nodemap(struct ctdb_node_map *nodemap)
1153 return talloc_zero_size(nodemap,
1154 offsetof(struct ctdb_node_map, nodes) +
1155 nodemap->num * sizeof(struct ctdb_node_and_flags));
1158 static struct ctdb_node_map *
1159 filter_nodemap_by_addrs(struct ctdb_context *ctdb,
1160 struct ctdb_node_map *nodemap,
1161 struct ctdb_node_map *natgw_nodes)
1164 struct ctdb_node_map *ret;
1166 ret = talloc_nodemap(nodemap);
1167 CTDB_NO_MEMORY_NULL(ctdb, ret);
1171 for (i = 0; i < nodemap->num; i++) {
1172 for(j = 0; j < natgw_nodes->num ; j++) {
1173 if (nodemap->nodes[j].flags & NODE_FLAGS_DELETED) {
1176 if (ctdb_same_ip(&natgw_nodes->nodes[j].addr,
1177 &nodemap->nodes[i].addr)) {
1179 ret->nodes[ret->num] = nodemap->nodes[i];
1189 static struct ctdb_node_map *
1190 filter_nodemap_by_capabilities(struct ctdb_context *ctdb,
1191 struct ctdb_node_map *nodemap,
1192 uint32_t required_capabilities,
1196 uint32_t capabilities;
1197 struct ctdb_node_map *ret;
1199 ret = talloc_nodemap(nodemap);
1200 CTDB_NO_MEMORY_NULL(ctdb, ret);
1204 for (i = 0; i < nodemap->num; i++) {
1207 /* Disconnected nodes have no capabilities! */
1208 if (nodemap->nodes[i].flags & NODE_FLAGS_DISCONNECTED) {
1212 res = ctdb_ctrl_getcapabilities(ctdb, TIMELIMIT(),
1213 nodemap->nodes[i].pnn,
1216 DEBUG(DEBUG_ERR, ("Unable to get capabilities from node %u\n",
1217 nodemap->nodes[i].pnn));
1221 if (!(capabilities & required_capabilities)) {
1225 ret->nodes[ret->num] = nodemap->nodes[i];
1235 static struct ctdb_node_map *
1236 filter_nodemap_by_flags(struct ctdb_context *ctdb,
1237 struct ctdb_node_map *nodemap,
1238 uint32_t flags_mask)
1241 struct ctdb_node_map *ret;
1243 ret = talloc_nodemap(nodemap);
1244 CTDB_NO_MEMORY_NULL(ctdb, ret);
1248 for (i = 0; i < nodemap->num; i++) {
1249 if (nodemap->nodes[i].flags & flags_mask) {
1253 ret->nodes[ret->num] = nodemap->nodes[i];
1261 display the list of nodes belonging to this natgw configuration
1263 static int control_natgwlist(struct ctdb_context *ctdb, int argc, const char **argv)
1265 TALLOC_CTX *tmp_ctx = talloc_new(ctdb);
1267 struct ctdb_node_map *natgw_nodes = NULL;
1268 struct ctdb_node_map *orig_nodemap=NULL;
1269 struct ctdb_node_map *nodemap;
1270 uint32_t mypnn, pnn;
1273 /* When we have some nodes that could be the NATGW, make a
1274 * series of attempts to find the first node that doesn't have
1275 * certain status flags set.
1277 uint32_t exclude_flags[] = {
1278 /* Look for a nice healthy node */
1279 NODE_FLAGS_DISCONNECTED|NODE_FLAGS_STOPPED|NODE_FLAGS_DELETED|NODE_FLAGS_BANNED|NODE_FLAGS_UNHEALTHY,
1280 /* If not found, an UNHEALTHY/BANNED node will do */
1281 NODE_FLAGS_DISCONNECTED|NODE_FLAGS_STOPPED|NODE_FLAGS_DELETED,
1282 /* If not found, a STOPPED node will do */
1283 NODE_FLAGS_DISCONNECTED|NODE_FLAGS_DELETED,
1287 /* read the natgw nodes file into a linked list */
1288 natgw_nodes = read_natgw_nodes_file(ctdb, tmp_ctx);
1289 if (natgw_nodes == NULL) {
1294 ret = ctdb_ctrl_getnodemap(ctdb, TIMELIMIT(), CTDB_CURRENT_NODE,
1295 tmp_ctx, &orig_nodemap);
1297 DEBUG(DEBUG_ERR, ("Unable to get nodemap from local node.\n"));
1298 talloc_free(tmp_ctx);
1302 /* Get a nodemap that includes only the nodes in the NATGW
1304 nodemap = filter_nodemap_by_addrs(ctdb, orig_nodemap, natgw_nodes);
1305 if (nodemap == NULL) {
1310 ret = 2; /* matches ENOENT */
1313 /* For each flag mask... */
1314 for (i = 0; exclude_flags[i] != 0; i++) {
1315 /* ... get a nodemap that excludes nodes with with
1316 * masked flags... */
1317 struct ctdb_node_map *t =
1318 filter_nodemap_by_flags(ctdb, nodemap,
1326 /* ... and find the first node with the NATGW
1328 struct ctdb_node_map *n;
1329 n = filter_nodemap_by_capabilities(ctdb, t,
1339 pnn = n->nodes[0].pnn;
1340 ip = ctdb_addr_to_str(&n->nodes[0].addr);
1347 if (options.machinereadable) {
1348 printm(":Node:IP:\n");
1349 printm(":%d:%s:\n", pnn, ip);
1351 printf("%d %s\n", pnn, ip);
1354 /* print the pruned list of nodes belonging to this natgw list */
1355 mypnn = getpnn(ctdb);
1356 if (options.machinereadable) {
1357 control_status_header_machine();
1359 printf("Number of nodes:%d\n", nodemap->num);
1361 for(i=0;i<nodemap->num;i++){
1362 if (nodemap->nodes[i].flags & NODE_FLAGS_DELETED) {
1365 if (options.machinereadable) {
1366 control_status_1_machine(ctdb, mypnn, &(nodemap->nodes[i]));
1368 control_status_1_human(ctdb, mypnn, &(nodemap->nodes[i]));
1373 talloc_free(tmp_ctx);
1378 display the status of the scripts for monitoring (or other events)
1380 static int control_one_scriptstatus(struct ctdb_context *ctdb,
1381 enum ctdb_eventscript_call type)
1383 struct ctdb_scripts_wire *script_status;
1386 ret = ctdb_ctrl_getscriptstatus(ctdb, TIMELIMIT(), options.pnn, ctdb, type, &script_status);
1388 DEBUG(DEBUG_ERR, ("Unable to get script status from node %u\n", options.pnn));
1392 if (script_status == NULL) {
1393 if (!options.machinereadable) {
1394 printf("%s cycle never run\n",
1395 ctdb_eventscript_call_names[type]);
1400 if (!options.machinereadable) {
1402 for (i=0; i<script_status->num_scripts; i++) {
1403 if (script_status->scripts[i].status != -ENOEXEC) {
1407 printf("%d scripts were executed last %s cycle\n",
1409 ctdb_eventscript_call_names[type]);
1411 for (i=0; i<script_status->num_scripts; i++) {
1412 const char *status = NULL;
1414 switch (script_status->scripts[i].status) {
1416 status = "TIMEDOUT";
1419 status = "DISABLED";
1425 if (script_status->scripts[i].status > 0)
1429 if (options.machinereadable) {
1430 printm(":%s:%s:%i:%s:%lu.%06lu:%lu.%06lu:%s:\n",
1431 ctdb_eventscript_call_names[type],
1432 script_status->scripts[i].name,
1433 script_status->scripts[i].status,
1435 (long)script_status->scripts[i].start.tv_sec,
1436 (long)script_status->scripts[i].start.tv_usec,
1437 (long)script_status->scripts[i].finished.tv_sec,
1438 (long)script_status->scripts[i].finished.tv_usec,
1439 script_status->scripts[i].output);
1443 printf("%-20s Status:%s ",
1444 script_status->scripts[i].name, status);
1446 /* Some other error, eg from stat. */
1447 printf("%-20s Status:CANNOT RUN (%s)",
1448 script_status->scripts[i].name,
1449 strerror(-script_status->scripts[i].status));
1451 if (script_status->scripts[i].status >= 0) {
1452 printf("Duration:%.3lf ",
1453 timeval_delta(&script_status->scripts[i].finished,
1454 &script_status->scripts[i].start));
1456 if (script_status->scripts[i].status != -ENOEXEC) {
1458 ctime(&script_status->scripts[i].start.tv_sec));
1459 if (script_status->scripts[i].status != 0) {
1460 printf(" OUTPUT:%s\n",
1461 script_status->scripts[i].output);
1471 static int control_scriptstatus(struct ctdb_context *ctdb,
1472 int argc, const char **argv)
1475 enum ctdb_eventscript_call type, min, max;
1479 DEBUG(DEBUG_ERR, ("Unknown arguments to scriptstatus\n"));
1484 arg = ctdb_eventscript_call_names[CTDB_EVENT_MONITOR];
1488 for (type = 0; type < CTDB_EVENT_MAX; type++) {
1489 if (strcmp(arg, ctdb_eventscript_call_names[type]) == 0) {
1495 if (type == CTDB_EVENT_MAX) {
1496 if (strcmp(arg, "all") == 0) {
1498 max = CTDB_EVENT_MAX;
1500 DEBUG(DEBUG_ERR, ("Unknown event type %s\n", argv[0]));
1505 if (options.machinereadable) {
1506 printm(":Type:Name:Code:Status:Start:End:Error Output...:\n");
1509 for (type = min; type < max; type++) {
1510 ret = control_one_scriptstatus(ctdb, type);
1520 enable an eventscript
1522 static int control_enablescript(struct ctdb_context *ctdb, int argc, const char **argv)
1530 ret = ctdb_ctrl_enablescript(ctdb, TIMELIMIT(), options.pnn, argv[0]);
1532 DEBUG(DEBUG_ERR, ("Unable to enable script %s on node %u\n", argv[0], options.pnn));
1540 disable an eventscript
1542 static int control_disablescript(struct ctdb_context *ctdb, int argc, const char **argv)
1550 ret = ctdb_ctrl_disablescript(ctdb, TIMELIMIT(), options.pnn, argv[0]);
1552 DEBUG(DEBUG_ERR, ("Unable to disable script %s on node %u\n", argv[0], options.pnn));
1560 display the pnn of the recovery master
1562 static int control_recmaster(struct ctdb_context *ctdb, int argc, const char **argv)
1567 ret = ctdb_ctrl_getrecmaster(ctdb, ctdb, TIMELIMIT(), options.pnn, &recmaster);
1569 DEBUG(DEBUG_ERR, ("Unable to get recmaster from node %u\n", options.pnn));
1572 printf("%d\n",recmaster);
1578 add a tickle to a public address
1580 static int control_add_tickle(struct ctdb_context *ctdb, int argc, const char **argv)
1582 struct ctdb_tcp_connection t;
1586 assert_single_node_only();
1592 if (parse_ip_port(argv[0], &t.src_addr) == 0) {
1593 DEBUG(DEBUG_ERR,("Wrongly formed ip address '%s'\n", argv[0]));
1596 if (parse_ip_port(argv[1], &t.dst_addr) == 0) {
1597 DEBUG(DEBUG_ERR,("Wrongly formed ip address '%s'\n", argv[1]));
1601 data.dptr = (uint8_t *)&t;
1602 data.dsize = sizeof(t);
1604 /* tell all nodes about this tcp connection */
1605 ret = ctdb_control(ctdb, options.pnn, 0, CTDB_CONTROL_TCP_ADD_DELAYED_UPDATE,
1606 0, data, ctdb, NULL, NULL, NULL, NULL);
1608 DEBUG(DEBUG_ERR,("Failed to add tickle\n"));
1617 delete a tickle from a node
1619 static int control_del_tickle(struct ctdb_context *ctdb, int argc, const char **argv)
1621 struct ctdb_tcp_connection t;
1625 assert_single_node_only();
1631 if (parse_ip_port(argv[0], &t.src_addr) == 0) {
1632 DEBUG(DEBUG_ERR,("Wrongly formed ip address '%s'\n", argv[0]));
1635 if (parse_ip_port(argv[1], &t.dst_addr) == 0) {
1636 DEBUG(DEBUG_ERR,("Wrongly formed ip address '%s'\n", argv[1]));
1640 data.dptr = (uint8_t *)&t;
1641 data.dsize = sizeof(t);
1643 /* tell all nodes about this tcp connection */
1644 ret = ctdb_control(ctdb, options.pnn, 0, CTDB_CONTROL_TCP_REMOVE,
1645 0, data, ctdb, NULL, NULL, NULL, NULL);
1647 DEBUG(DEBUG_ERR,("Failed to remove tickle\n"));
1656 get a list of all tickles for this pnn
1658 static int control_get_tickles(struct ctdb_context *ctdb, int argc, const char **argv)
1660 struct ctdb_control_tcp_tickle_list *list;
1661 ctdb_sock_addr addr;
1665 assert_single_node_only();
1672 port = atoi(argv[1]);
1675 if (parse_ip(argv[0], NULL, 0, &addr) == 0) {
1676 DEBUG(DEBUG_ERR,("Wrongly formed ip address '%s'\n", argv[0]));
1680 ret = ctdb_ctrl_get_tcp_tickles(ctdb, TIMELIMIT(), options.pnn, ctdb, &addr, &list);
1682 DEBUG(DEBUG_ERR, ("Unable to list tickles\n"));
1686 if (options.machinereadable){
1687 printm(":source ip:port:destination ip:port:\n");
1688 for (i=0;i<list->tickles.num;i++) {
1689 if (port && port != ntohs(list->tickles.connections[i].dst_addr.ip.sin_port)) {
1692 printm(":%s:%u", ctdb_addr_to_str(&list->tickles.connections[i].src_addr), ntohs(list->tickles.connections[i].src_addr.ip.sin_port));
1693 printm(":%s:%u:\n", ctdb_addr_to_str(&list->tickles.connections[i].dst_addr), ntohs(list->tickles.connections[i].dst_addr.ip.sin_port));
1696 printf("Tickles for ip:%s\n", ctdb_addr_to_str(&list->addr));
1697 printf("Num tickles:%u\n", list->tickles.num);
1698 for (i=0;i<list->tickles.num;i++) {
1699 if (port && port != ntohs(list->tickles.connections[i].dst_addr.ip.sin_port)) {
1702 printf("SRC: %s:%u ", ctdb_addr_to_str(&list->tickles.connections[i].src_addr), ntohs(list->tickles.connections[i].src_addr.ip.sin_port));
1703 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));
1713 static int move_ip(struct ctdb_context *ctdb, ctdb_sock_addr *addr, uint32_t pnn)
1715 struct ctdb_all_public_ips *ips;
1716 struct ctdb_public_ip ip;
1719 uint32_t disable_time;
1721 struct ctdb_node_map *nodemap=NULL;
1722 TALLOC_CTX *tmp_ctx = talloc_new(ctdb);
1725 data.dptr = (uint8_t*)&disable_time;
1726 data.dsize = sizeof(disable_time);
1727 ret = ctdb_client_send_message(ctdb, CTDB_BROADCAST_CONNECTED, CTDB_SRVID_DISABLE_IP_CHECK, data);
1729 DEBUG(DEBUG_ERR,("Failed to send message to disable ipcheck\n"));
1735 /* read the public ip list from the node */
1736 ret = ctdb_ctrl_get_public_ips(ctdb, TIMELIMIT(), pnn, ctdb, &ips);
1738 DEBUG(DEBUG_ERR, ("Unable to get public ip list from node %u\n", pnn));
1739 talloc_free(tmp_ctx);
1743 for (i=0;i<ips->num;i++) {
1744 if (ctdb_same_ip(addr, &ips->ips[i].addr)) {
1749 DEBUG(DEBUG_ERR, ("Node %u can not host ip address '%s'\n",
1750 pnn, ctdb_addr_to_str(addr)));
1751 talloc_free(tmp_ctx);
1758 data.dptr = (uint8_t *)&ip;
1759 data.dsize = sizeof(ip);
1761 ret = ctdb_ctrl_getnodemap(ctdb, TIMELIMIT(), options.pnn, tmp_ctx, &nodemap);
1763 DEBUG(DEBUG_ERR, ("Unable to get nodemap from node %u\n", options.pnn));
1764 talloc_free(tmp_ctx);
1768 nodes = list_of_nodes(ctdb, nodemap, tmp_ctx, NODE_FLAGS_INACTIVE, pnn);
1769 ret = ctdb_client_async_control(ctdb, CTDB_CONTROL_RELEASE_IP,
1776 DEBUG(DEBUG_ERR,("Failed to release IP on nodes\n"));
1777 talloc_free(tmp_ctx);
1781 ret = ctdb_ctrl_takeover_ip(ctdb, LONGTIMELIMIT(), pnn, &ip);
1783 DEBUG(DEBUG_ERR,("Failed to take over IP on node %d\n", pnn));
1784 talloc_free(tmp_ctx);
1788 /* update the recovery daemon so it now knows to expect the new
1789 node assignment for this ip.
1791 ret = ctdb_client_send_message(ctdb, CTDB_BROADCAST_CONNECTED, CTDB_SRVID_RECD_UPDATE_IP, data);
1793 DEBUG(DEBUG_ERR,("Failed to send message to update the ip on the recovery master.\n"));
1797 talloc_free(tmp_ctx);
1803 * scans all other nodes and returns a pnn for another node that can host this
1807 find_other_host_for_public_ip(struct ctdb_context *ctdb, ctdb_sock_addr *addr)
1809 TALLOC_CTX *tmp_ctx = talloc_new(ctdb);
1810 struct ctdb_all_public_ips *ips;
1811 struct ctdb_node_map *nodemap=NULL;
1815 ret = ctdb_ctrl_getnodemap(ctdb, TIMELIMIT(), CTDB_CURRENT_NODE, tmp_ctx, &nodemap);
1817 DEBUG(DEBUG_ERR, ("Unable to get nodemap from node %u\n", options.pnn));
1818 talloc_free(tmp_ctx);
1822 for(i=0;i<nodemap->num;i++){
1823 if (nodemap->nodes[i].flags & NODE_FLAGS_INACTIVE) {
1826 if (nodemap->nodes[i].pnn == options.pnn) {
1830 /* read the public ip list from this node */
1831 ret = ctdb_ctrl_get_public_ips(ctdb, TIMELIMIT(), nodemap->nodes[i].pnn, tmp_ctx, &ips);
1833 DEBUG(DEBUG_ERR, ("Unable to get public ip list from node %u\n", nodemap->nodes[i].pnn));
1837 for (j=0;j<ips->num;j++) {
1838 if (ctdb_same_ip(addr, &ips->ips[j].addr)) {
1839 pnn = nodemap->nodes[i].pnn;
1840 talloc_free(tmp_ctx);
1847 talloc_free(tmp_ctx);
1851 /* If pnn is -1 then try to find a node to move IP to... */
1852 static bool try_moveip(struct ctdb_context *ctdb, ctdb_sock_addr *addr, uint32_t pnn)
1854 bool pnn_specified = (pnn == -1 ? false : true);
1857 while (retries < 5) {
1858 if (!pnn_specified) {
1859 pnn = find_other_host_for_public_ip(ctdb, addr);
1864 ("Trying to move public IP to node %u\n", pnn));
1867 if (move_ip(ctdb, addr, pnn) == 0) {
1880 move/failover an ip address to a specific node
1882 static int control_moveip(struct ctdb_context *ctdb, int argc, const char **argv)
1885 ctdb_sock_addr addr;
1887 assert_single_node_only();
1894 if (parse_ip(argv[0], NULL, 0, &addr) == 0) {
1895 DEBUG(DEBUG_ERR,("Wrongly formed ip address '%s'\n", argv[0]));
1900 if (sscanf(argv[1], "%u", &pnn) != 1) {
1901 DEBUG(DEBUG_ERR, ("Badly formed pnn\n"));
1905 if (!try_moveip(ctdb, &addr, pnn)) {
1906 DEBUG(DEBUG_ERR,("Failed to move IP to node %d.\n", pnn));
1913 static int rebalance_node(struct ctdb_context *ctdb, uint32_t pnn)
1917 data.dptr = (uint8_t *)&pnn;
1918 data.dsize = sizeof(uint32_t);
1919 if (ctdb_client_send_message(ctdb, CTDB_BROADCAST_CONNECTED, CTDB_SRVID_REBALANCE_NODE, data) != 0) {
1921 ("Failed to send message to force node %u to be a rebalancing target\n",
1931 rebalance a node by setting it to allow failback and triggering a
1934 static int control_rebalancenode(struct ctdb_context *ctdb, int argc, const char **argv)
1936 TALLOC_CTX *tmp_ctx = talloc_new(ctdb);
1941 assert_single_node_only();
1947 /* Determine the nodes where IPs need to be reloaded */
1948 if (!parse_nodestring(ctdb, tmp_ctx, argc == 1 ? argv[0] : NULL,
1949 options.pnn, true, &nodes, &pnn_mode)) {
1954 for (i = 0; i < talloc_array_length(nodes); i++) {
1955 if (!rebalance_node(ctdb, nodes[i])) {
1961 talloc_free(tmp_ctx);
1965 static int rebalance_ip(struct ctdb_context *ctdb, ctdb_sock_addr *addr)
1967 struct ctdb_public_ip ip;
1970 uint32_t disable_time;
1972 struct ctdb_node_map *nodemap=NULL;
1973 TALLOC_CTX *tmp_ctx = talloc_new(ctdb);
1976 data.dptr = (uint8_t*)&disable_time;
1977 data.dsize = sizeof(disable_time);
1978 ret = ctdb_client_send_message(ctdb, CTDB_BROADCAST_CONNECTED, CTDB_SRVID_DISABLE_IP_CHECK, data);
1980 DEBUG(DEBUG_ERR,("Failed to send message to disable ipcheck\n"));
1987 data.dptr = (uint8_t *)&ip;
1988 data.dsize = sizeof(ip);
1990 ret = ctdb_ctrl_getnodemap(ctdb, TIMELIMIT(), options.pnn, tmp_ctx, &nodemap);
1992 DEBUG(DEBUG_ERR, ("Unable to get nodemap from node %u\n", options.pnn));
1993 talloc_free(tmp_ctx);
1997 nodes = list_of_active_nodes(ctdb, nodemap, tmp_ctx, true);
1998 ret = ctdb_client_async_control(ctdb, CTDB_CONTROL_RELEASE_IP,
2005 DEBUG(DEBUG_ERR,("Failed to release IP on nodes\n"));
2006 talloc_free(tmp_ctx);
2010 talloc_free(tmp_ctx);
2015 release an ip form all nodes and have it re-assigned by recd
2017 static int control_rebalanceip(struct ctdb_context *ctdb, int argc, const char **argv)
2019 ctdb_sock_addr addr;
2021 assert_single_node_only();
2028 if (parse_ip(argv[0], NULL, 0, &addr) == 0) {
2029 DEBUG(DEBUG_ERR,("Wrongly formed ip address '%s'\n", argv[0]));
2033 if (rebalance_ip(ctdb, &addr) != 0) {
2034 DEBUG(DEBUG_ERR,("Error when trying to reassign ip\n"));
2041 static int getips_store_callback(void *param, void *data)
2043 struct ctdb_public_ip *node_ip = (struct ctdb_public_ip *)data;
2044 struct ctdb_all_public_ips *ips = param;
2048 ips->ips[i].pnn = node_ip->pnn;
2049 ips->ips[i].addr = node_ip->addr;
2053 static int getips_count_callback(void *param, void *data)
2055 uint32_t *count = param;
2062 static uint32_t *ip_key(ctdb_sock_addr *ip)
2064 static uint32_t key[IP_KEYLEN];
2066 bzero(key, sizeof(key));
2068 switch (ip->sa.sa_family) {
2070 key[0] = ip->ip.sin_addr.s_addr;
2073 uint32_t *s6_a32 = (uint32_t *)&(ip->ip6.sin6_addr.s6_addr);
2081 DEBUG(DEBUG_ERR, (__location__ " ERROR, unknown family passed :%u\n", ip->sa.sa_family));
2088 static void *add_ip_callback(void *parm, void *data)
2094 control_get_all_public_ips(struct ctdb_context *ctdb, TALLOC_CTX *tmp_ctx, struct ctdb_all_public_ips **ips)
2096 struct ctdb_all_public_ips *tmp_ips;
2097 struct ctdb_node_map *nodemap=NULL;
2098 trbt_tree_t *ip_tree;
2102 ret = ctdb_ctrl_getnodemap(ctdb, TIMELIMIT(), CTDB_CURRENT_NODE, tmp_ctx, &nodemap);
2104 DEBUG(DEBUG_ERR, ("Unable to get nodemap from node %u\n", options.pnn));
2108 ip_tree = trbt_create(tmp_ctx, 0);
2110 for(i=0;i<nodemap->num;i++){
2111 if (nodemap->nodes[i].flags & NODE_FLAGS_DELETED) {
2114 if (nodemap->nodes[i].flags & NODE_FLAGS_DISCONNECTED) {
2118 /* read the public ip list from this node */
2119 ret = ctdb_ctrl_get_public_ips(ctdb, TIMELIMIT(), nodemap->nodes[i].pnn, tmp_ctx, &tmp_ips);
2121 DEBUG(DEBUG_ERR, ("Unable to get public ip list from node %u\n", nodemap->nodes[i].pnn));
2125 for (j=0; j<tmp_ips->num;j++) {
2126 struct ctdb_public_ip *node_ip;
2128 node_ip = talloc(tmp_ctx, struct ctdb_public_ip);
2129 node_ip->pnn = tmp_ips->ips[j].pnn;
2130 node_ip->addr = tmp_ips->ips[j].addr;
2132 trbt_insertarray32_callback(ip_tree,
2133 IP_KEYLEN, ip_key(&tmp_ips->ips[j].addr),
2137 talloc_free(tmp_ips);
2142 trbt_traversearray32(ip_tree, IP_KEYLEN, getips_count_callback, &count);
2144 len = offsetof(struct ctdb_all_public_ips, ips) +
2145 count*sizeof(struct ctdb_public_ip);
2146 tmp_ips = talloc_zero_size(tmp_ctx, len);
2147 trbt_traversearray32(ip_tree, IP_KEYLEN, getips_store_callback, tmp_ips);
2155 static void ctdb_every_second(struct event_context *ev, struct timed_event *te, struct timeval t, void *p)
2157 struct ctdb_context *ctdb = talloc_get_type(p, struct ctdb_context);
2159 event_add_timed(ctdb->ev, ctdb,
2160 timeval_current_ofs(1, 0),
2161 ctdb_every_second, ctdb);
2164 struct srvid_reply_handler_data {
2168 const char *srvid_str;
2171 static void srvid_broadcast_reply_handler(struct ctdb_context *ctdb,
2176 struct srvid_reply_handler_data *d =
2177 (struct srvid_reply_handler_data *)private_data;
2181 if (data.dsize != sizeof(ret)) {
2182 DEBUG(DEBUG_ERR, (__location__ " Wrong reply size\n"));
2186 /* ret will be a PNN (i.e. >=0) on success, or negative on error */
2187 ret = *(int32_t *)data.dptr;
2190 ("%s failed with result %d\n", d->srvid_str, ret));
2194 if (!d->wait_for_all) {
2199 /* Wait for all replies */
2201 for (i = 0; i < talloc_array_length(d->nodes); i++) {
2202 if (d->nodes[i] == ret) {
2204 ("%s reply received from node %u\n",
2205 d->srvid_str, ret));
2208 if (d->nodes[i] != -1) {
2209 /* Found a node that hasn't yet replied */
2215 /* Broadcast the given SRVID to all connected nodes. Wait for 1 reply
2216 * or replies from all connected nodes. arg is the data argument to
2217 * pass in the srvid_request structure - pass 0 if this isn't needed.
2219 static int srvid_broadcast(struct ctdb_context *ctdb,
2220 uint64_t srvid, uint32_t *arg,
2221 const char *srvid_str, bool wait_for_all)
2226 uint64_t reply_srvid;
2227 struct srvid_request request;
2228 struct srvid_request_data request_data;
2229 struct srvid_reply_handler_data reply_data;
2232 ZERO_STRUCT(request);
2234 /* Time ticks to enable timeouts to be processed */
2235 event_add_timed(ctdb->ev, ctdb,
2236 timeval_current_ofs(1, 0),
2237 ctdb_every_second, ctdb);
2239 pnn = ctdb_get_pnn(ctdb);
2240 reply_srvid = getpid();
2244 request.srvid = reply_srvid;
2246 data.dptr = (uint8_t *)&request;
2247 data.dsize = sizeof(request);
2249 request_data.pnn = pnn;
2250 request_data.srvid = reply_srvid;
2251 request_data.data = *arg;
2253 data.dptr = (uint8_t *)&request_data;
2254 data.dsize = sizeof(request_data);
2257 /* Register message port for reply from recovery master */
2258 ctdb_client_set_message_handler(ctdb, reply_srvid,
2259 srvid_broadcast_reply_handler,
2262 reply_data.wait_for_all = wait_for_all;
2263 reply_data.nodes = NULL;
2264 reply_data.srvid_str = srvid_str;
2267 reply_data.done = false;
2270 struct ctdb_node_map *nodemap;
2272 ret = ctdb_ctrl_getnodemap(ctdb, TIMELIMIT(),
2273 CTDB_CURRENT_NODE, ctdb, &nodemap);
2276 ("Unable to get nodemap from current node, try again\n"));
2281 if (reply_data.nodes != NULL) {
2282 talloc_free(reply_data.nodes);
2284 reply_data.nodes = list_of_connected_nodes(ctdb, nodemap,
2287 talloc_free(nodemap);
2290 /* Send to all connected nodes. Only recmaster replies */
2291 ret = ctdb_client_send_message(ctdb, CTDB_BROADCAST_CONNECTED,
2294 /* This can only happen if the socket is closed and
2295 * there's no way to recover from that, so don't try
2299 ("Failed to send %s request to connected nodes\n",
2304 tv = timeval_current();
2305 /* This loop terminates the reply is received */
2306 while (timeval_elapsed(&tv) < 5.0 && !reply_data.done) {
2307 event_loop_once(ctdb->ev);
2310 if (!reply_data.done) {
2312 ("Still waiting for confirmation of %s\n", srvid_str));
2317 ctdb_client_remove_message_handler(ctdb, reply_srvid, &reply_data);
2319 talloc_free(reply_data.nodes);
2324 static int ipreallocate(struct ctdb_context *ctdb)
2326 return srvid_broadcast(ctdb, CTDB_SRVID_TAKEOVER_RUN, NULL,
2327 "IP reallocation", false);
2331 static int control_ipreallocate(struct ctdb_context *ctdb, int argc, const char **argv)
2333 return ipreallocate(ctdb);
2337 add a public ip address to a node
2339 static int control_addip(struct ctdb_context *ctdb, int argc, const char **argv)
2342 int len, retries = 0;
2344 ctdb_sock_addr addr;
2345 struct ctdb_control_ip_iface *pub;
2346 TALLOC_CTX *tmp_ctx = talloc_new(ctdb);
2347 struct ctdb_all_public_ips *ips;
2351 talloc_free(tmp_ctx);
2355 if (!parse_ip_mask(argv[0], argv[1], &addr, &mask)) {
2356 DEBUG(DEBUG_ERR, ("Badly formed ip/mask : %s\n", argv[0]));
2357 talloc_free(tmp_ctx);
2361 /* read the public ip list from the node */
2362 ret = ctdb_ctrl_get_public_ips(ctdb, TIMELIMIT(), options.pnn, tmp_ctx, &ips);
2364 DEBUG(DEBUG_ERR, ("Unable to get public ip list from node %u\n", options.pnn));
2365 talloc_free(tmp_ctx);
2368 for (i=0;i<ips->num;i++) {
2369 if (ctdb_same_ip(&addr, &ips->ips[i].addr)) {
2370 DEBUG(DEBUG_ERR,("Can not add ip to node. Node already hosts this ip\n"));
2377 /* Dont timeout. This command waits for an ip reallocation
2378 which sometimes can take wuite a while if there has
2379 been a recent recovery
2383 len = offsetof(struct ctdb_control_ip_iface, iface) + strlen(argv[1]) + 1;
2384 pub = talloc_size(tmp_ctx, len);
2385 CTDB_NO_MEMORY(ctdb, pub);
2389 pub->len = strlen(argv[1])+1;
2390 memcpy(&pub->iface[0], argv[1], strlen(argv[1])+1);
2393 ret = ctdb_ctrl_add_public_ip(ctdb, TIMELIMIT(), options.pnn, pub);
2395 DEBUG(DEBUG_ERR, ("Unable to add public ip to node %u. Wait 3 seconds and try again.\n", options.pnn));
2399 } while (retries < 5 && ret != 0);
2401 DEBUG(DEBUG_ERR, ("Unable to add public ip to node %u. Giving up.\n", options.pnn));
2402 talloc_free(tmp_ctx);
2406 if (rebalance_node(ctdb, options.pnn) != 0) {
2407 DEBUG(DEBUG_ERR,("Error when trying to rebalance node\n"));
2411 talloc_free(tmp_ctx);
2416 add a public ip address to a node
2418 static int control_ipiface(struct ctdb_context *ctdb, int argc, const char **argv)
2420 ctdb_sock_addr addr;
2427 if (!parse_ip(argv[0], NULL, 0, &addr)) {
2428 printf("Badly formed ip : %s\n", argv[0]);
2432 iface = ctdb_sys_find_ifname(&addr);
2433 if (iface == NULL) {
2434 printf("Failed to get interface name for ip: %s", argv[0]);
2438 printf("IP on interface %s\n", iface);
2445 static int control_delip(struct ctdb_context *ctdb, int argc, const char **argv);
2447 static int control_delip_all(struct ctdb_context *ctdb, int argc, const char **argv, ctdb_sock_addr *addr)
2449 TALLOC_CTX *tmp_ctx = talloc_new(ctdb);
2450 struct ctdb_node_map *nodemap=NULL;
2451 struct ctdb_all_public_ips *ips;
2454 ret = ctdb_ctrl_getnodemap(ctdb, TIMELIMIT(), CTDB_CURRENT_NODE, tmp_ctx, &nodemap);
2456 DEBUG(DEBUG_ERR, ("Unable to get nodemap from current node\n"));
2460 /* remove it from the nodes that are not hosting the ip currently */
2461 for(i=0;i<nodemap->num;i++){
2462 if (nodemap->nodes[i].flags & NODE_FLAGS_INACTIVE) {
2465 ret = ctdb_ctrl_get_public_ips(ctdb, TIMELIMIT(), nodemap->nodes[i].pnn, tmp_ctx, &ips);
2467 DEBUG(DEBUG_ERR, ("Unable to get public ip list from node %d\n", nodemap->nodes[i].pnn));
2471 for (j=0;j<ips->num;j++) {
2472 if (ctdb_same_ip(addr, &ips->ips[j].addr)) {
2480 if (ips->ips[j].pnn == nodemap->nodes[i].pnn) {
2484 options.pnn = nodemap->nodes[i].pnn;
2485 control_delip(ctdb, argc, argv);
2489 /* remove it from every node (also the one hosting it) */
2490 for(i=0;i<nodemap->num;i++){
2491 if (nodemap->nodes[i].flags & NODE_FLAGS_INACTIVE) {
2494 ret = ctdb_ctrl_get_public_ips(ctdb, TIMELIMIT(), nodemap->nodes[i].pnn, tmp_ctx, &ips);
2496 DEBUG(DEBUG_ERR, ("Unable to get public ip list from node %d\n", nodemap->nodes[i].pnn));
2500 for (j=0;j<ips->num;j++) {
2501 if (ctdb_same_ip(addr, &ips->ips[j].addr)) {
2509 options.pnn = nodemap->nodes[i].pnn;
2510 control_delip(ctdb, argc, argv);
2513 talloc_free(tmp_ctx);
2518 delete a public ip address from a node
2520 static int control_delip(struct ctdb_context *ctdb, int argc, const char **argv)
2523 ctdb_sock_addr addr;
2524 struct ctdb_control_ip_iface pub;
2525 TALLOC_CTX *tmp_ctx = talloc_new(ctdb);
2526 struct ctdb_all_public_ips *ips;
2529 talloc_free(tmp_ctx);
2533 if (parse_ip(argv[0], NULL, 0, &addr) == 0) {
2534 DEBUG(DEBUG_ERR,("Wrongly formed ip address '%s'\n", argv[0]));
2538 if (options.pnn == CTDB_BROADCAST_ALL) {
2539 return control_delip_all(ctdb, argc, argv, &addr);
2546 ret = ctdb_ctrl_get_public_ips(ctdb, TIMELIMIT(), options.pnn, tmp_ctx, &ips);
2548 DEBUG(DEBUG_ERR, ("Unable to get public ip list from cluster\n"));
2549 talloc_free(tmp_ctx);
2553 for (i=0;i<ips->num;i++) {
2554 if (ctdb_same_ip(&addr, &ips->ips[i].addr)) {
2560 DEBUG(DEBUG_ERR, ("This node does not support this public address '%s'\n",
2561 ctdb_addr_to_str(&addr)));
2562 talloc_free(tmp_ctx);
2566 /* This is an optimisation. If this node is hosting the IP
2567 * then try to move it somewhere else without invoking a full
2568 * takeover run. We don't care if this doesn't work!
2570 if (ips->ips[i].pnn == options.pnn) {
2571 (void) try_moveip(ctdb, &addr, -1);
2574 ret = ctdb_ctrl_del_public_ip(ctdb, TIMELIMIT(), options.pnn, &pub);
2576 DEBUG(DEBUG_ERR, ("Unable to del public ip from node %u\n", options.pnn));
2577 talloc_free(tmp_ctx);
2581 talloc_free(tmp_ctx);
2585 static int kill_tcp_from_file(struct ctdb_context *ctdb,
2586 int argc, const char **argv)
2588 struct ctdb_control_killtcp *killtcp;
2589 int max_entries, current, i;
2590 struct timeval timeout;
2591 char line[128], src[128], dst[128];
2594 struct client_async_data *async_data;
2595 struct ctdb_client_control_state *state;
2605 while (!feof(stdin)) {
2606 if (fgets(line, sizeof(line), stdin) == NULL) {
2610 /* Silently skip empty lines */
2611 if (line[0] == '\n') {
2615 if (sscanf(line, "%s %s\n", src, dst) != 2) {
2616 DEBUG(DEBUG_ERR, ("Bad line [%d]: '%s'\n",
2618 talloc_free(killtcp);
2622 if (current >= max_entries) {
2623 max_entries += 1024;
2624 killtcp = talloc_realloc(ctdb, killtcp,
2625 struct ctdb_control_killtcp,
2627 CTDB_NO_MEMORY(ctdb, killtcp);
2630 if (!parse_ip_port(src, &killtcp[current].src_addr)) {
2631 DEBUG(DEBUG_ERR, ("Bad IP:port on line [%d]: '%s'\n",
2633 talloc_free(killtcp);
2637 if (!parse_ip_port(dst, &killtcp[current].dst_addr)) {
2638 DEBUG(DEBUG_ERR, ("Bad IP:port on line [%d]: '%s'\n",
2640 talloc_free(killtcp);
2647 async_data = talloc_zero(ctdb, struct client_async_data);
2648 if (async_data == NULL) {
2649 talloc_free(killtcp);
2653 for (i = 0; i < current; i++) {
2655 data.dsize = sizeof(struct ctdb_control_killtcp);
2656 data.dptr = (unsigned char *)&killtcp[i];
2658 timeout = TIMELIMIT();
2659 state = ctdb_control_send(ctdb, options.pnn, 0,
2660 CTDB_CONTROL_KILL_TCP, 0, data,
2661 async_data, &timeout, NULL);
2663 if (state == NULL) {
2665 ("Failed to call async killtcp control to node %u\n",
2667 talloc_free(killtcp);
2671 ctdb_client_async_add(async_data, state);
2674 if (ctdb_client_async_wait(ctdb, async_data) != 0) {
2675 DEBUG(DEBUG_ERR,("killtcp failed\n"));
2676 talloc_free(killtcp);
2680 talloc_free(killtcp);
2686 kill a tcp connection
2688 static int kill_tcp(struct ctdb_context *ctdb, int argc, const char **argv)
2691 struct ctdb_control_killtcp killtcp;
2693 assert_single_node_only();
2696 return kill_tcp_from_file(ctdb, argc, argv);
2703 if (!parse_ip_port(argv[0], &killtcp.src_addr)) {
2704 DEBUG(DEBUG_ERR, ("Bad IP:port '%s'\n", argv[0]));
2708 if (!parse_ip_port(argv[1], &killtcp.dst_addr)) {
2709 DEBUG(DEBUG_ERR, ("Bad IP:port '%s'\n", argv[1]));
2713 ret = ctdb_ctrl_killtcp(ctdb, TIMELIMIT(), options.pnn, &killtcp);
2715 DEBUG(DEBUG_ERR, ("Unable to killtcp from node %u\n", options.pnn));
2726 static int control_gratious_arp(struct ctdb_context *ctdb, int argc, const char **argv)
2729 ctdb_sock_addr addr;
2731 assert_single_node_only();
2737 if (!parse_ip(argv[0], NULL, 0, &addr)) {
2738 DEBUG(DEBUG_ERR, ("Bad IP '%s'\n", argv[0]));
2742 ret = ctdb_ctrl_gratious_arp(ctdb, TIMELIMIT(), options.pnn, &addr, argv[1]);
2744 DEBUG(DEBUG_ERR, ("Unable to send gratious_arp from node %u\n", options.pnn));
2752 register a server id
2754 static int regsrvid(struct ctdb_context *ctdb, int argc, const char **argv)
2757 struct ctdb_server_id server_id;
2763 server_id.pnn = strtoul(argv[0], NULL, 0);
2764 server_id.type = strtoul(argv[1], NULL, 0);
2765 server_id.server_id = strtoul(argv[2], NULL, 0);
2767 ret = ctdb_ctrl_register_server_id(ctdb, TIMELIMIT(), &server_id);
2769 DEBUG(DEBUG_ERR, ("Unable to register server_id from node %u\n", options.pnn));
2772 DEBUG(DEBUG_ERR,("Srvid registered. Sleeping for 999 seconds\n"));
2778 unregister a server id
2780 static int unregsrvid(struct ctdb_context *ctdb, int argc, const char **argv)
2783 struct ctdb_server_id server_id;
2789 server_id.pnn = strtoul(argv[0], NULL, 0);
2790 server_id.type = strtoul(argv[1], NULL, 0);
2791 server_id.server_id = strtoul(argv[2], NULL, 0);
2793 ret = ctdb_ctrl_unregister_server_id(ctdb, TIMELIMIT(), &server_id);
2795 DEBUG(DEBUG_ERR, ("Unable to unregister server_id from node %u\n", options.pnn));
2802 check if a server id exists
2804 static int chksrvid(struct ctdb_context *ctdb, int argc, const char **argv)
2806 uint32_t status = 0;
2808 struct ctdb_server_id server_id;
2814 server_id.pnn = strtoul(argv[0], NULL, 0);
2815 server_id.type = strtoul(argv[1], NULL, 0);
2816 server_id.server_id = strtoul(argv[2], NULL, 0);
2818 ret = ctdb_ctrl_check_server_id(ctdb, TIMELIMIT(), options.pnn, &server_id, &status);
2820 DEBUG(DEBUG_ERR, ("Unable to check server_id from node %u\n", options.pnn));
2825 printf("Server id %d:%d:%d EXISTS\n", server_id.pnn, server_id.type, server_id.server_id);
2827 printf("Server id %d:%d:%d does NOT exist\n", server_id.pnn, server_id.type, server_id.server_id);
2833 get a list of all server ids that are registered on a node
2835 static int getsrvids(struct ctdb_context *ctdb, int argc, const char **argv)
2838 struct ctdb_server_id_list *server_ids;
2840 ret = ctdb_ctrl_get_server_id_list(ctdb, ctdb, TIMELIMIT(), options.pnn, &server_ids);
2842 DEBUG(DEBUG_ERR, ("Unable to get server_id list from node %u\n", options.pnn));
2846 for (i=0; i<server_ids->num; i++) {
2847 printf("Server id %d:%d:%d\n",
2848 server_ids->server_ids[i].pnn,
2849 server_ids->server_ids[i].type,
2850 server_ids->server_ids[i].server_id);
2857 check if a server id exists
2859 static int check_srvids(struct ctdb_context *ctdb, int argc, const char **argv)
2861 TALLOC_CTX *tmp_ctx = talloc_new(NULL);
2867 talloc_free(tmp_ctx);
2871 ids = talloc_array(tmp_ctx, uint64_t, argc);
2872 result = talloc_array(tmp_ctx, uint8_t, argc);
2874 for (i = 0; i < argc; i++) {
2875 ids[i] = strtoull(argv[i], NULL, 0);
2878 if (!ctdb_client_check_message_handlers(ctdb, ids, argc, result)) {
2879 DEBUG(DEBUG_ERR, ("Unable to check server_id from node %u\n",
2881 talloc_free(tmp_ctx);
2885 for (i=0; i < argc; i++) {
2886 printf("Server id %d:%llu %s\n", options.pnn, (long long)ids[i],
2887 result[i] ? "exists" : "does not exist");
2890 talloc_free(tmp_ctx);
2895 send a tcp tickle ack
2897 static int tickle_tcp(struct ctdb_context *ctdb, int argc, const char **argv)
2900 ctdb_sock_addr src, dst;
2906 if (!parse_ip_port(argv[0], &src)) {
2907 DEBUG(DEBUG_ERR, ("Bad IP:port '%s'\n", argv[0]));
2911 if (!parse_ip_port(argv[1], &dst)) {
2912 DEBUG(DEBUG_ERR, ("Bad IP:port '%s'\n", argv[1]));
2916 ret = ctdb_sys_send_tcp(&src, &dst, 0, 0, 0);
2920 DEBUG(DEBUG_ERR, ("Error while sending tickle ack\n"));
2927 display public ip status
2929 static int control_ip(struct ctdb_context *ctdb, int argc, const char **argv)
2932 TALLOC_CTX *tmp_ctx = talloc_new(ctdb);
2933 struct ctdb_all_public_ips *ips;
2935 if (options.pnn == CTDB_BROADCAST_ALL) {
2936 /* read the list of public ips from all nodes */
2937 ret = control_get_all_public_ips(ctdb, tmp_ctx, &ips);
2939 /* read the public ip list from this node */
2940 ret = ctdb_ctrl_get_public_ips(ctdb, TIMELIMIT(), options.pnn, tmp_ctx, &ips);
2943 DEBUG(DEBUG_ERR, ("Unable to get public ips from node %u\n", options.pnn));
2944 talloc_free(tmp_ctx);
2948 if (options.machinereadable){
2949 printm(":Public IP:Node:");
2950 if (options.verbose){
2951 printm("ActiveInterface:AvailableInterfaces:ConfiguredInterfaces:");
2955 if (options.pnn == CTDB_BROADCAST_ALL) {
2956 printf("Public IPs on ALL nodes\n");
2958 printf("Public IPs on node %u\n", options.pnn);
2962 for (i=1;i<=ips->num;i++) {
2963 struct ctdb_control_public_ip_info *info = NULL;
2965 char *aciface = NULL;
2966 char *avifaces = NULL;
2967 char *cifaces = NULL;
2969 if (options.pnn == CTDB_BROADCAST_ALL) {
2970 pnn = ips->ips[ips->num-i].pnn;
2976 ret = ctdb_ctrl_get_public_ip_info(ctdb, TIMELIMIT(), pnn, ctdb,
2977 &ips->ips[ips->num-i].addr, &info);
2984 for (j=0; j < info->num; j++) {
2985 if (cifaces == NULL) {
2986 cifaces = talloc_strdup(info,
2987 info->ifaces[j].name);
2989 cifaces = talloc_asprintf_append(cifaces,
2991 info->ifaces[j].name);
2994 if (info->active_idx == j) {
2995 aciface = info->ifaces[j].name;
2998 if (info->ifaces[j].link_state == 0) {
3002 if (avifaces == NULL) {
3003 avifaces = talloc_strdup(info, info->ifaces[j].name);
3005 avifaces = talloc_asprintf_append(avifaces,
3007 info->ifaces[j].name);
3012 if (options.machinereadable){
3014 ctdb_addr_to_str(&ips->ips[ips->num-i].addr),
3015 ips->ips[ips->num-i].pnn);
3016 if (options.verbose){
3019 avifaces?avifaces:"",
3020 cifaces?cifaces:"");
3024 if (options.verbose) {
3025 printf("%s node[%d] active[%s] available[%s] configured[%s]\n",
3026 ctdb_addr_to_str(&ips->ips[ips->num-i].addr),
3027 ips->ips[ips->num-i].pnn,
3029 avifaces?avifaces:"",
3030 cifaces?cifaces:"");
3033 ctdb_addr_to_str(&ips->ips[ips->num-i].addr),
3034 ips->ips[ips->num-i].pnn);
3040 talloc_free(tmp_ctx);
3047 static int control_ipinfo(struct ctdb_context *ctdb, int argc, const char **argv)
3050 ctdb_sock_addr addr;
3051 TALLOC_CTX *tmp_ctx = talloc_new(ctdb);
3052 struct ctdb_control_public_ip_info *info;
3055 talloc_free(tmp_ctx);
3059 if (parse_ip(argv[0], NULL, 0, &addr) == 0) {
3060 DEBUG(DEBUG_ERR,("Wrongly formed ip address '%s'\n", argv[0]));
3064 /* read the public ip info from this node */
3065 ret = ctdb_ctrl_get_public_ip_info(ctdb, TIMELIMIT(), options.pnn,
3066 tmp_ctx, &addr, &info);
3068 DEBUG(DEBUG_ERR, ("Unable to get public ip[%s]info from node %u\n",
3069 argv[0], options.pnn));
3070 talloc_free(tmp_ctx);
3074 printf("Public IP[%s] info on node %u\n",
3075 ctdb_addr_to_str(&info->ip.addr),
3078 printf("IP:%s\nCurrentNode:%d\nNumInterfaces:%u\n",
3079 ctdb_addr_to_str(&info->ip.addr),
3080 info->ip.pnn, info->num);
3082 for (i=0; i<info->num; i++) {
3083 info->ifaces[i].name[CTDB_IFACE_SIZE] = '\0';
3085 printf("Interface[%u]: Name:%s Link:%s References:%u%s\n",
3086 i+1, info->ifaces[i].name,
3087 info->ifaces[i].link_state?"up":"down",
3088 (unsigned int)info->ifaces[i].references,
3089 (i==info->active_idx)?" (active)":"");
3092 talloc_free(tmp_ctx);
3097 display interfaces status
3099 static int control_ifaces(struct ctdb_context *ctdb, int argc, const char **argv)
3101 TALLOC_CTX *tmp_ctx = talloc_new(ctdb);
3103 struct ctdb_control_get_ifaces *ifaces;
3106 /* read the public ip list from this node */
3107 ret = ctdb_ctrl_get_ifaces(ctdb, TIMELIMIT(), options.pnn, tmp_ctx, &ifaces);
3109 DEBUG(DEBUG_ERR, ("Unable to get interfaces from node %u\n",
3111 talloc_free(tmp_ctx);
3115 if (options.machinereadable){
3116 printm(":Name:LinkStatus:References:\n");
3118 printf("Interfaces on node %u\n", options.pnn);
3121 for (i=0; i<ifaces->num; i++) {
3122 if (options.machinereadable){
3123 printm(":%s:%s:%u:\n",
3124 ifaces->ifaces[i].name,
3125 ifaces->ifaces[i].link_state?"1":"0",
3126 (unsigned int)ifaces->ifaces[i].references);
3128 printf("name:%s link:%s references:%u\n",
3129 ifaces->ifaces[i].name,
3130 ifaces->ifaces[i].link_state?"up":"down",
3131 (unsigned int)ifaces->ifaces[i].references);
3135 talloc_free(tmp_ctx);
3141 set link status of an interface
3143 static int control_setifacelink(struct ctdb_context *ctdb, int argc, const char **argv)
3146 TALLOC_CTX *tmp_ctx = talloc_new(ctdb);
3147 struct ctdb_control_iface_info info;
3155 if (strlen(argv[0]) > CTDB_IFACE_SIZE) {
3156 DEBUG(DEBUG_ERR, ("interfaces name '%s' too long\n",
3158 talloc_free(tmp_ctx);
3161 strcpy(info.name, argv[0]);
3163 if (strcmp(argv[1], "up") == 0) {
3164 info.link_state = 1;
3165 } else if (strcmp(argv[1], "down") == 0) {
3166 info.link_state = 0;
3168 DEBUG(DEBUG_ERR, ("link state invalid '%s' should be 'up' or 'down'\n",
3170 talloc_free(tmp_ctx);
3174 /* read the public ip list from this node */
3175 ret = ctdb_ctrl_set_iface_link(ctdb, TIMELIMIT(), options.pnn,
3178 DEBUG(DEBUG_ERR, ("Unable to set link state for interfaces %s node %u\n",
3179 argv[0], options.pnn));
3180 talloc_free(tmp_ctx);
3184 talloc_free(tmp_ctx);
3189 display pid of a ctdb daemon
3191 static int control_getpid(struct ctdb_context *ctdb, int argc, const char **argv)
3196 ret = ctdb_ctrl_getpid(ctdb, TIMELIMIT(), options.pnn, &pid);
3198 DEBUG(DEBUG_ERR, ("Unable to get daemon pid from node %u\n", options.pnn));
3201 printf("Pid:%d\n", pid);
3206 typedef bool update_flags_handler_t(struct ctdb_context *ctdb, void *data);
3208 static int update_flags_and_ipreallocate(struct ctdb_context *ctdb,
3210 update_flags_handler_t handler,
3215 struct ctdb_node_map *nodemap = NULL;
3219 /* Check if the node is already in the desired state */
3220 ret = ctdb_ctrl_getnodemap(ctdb, TIMELIMIT(), CTDB_CURRENT_NODE, ctdb, &nodemap);
3222 DEBUG(DEBUG_ERR, ("Unable to get nodemap from local node\n"));
3225 flag_is_set = nodemap->nodes[options.pnn].flags & flag;
3226 if (set_flag == flag_is_set) {
3227 DEBUG(DEBUG_NOTICE, ("Node %d is %s %s\n", options.pnn,
3228 (set_flag ? "already" : "not"), desc));
3233 if (!handler(ctdb, data)) {
3234 DEBUG(DEBUG_WARNING,
3235 ("Failed to send control to set state %s on node %u, try again\n",
3236 desc, options.pnn));
3241 /* Read the nodemap and verify the change took effect.
3242 * Even if the above control/hanlder timed out then it
3243 * could still have worked!
3245 ret = ctdb_ctrl_getnodemap(ctdb, TIMELIMIT(), CTDB_CURRENT_NODE,
3248 DEBUG(DEBUG_WARNING,
3249 ("Unable to get nodemap from local node, try again\n"));
3251 flag_is_set = nodemap->nodes[options.pnn].flags & flag;
3252 } while (nodemap == NULL || (set_flag != flag_is_set));
3254 return ipreallocate(ctdb);
3257 /* Administratively disable a node */
3258 static bool update_flags_disabled(struct ctdb_context *ctdb, void *data)
3262 ret = ctdb_ctrl_modflags(ctdb, TIMELIMIT(), options.pnn,
3263 NODE_FLAGS_PERMANENTLY_DISABLED, 0);
3267 static int control_disable(struct ctdb_context *ctdb, int argc, const char **argv)
3269 return update_flags_and_ipreallocate(ctdb, NULL,
3270 update_flags_disabled,
3271 NODE_FLAGS_PERMANENTLY_DISABLED,
3273 true /* set_flag*/);
3276 /* Administratively re-enable a node */
3277 static bool update_flags_not_disabled(struct ctdb_context *ctdb, void *data)
3281 ret = ctdb_ctrl_modflags(ctdb, TIMELIMIT(), options.pnn,
3282 0, NODE_FLAGS_PERMANENTLY_DISABLED);
3286 static int control_enable(struct ctdb_context *ctdb, int argc, const char **argv)
3288 return update_flags_and_ipreallocate(ctdb, NULL,
3289 update_flags_not_disabled,
3290 NODE_FLAGS_PERMANENTLY_DISABLED,
3292 false /* set_flag*/);
3296 static bool update_flags_stopped(struct ctdb_context *ctdb, void *data)
3300 ret = ctdb_ctrl_stop_node(ctdb, TIMELIMIT(), options.pnn);
3305 static int control_stop(struct ctdb_context *ctdb, int argc, const char **argv)
3307 return update_flags_and_ipreallocate(ctdb, NULL,
3308 update_flags_stopped,
3311 true /* set_flag*/);
3314 /* Continue a stopped node */
3315 static bool update_flags_not_stopped(struct ctdb_context *ctdb, void *data)
3319 ret = ctdb_ctrl_continue_node(ctdb, TIMELIMIT(), options.pnn);
3324 static int control_continue(struct ctdb_context *ctdb, int argc, const char **argv)
3326 return update_flags_and_ipreallocate(ctdb, NULL,
3327 update_flags_not_stopped,
3330 false /* set_flag */);
3333 static uint32_t get_generation(struct ctdb_context *ctdb)
3335 TALLOC_CTX *tmp_ctx = talloc_new(ctdb);
3336 struct ctdb_vnn_map *vnnmap=NULL;
3338 uint32_t generation;
3340 /* wait until the recmaster is not in recovery mode */
3342 uint32_t recmode, recmaster;
3344 if (vnnmap != NULL) {
3345 talloc_free(vnnmap);
3349 /* get the recmaster */
3350 ret = ctdb_ctrl_getrecmaster(ctdb, tmp_ctx, TIMELIMIT(), CTDB_CURRENT_NODE, &recmaster);
3352 DEBUG(DEBUG_ERR, ("Unable to get recmaster from node %u\n", options.pnn));
3353 talloc_free(tmp_ctx);
3357 /* get recovery mode */
3358 ret = ctdb_ctrl_getrecmode(ctdb, tmp_ctx, TIMELIMIT(), recmaster, &recmode);
3360 DEBUG(DEBUG_ERR, ("Unable to get recmode from node %u\n", options.pnn));
3361 talloc_free(tmp_ctx);
3365 /* get the current generation number */
3366 ret = ctdb_ctrl_getvnnmap(ctdb, TIMELIMIT(), recmaster, tmp_ctx, &vnnmap);
3368 DEBUG(DEBUG_ERR, ("Unable to get vnnmap from recmaster (%u)\n", recmaster));
3369 talloc_free(tmp_ctx);
3373 if ((recmode == CTDB_RECOVERY_NORMAL) && (vnnmap->generation != 1)) {
3374 generation = vnnmap->generation;
3375 talloc_free(tmp_ctx);
3383 static bool update_state_banned(struct ctdb_context *ctdb, void *data)
3385 struct ctdb_ban_time *bantime = (struct ctdb_ban_time *)data;
3388 ret = ctdb_ctrl_set_ban(ctdb, TIMELIMIT(), options.pnn, bantime);
3393 static int control_ban(struct ctdb_context *ctdb, int argc, const char **argv)
3395 struct ctdb_ban_time bantime;
3401 bantime.pnn = options.pnn;
3402 bantime.time = strtoul(argv[0], NULL, 0);
3404 if (bantime.time == 0) {
3405 DEBUG(DEBUG_ERR, ("Invalid ban time specified - must be >0\n"));
3409 return update_flags_and_ipreallocate(ctdb, &bantime,
3410 update_state_banned,
3413 true /* set_flag*/);
3418 static int control_unban(struct ctdb_context *ctdb, int argc, const char **argv)
3420 struct ctdb_ban_time bantime;
3422 bantime.pnn = options.pnn;
3425 return update_flags_and_ipreallocate(ctdb, &bantime,
3426 update_state_banned,
3429 false /* set_flag*/);
3433 show ban information for a node
3435 static int control_showban(struct ctdb_context *ctdb, int argc, const char **argv)
3438 struct ctdb_node_map *nodemap=NULL;
3439 struct ctdb_ban_time *bantime;
3441 /* verify the node exists */
3442 ret = ctdb_ctrl_getnodemap(ctdb, TIMELIMIT(), CTDB_CURRENT_NODE, ctdb, &nodemap);
3444 DEBUG(DEBUG_ERR, ("Unable to get nodemap from local node\n"));
3448 ret = ctdb_ctrl_get_ban(ctdb, TIMELIMIT(), options.pnn, ctdb, &bantime);
3450 DEBUG(DEBUG_ERR,("Showing ban info for node %d failed.\n", options.pnn));
3454 if (bantime->time == 0) {
3455 printf("Node %u is not banned\n", bantime->pnn);
3457 printf("Node %u is banned, %d seconds remaining\n",
3458 bantime->pnn, bantime->time);
3467 static int control_shutdown(struct ctdb_context *ctdb, int argc, const char **argv)
3471 ret = ctdb_ctrl_shutdown(ctdb, TIMELIMIT(), options.pnn);
3473 DEBUG(DEBUG_ERR, ("Unable to shutdown node %u\n", options.pnn));
3483 static int control_recover(struct ctdb_context *ctdb, int argc, const char **argv)
3486 uint32_t generation, next_generation;
3488 /* record the current generation number */
3489 generation = get_generation(ctdb);
3491 ret = ctdb_ctrl_setrecmode(ctdb, TIMELIMIT(), options.pnn, CTDB_RECOVERY_ACTIVE);
3493 DEBUG(DEBUG_ERR, ("Unable to set recovery mode\n"));
3497 /* wait until we are in a new generation */
3499 next_generation = get_generation(ctdb);
3500 if (next_generation != generation) {
3511 display monitoring mode of a remote node
3513 static int control_getmonmode(struct ctdb_context *ctdb, int argc, const char **argv)
3518 ret = ctdb_ctrl_getmonmode(ctdb, TIMELIMIT(), options.pnn, &monmode);
3520 DEBUG(DEBUG_ERR, ("Unable to get monmode from node %u\n", options.pnn));
3523 if (!options.machinereadable){
3524 printf("Monitoring mode:%s (%d)\n",monmode==CTDB_MONITORING_ACTIVE?"ACTIVE":"DISABLED",monmode);
3527 printm(":%d:\n",monmode);
3534 display capabilities of a remote node
3536 static int control_getcapabilities(struct ctdb_context *ctdb, int argc, const char **argv)
3538 uint32_t capabilities;
3541 ret = ctdb_ctrl_getcapabilities(ctdb, TIMELIMIT(), options.pnn, &capabilities);
3543 DEBUG(DEBUG_ERR, ("Unable to get capabilities from node %u\n", options.pnn));
3547 if (!options.machinereadable){
3548 printf("RECMASTER: %s\n", (capabilities&CTDB_CAP_RECMASTER)?"YES":"NO");
3549 printf("LMASTER: %s\n", (capabilities&CTDB_CAP_LMASTER)?"YES":"NO");
3550 printf("LVS: %s\n", (capabilities&CTDB_CAP_LVS)?"YES":"NO");
3551 printf("NATGW: %s\n", (capabilities&CTDB_CAP_NATGW)?"YES":"NO");
3553 printm(":RECMASTER:LMASTER:LVS:NATGW:\n");
3554 printm(":%d:%d:%d:%d:\n",
3555 !!(capabilities&CTDB_CAP_RECMASTER),
3556 !!(capabilities&CTDB_CAP_LMASTER),
3557 !!(capabilities&CTDB_CAP_LVS),
3558 !!(capabilities&CTDB_CAP_NATGW));
3564 display lvs configuration
3567 static uint32_t lvs_exclude_flags[] = {
3568 /* Look for a nice healthy node */
3569 NODE_FLAGS_INACTIVE|NODE_FLAGS_DISABLED,
3570 /* If not found, an UNHEALTHY node will do */
3571 NODE_FLAGS_INACTIVE|NODE_FLAGS_PERMANENTLY_DISABLED,
3575 static int control_lvs(struct ctdb_context *ctdb, int argc, const char **argv)
3577 TALLOC_CTX *tmp_ctx = talloc_new(ctdb);
3578 struct ctdb_node_map *orig_nodemap=NULL;
3579 struct ctdb_node_map *nodemap;
3582 ret = ctdb_ctrl_getnodemap(ctdb, TIMELIMIT(), options.pnn,
3583 tmp_ctx, &orig_nodemap);
3585 DEBUG(DEBUG_ERR, ("Unable to get nodemap from node %u\n", options.pnn));
3586 talloc_free(tmp_ctx);
3590 nodemap = filter_nodemap_by_capabilities(ctdb, orig_nodemap,
3591 CTDB_CAP_LVS, false);
3592 if (nodemap == NULL) {
3600 for (i = 0; lvs_exclude_flags[i] != 0; i++) {
3601 struct ctdb_node_map *t =
3602 filter_nodemap_by_flags(ctdb, nodemap,
3603 lvs_exclude_flags[i]);
3610 /* At least 1 node without excluded flags */
3612 for (j = 0; j < t->num; j++) {
3613 printf("%d:%s\n", t->nodes[j].pnn,
3614 ctdb_addr_to_str(&t->nodes[j].addr));
3621 talloc_free(tmp_ctx);
3626 display who is the lvs master
3628 static int control_lvsmaster(struct ctdb_context *ctdb, int argc, const char **argv)
3630 TALLOC_CTX *tmp_ctx = talloc_new(ctdb);
3631 struct ctdb_node_map *nodemap=NULL;
3634 ret = ctdb_ctrl_getnodemap(ctdb, TIMELIMIT(), options.pnn,
3637 DEBUG(DEBUG_ERR, ("Unable to get nodemap from node %u\n", options.pnn));
3638 talloc_free(tmp_ctx);
3642 for (i = 0; lvs_exclude_flags[i] != 0; i++) {
3643 struct ctdb_node_map *t =
3644 filter_nodemap_by_flags(ctdb, nodemap,
3645 lvs_exclude_flags[i]);
3652 struct ctdb_node_map *n;
3653 n = filter_nodemap_by_capabilities(ctdb,
3664 if (options.machinereadable) {
3665 printm("%d\n", n->nodes[0].pnn);
3667 printf("Node %d is LVS master\n", n->nodes[0].pnn);
3675 printf("There is no LVS master\n");
3678 talloc_free(tmp_ctx);
3683 disable monitoring on a node
3685 static int control_disable_monmode(struct ctdb_context *ctdb, int argc, const char **argv)
3690 ret = ctdb_ctrl_disable_monmode(ctdb, TIMELIMIT(), options.pnn);
3692 DEBUG(DEBUG_ERR, ("Unable to disable monmode on node %u\n", options.pnn));
3695 printf("Monitoring mode:%s\n","DISABLED");
3701 enable monitoring on a node
3703 static int control_enable_monmode(struct ctdb_context *ctdb, int argc, const char **argv)
3708 ret = ctdb_ctrl_enable_monmode(ctdb, TIMELIMIT(), options.pnn);
3710 DEBUG(DEBUG_ERR, ("Unable to enable monmode on node %u\n", options.pnn));
3713 printf("Monitoring mode:%s\n","ACTIVE");
3719 display remote list of keys/data for a db
3721 static int control_catdb(struct ctdb_context *ctdb, int argc, const char **argv)
3723 const char *db_name;
3724 struct ctdb_db_context *ctdb_db;
3726 struct ctdb_dump_db_context c;
3733 if (!db_exists(ctdb, argv[0], NULL, &db_name, &flags)) {
3737 ctdb_db = ctdb_attach(ctdb, TIMELIMIT(), db_name, flags & CTDB_DB_FLAGS_PERSISTENT, 0);
3738 if (ctdb_db == NULL) {
3739 DEBUG(DEBUG_ERR,("Unable to attach to database '%s'\n", db_name));
3743 if (options.printlmaster) {
3744 ret = ctdb_ctrl_getvnnmap(ctdb, TIMELIMIT(), options.pnn,
3745 ctdb, &ctdb->vnn_map);
3747 DEBUG(DEBUG_ERR, ("Unable to get vnnmap from node %u\n",
3755 c.printemptyrecords = (bool)options.printemptyrecords;
3756 c.printdatasize = (bool)options.printdatasize;
3757 c.printlmaster = (bool)options.printlmaster;
3758 c.printhash = (bool)options.printhash;
3759 c.printrecordflags = (bool)options.printrecordflags;
3761 /* traverse and dump the cluster tdb */
3762 ret = ctdb_dump_db(ctdb_db, &c);
3764 DEBUG(DEBUG_ERR, ("Unable to dump database\n"));
3765 DEBUG(DEBUG_ERR, ("Maybe try 'ctdb getdbstatus %s'"
3766 " and 'ctdb getvar AllowUnhealthyDBRead'\n",
3770 talloc_free(ctdb_db);
3772 printf("Dumped %d records\n", ret);
3776 struct cattdb_data {
3777 struct ctdb_context *ctdb;
3781 static int cattdb_traverse(struct tdb_context *tdb, TDB_DATA key, TDB_DATA data, void *private_data)
3783 struct cattdb_data *d = private_data;
3784 struct ctdb_dump_db_context c;
3790 c.printemptyrecords = (bool)options.printemptyrecords;
3791 c.printdatasize = (bool)options.printdatasize;
3792 c.printlmaster = false;
3793 c.printhash = (bool)options.printhash;
3794 c.printrecordflags = true;
3796 return ctdb_dumpdb_record(d->ctdb, key, data, &c);
3800 cat the local tdb database using same format as catdb
3802 static int control_cattdb(struct ctdb_context *ctdb, int argc, const char **argv)
3804 const char *db_name;
3805 struct ctdb_db_context *ctdb_db;
3806 struct cattdb_data d;
3813 if (!db_exists(ctdb, argv[0], NULL, &db_name, &flags)) {
3817 ctdb_db = ctdb_attach(ctdb, TIMELIMIT(), db_name, flags & CTDB_DB_FLAGS_PERSISTENT, 0);
3818 if (ctdb_db == NULL) {
3819 DEBUG(DEBUG_ERR,("Unable to attach to database '%s'\n", db_name));
3823 /* traverse the local tdb */
3826 if (tdb_traverse_read(ctdb_db->ltdb->tdb, cattdb_traverse, &d) == -1) {
3827 printf("Failed to cattdb data\n");
3830 talloc_free(ctdb_db);
3832 printf("Dumped %d records\n", d.count);
3837 display the content of a database key
3839 static int control_readkey(struct ctdb_context *ctdb, int argc, const char **argv)
3841 const char *db_name;
3842 struct ctdb_db_context *ctdb_db;
3843 struct ctdb_record_handle *h;
3844 TALLOC_CTX *tmp_ctx = talloc_new(ctdb);
3852 if (!db_exists(ctdb, argv[0], NULL, &db_name, &flags)) {
3856 ctdb_db = ctdb_attach(ctdb, TIMELIMIT(), db_name, flags & CTDB_DB_FLAGS_PERSISTENT, 0);
3857 if (ctdb_db == NULL) {
3858 DEBUG(DEBUG_ERR,("Unable to attach to database '%s'\n", db_name));
3862 key.dptr = discard_const(argv[1]);
3863 key.dsize = strlen((char *)key.dptr);
3865 h = ctdb_fetch_lock(ctdb_db, tmp_ctx, key, &data);
3867 printf("Failed to fetch record '%s' on node %d\n",
3868 (const char *)key.dptr, ctdb_get_pnn(ctdb));
3869 talloc_free(tmp_ctx);
3873 printf("Data: size:%d ptr:[%.*s]\n", (int)data.dsize, (int)data.dsize, data.dptr);
3875 talloc_free(tmp_ctx);
3876 talloc_free(ctdb_db);
3881 display the content of a database key
3883 static int control_writekey(struct ctdb_context *ctdb, int argc, const char **argv)
3885 const char *db_name;
3886 struct ctdb_db_context *ctdb_db;
3887 struct ctdb_record_handle *h;
3888 TALLOC_CTX *tmp_ctx = talloc_new(ctdb);
3896 if (!db_exists(ctdb, argv[0], NULL, &db_name, &flags)) {
3900 ctdb_db = ctdb_attach(ctdb, TIMELIMIT(), db_name, flags & CTDB_DB_FLAGS_PERSISTENT, 0);
3901 if (ctdb_db == NULL) {
3902 DEBUG(DEBUG_ERR,("Unable to attach to database '%s'\n", db_name));
3906 key.dptr = discard_const(argv[1]);
3907 key.dsize = strlen((char *)key.dptr);
3909 h = ctdb_fetch_lock(ctdb_db, tmp_ctx, key, &data);
3911 printf("Failed to fetch record '%s' on node %d\n",
3912 (const char *)key.dptr, ctdb_get_pnn(ctdb));
3913 talloc_free(tmp_ctx);
3917 data.dptr = discard_const(argv[2]);
3918 data.dsize = strlen((char *)data.dptr);
3920 if (ctdb_record_store(h, data) != 0) {
3921 printf("Failed to store record\n");
3925 talloc_free(tmp_ctx);
3926 talloc_free(ctdb_db);
3931 fetch a record from a persistent database
3933 static int control_pfetch(struct ctdb_context *ctdb, int argc, const char **argv)
3935 const char *db_name;
3936 struct ctdb_db_context *ctdb_db;
3937 TALLOC_CTX *tmp_ctx = talloc_new(ctdb);
3938 struct ctdb_transaction_handle *h;
3945 talloc_free(tmp_ctx);
3949 if (!db_exists(ctdb, argv[0], NULL, &db_name, &flags)) {
3950 talloc_free(tmp_ctx);
3954 persistent = flags & CTDB_DB_FLAGS_PERSISTENT;
3956 DEBUG(DEBUG_ERR,("Database '%s' is not persistent\n", db_name));
3957 talloc_free(tmp_ctx);
3961 ctdb_db = ctdb_attach(ctdb, TIMELIMIT(), db_name, persistent, 0);
3962 if (ctdb_db == NULL) {
3963 DEBUG(DEBUG_ERR,("Unable to attach to database '%s'\n", db_name));
3964 talloc_free(tmp_ctx);
3968 h = ctdb_transaction_start(ctdb_db, tmp_ctx);
3970 DEBUG(DEBUG_ERR,("Failed to start transaction on database %s\n", db_name));
3971 talloc_free(tmp_ctx);
3975 key.dptr = discard_const(argv[1]);
3976 key.dsize = strlen(argv[1]);
3977 ret = ctdb_transaction_fetch(h, tmp_ctx, key, &data);
3979 DEBUG(DEBUG_ERR,("Failed to fetch record\n"));
3980 talloc_free(tmp_ctx);
3984 if (data.dsize == 0 || data.dptr == NULL) {
3985 DEBUG(DEBUG_ERR,("Record is empty\n"));
3986 talloc_free(tmp_ctx);
3991 fd = open(argv[2], O_WRONLY|O_CREAT|O_TRUNC, 0600);
3993 DEBUG(DEBUG_ERR,("Failed to open output file %s\n", argv[2]));
3994 talloc_free(tmp_ctx);
3997 sys_write(fd, data.dptr, data.dsize);
4000 sys_write(1, data.dptr, data.dsize);
4003 /* abort the transaction */
4007 talloc_free(tmp_ctx);
4012 fetch a record from a tdb-file
4014 static int control_tfetch(struct ctdb_context *ctdb, int argc, const char **argv)
4016 const char *tdb_file;
4019 TALLOC_CTX *tmp_ctx = talloc_new(NULL);
4028 tdb = tdb_open(tdb_file, 0, 0, O_RDONLY, 0);
4030 printf("Failed to open TDB file %s\n", tdb_file);
4034 if (!strncmp(argv[1], "0x", 2)) {
4035 key = hextodata(tmp_ctx, argv[1] + 2);
4036 if (key.dsize == 0) {
4037 printf("Failed to convert \"%s\" into a TDB_DATA\n", argv[1]);
4041 key.dptr = discard_const(argv[1]);
4042 key.dsize = strlen(argv[1]);
4045 data = tdb_fetch(tdb, key);
4046 if (data.dptr == NULL || data.dsize < sizeof(struct ctdb_ltdb_header)) {
4047 printf("Failed to read record %s from tdb %s\n", argv[1], tdb_file);
4055 fd = open(argv[2], O_WRONLY|O_CREAT|O_TRUNC, 0600);
4057 printf("Failed to open output file %s\n", argv[2]);
4060 if (options.verbose){
4061 sys_write(fd, data.dptr, data.dsize);
4063 sys_write(fd, data.dptr+sizeof(struct ctdb_ltdb_header), data.dsize-sizeof(struct ctdb_ltdb_header));
4067 if (options.verbose){
4068 sys_write(1, data.dptr, data.dsize);
4070 sys_write(1, data.dptr+sizeof(struct ctdb_ltdb_header), data.dsize-sizeof(struct ctdb_ltdb_header));
4074 talloc_free(tmp_ctx);
4079 store a record and header to a tdb-file
4081 static int control_tstore(struct ctdb_context *ctdb, int argc, const char **argv)
4083 const char *tdb_file;
4085 TDB_DATA key, value, data;
4086 TALLOC_CTX *tmp_ctx = talloc_new(NULL);
4087 struct ctdb_ltdb_header header;
4095 tdb = tdb_open(tdb_file, 0, 0, O_RDWR, 0);
4097 printf("Failed to open TDB file %s\n", tdb_file);
4101 if (!strncmp(argv[1], "0x", 2)) {
4102 key = hextodata(tmp_ctx, argv[1] + 2);
4103 if (key.dsize == 0) {
4104 printf("Failed to convert \"%s\" into a TDB_DATA\n", argv[1]);
4108 key.dptr = discard_const(argv[1]);
4109 key.dsize = strlen(argv[1]);
4112 if (!strncmp(argv[2], "0x", 2)) {
4113 value = hextodata(tmp_ctx, argv[2] + 2);
4114 if (value.dsize == 0) {
4115 printf("Failed to convert \"%s\" into a TDB_DATA\n", argv[2]);
4119 value.dptr = discard_const(argv[2]);
4120 value.dsize = strlen(argv[2]);
4123 ZERO_STRUCT(header);
4125 header.rsn = atoll(argv[3]);
4128 header.dmaster = atoi(argv[4]);
4131 header.flags = atoi(argv[5]);
4134 data.dsize = sizeof(struct ctdb_ltdb_header) + value.dsize;
4135 data.dptr = talloc_size(tmp_ctx, data.dsize);
4136 if (data.dptr == NULL) {
4137 printf("Failed to allocate header+value\n");
4141 *(struct ctdb_ltdb_header *)data.dptr = header;
4142 memcpy(data.dptr + sizeof(struct ctdb_ltdb_header), value.dptr, value.dsize);
4144 if (tdb_store(tdb, key, data, TDB_REPLACE) != 0) {
4145 printf("Failed to write record %s to tdb %s\n", argv[1], tdb_file);
4152 talloc_free(tmp_ctx);
4157 write a record to a persistent database
4159 static int control_pstore(struct ctdb_context *ctdb, int argc, const char **argv)
4161 const char *db_name;
4162 struct ctdb_db_context *ctdb_db;
4163 TALLOC_CTX *tmp_ctx = talloc_new(ctdb);
4164 struct ctdb_transaction_handle *h;
4170 talloc_free(tmp_ctx);
4174 fd = open(argv[2], O_RDONLY);
4176 DEBUG(DEBUG_ERR,("Failed to open file containing record data : %s %s\n", argv[2], strerror(errno)));
4177 talloc_free(tmp_ctx);
4181 ret = fstat(fd, &st);
4183 DEBUG(DEBUG_ERR,("fstat of file %s failed: %s\n", argv[2], strerror(errno)));
4185 talloc_free(tmp_ctx);
4189 if (!S_ISREG(st.st_mode)) {
4190 DEBUG(DEBUG_ERR,("Not a regular file %s\n", argv[2]));
4192 talloc_free(tmp_ctx);
4196 data.dsize = st.st_size;
4197 if (data.dsize == 0) {
4200 data.dptr = talloc_size(tmp_ctx, data.dsize);
4201 if (data.dptr == NULL) {
4202 DEBUG(DEBUG_ERR,("Failed to talloc %d of memory to store record data\n", (int)data.dsize));
4204 talloc_free(tmp_ctx);
4207 ret = sys_read(fd, data.dptr, data.dsize);
4208 if (ret != data.dsize) {
4209 DEBUG(DEBUG_ERR,("Failed to read %d bytes of record data\n", (int)data.dsize));
4211 talloc_free(tmp_ctx);
4220 ctdb_db = ctdb_attach(ctdb, TIMELIMIT(), db_name, true, 0);
4221 if (ctdb_db == NULL) {
4222 DEBUG(DEBUG_ERR,("Unable to attach to database '%s'\n", db_name));
4223 talloc_free(tmp_ctx);
4227 h = ctdb_transaction_start(ctdb_db, tmp_ctx);
4229 DEBUG(DEBUG_ERR,("Failed to start transaction on database %s\n", db_name));
4230 talloc_free(tmp_ctx);
4234 if (!strncmp(argv[1], "0x", 2)) {
4235 key = hextodata(tmp_ctx, argv[1] + 2);
4236 if (key.dsize == 0) {
4237 printf("Failed to convert \"%s\" into a TDB_DATA\n", argv[1]);
4241 key.dptr = discard_const(argv[1]);
4242 key.dsize = strlen(argv[1]);
4245 ret = ctdb_transaction_store(h, key, data);
4247 DEBUG(DEBUG_ERR,("Failed to store record\n"));
4248 talloc_free(tmp_ctx);
4252 ret = ctdb_transaction_commit(h);
4254 DEBUG(DEBUG_ERR,("Failed to commit transaction\n"));
4255 talloc_free(tmp_ctx);
4260 talloc_free(tmp_ctx);
4265 * delete a record from a persistent database
4267 static int control_pdelete(struct ctdb_context *ctdb, int argc, const char **argv)
4269 const char *db_name;
4270 struct ctdb_db_context *ctdb_db;
4271 TALLOC_CTX *tmp_ctx = talloc_new(ctdb);
4272 struct ctdb_transaction_handle *h;
4279 talloc_free(tmp_ctx);
4283 if (!db_exists(ctdb, argv[0], NULL, &db_name, &flags)) {
4284 talloc_free(tmp_ctx);
4288 persistent = flags & CTDB_DB_FLAGS_PERSISTENT;
4290 DEBUG(DEBUG_ERR, ("Database '%s' is not persistent\n", db_name));
4291 talloc_free(tmp_ctx);
4295 ctdb_db = ctdb_attach(ctdb, TIMELIMIT(), db_name, persistent, 0);
4296 if (ctdb_db == NULL) {
4297 DEBUG(DEBUG_ERR, ("Unable to attach to database '%s'\n", db_name));
4298 talloc_free(tmp_ctx);
4302 h = ctdb_transaction_start(ctdb_db, tmp_ctx);
4304 DEBUG(DEBUG_ERR, ("Failed to start transaction on database %s\n", db_name));
4305 talloc_free(tmp_ctx);
4309 key.dptr = discard_const(argv[1]);
4310 key.dsize = strlen(argv[1]);
4311 ret = ctdb_transaction_store(h, key, tdb_null);
4313 DEBUG(DEBUG_ERR, ("Failed to delete record\n"));
4314 talloc_free(tmp_ctx);
4318 ret = ctdb_transaction_commit(h);
4320 DEBUG(DEBUG_ERR, ("Failed to commit transaction\n"));
4321 talloc_free(tmp_ctx);
4325 talloc_free(tmp_ctx);
4329 static const char *ptrans_parse_string(TALLOC_CTX *mem_ctx, const char *s,
4334 const char *ret; /* Next byte after successfully parsed value */
4336 /* Error, unless someone says otherwise */
4338 /* Indicates no value to parse */
4341 /* Skip whitespace */
4342 n = strspn(s, " \t");
4346 /* Quoted ASCII string - no wide characters! */
4348 n = strcspn(t, "\"");
4352 data->dptr = talloc_memdup(mem_ctx, t, n);
4353 CTDB_NOMEM_ABORT(data->dptr);
4357 DEBUG(DEBUG_WARNING,("Unmatched \" in input %s\n", s));
4360 DEBUG(DEBUG_WARNING,("Unsupported input format in %s\n", s));
4366 static bool ptrans_get_key_value(TALLOC_CTX *mem_ctx, FILE *file,
4367 TDB_DATA *key, TDB_DATA *value)
4369 char line [1024]; /* FIXME: make this more flexible? */
4373 ptr = fgets(line, sizeof(line), file);
4380 t = ptrans_parse_string(mem_ctx, line, key);
4381 if (t == NULL || key->dptr == NULL) {
4382 /* Line Ignored but not EOF */
4387 t = ptrans_parse_string(mem_ctx, t, value);
4389 /* Line Ignored but not EOF */
4390 talloc_free(key->dptr);
4399 * Update a persistent database as per file/stdin
4401 static int control_ptrans(struct ctdb_context *ctdb,
4402 int argc, const char **argv)
4404 const char *db_name;
4405 struct ctdb_db_context *ctdb_db;
4406 TALLOC_CTX *tmp_ctx = talloc_new(ctdb);
4407 struct ctdb_transaction_handle *h;
4408 TDB_DATA key, value;
4413 talloc_free(tmp_ctx);
4419 file = fopen(argv[1], "r");
4421 DEBUG(DEBUG_ERR,("Unable to open file for reading '%s'\n", argv[1]));
4422 talloc_free(tmp_ctx);
4429 ctdb_db = ctdb_attach(ctdb, TIMELIMIT(), db_name, true, 0);
4430 if (ctdb_db == NULL) {
4431 DEBUG(DEBUG_ERR,("Unable to attach to database '%s'\n", db_name));
4435 h = ctdb_transaction_start(ctdb_db, tmp_ctx);
4437 DEBUG(DEBUG_ERR,("Failed to start transaction on database %s\n", db_name));
4441 while (ptrans_get_key_value(tmp_ctx, file, &key, &value)) {
4442 if (key.dsize != 0) {
4443 ret = ctdb_transaction_store(h, key, value);
4444 /* Minimise memory use */
4445 talloc_free(key.dptr);
4446 if (value.dptr != NULL) {
4447 talloc_free(value.dptr);
4450 DEBUG(DEBUG_ERR,("Failed to store record\n"));
4451 ctdb_transaction_cancel(h);
4457 ret = ctdb_transaction_commit(h);
4459 DEBUG(DEBUG_ERR,("Failed to commit transaction\n"));
4463 if (file != stdin) {
4466 talloc_free(tmp_ctx);
4470 if (file != stdin) {
4474 talloc_free(tmp_ctx);
4479 check if a service is bound to a port or not
4481 static int control_chktcpport(struct ctdb_context *ctdb, int argc, const char **argv)
4486 struct sockaddr_in sin;
4489 printf("Use: ctdb chktcport <port>\n");
4493 port = atoi(argv[0]);
4495 s = socket(PF_INET, SOCK_STREAM, IPPROTO_TCP);
4497 printf("Failed to open local socket\n");
4501 v = fcntl(s, F_GETFL, 0);
4502 if (v == -1 || fcntl(s, F_SETFL, v | O_NONBLOCK) != 0) {
4503 printf("Unable to set socket non-blocking: %s\n", strerror(errno));
4506 bzero(&sin, sizeof(sin));
4507 sin.sin_family = PF_INET;
4508 sin.sin_port = htons(port);
4509 ret = bind(s, (struct sockaddr *)&sin, sizeof(sin));
4512 printf("Failed to bind to local socket: %d %s\n", errno, strerror(errno));
4520 /* Reload public IPs on a specified nodes */
4521 static int control_reloadips(struct ctdb_context *ctdb, int argc, const char **argv)
4523 TALLOC_CTX *tmp_ctx = talloc_new(ctdb);
4529 assert_single_node_only();
4535 /* Determine the nodes where IPs need to be reloaded */
4536 if (!parse_nodestring(ctdb, tmp_ctx, argc == 1 ? argv[0] : NULL,
4537 options.pnn, true, &nodes, &pnn_mode)) {
4543 /* Disable takeover runs on all connected nodes. A reply
4544 * indicating success is needed from each node so all nodes
4545 * will need to be active. This will retry until maxruntime
4546 * is exceeded, hence no error handling.
4548 * A check could be added to not allow reloading of IPs when
4549 * there are disconnected nodes. However, this should
4550 * probably be left up to the administrator.
4552 timeout = LONGTIMEOUT;
4553 srvid_broadcast(ctdb, CTDB_SRVID_DISABLE_TAKEOVER_RUNS, &timeout,
4554 "Disable takeover runs", true);
4556 /* Now tell all the desired nodes to reload their public IPs.
4557 * Keep trying this until it succeeds. This assumes all
4558 * failures are transient, which might not be true...
4560 if (ctdb_client_async_control(ctdb, CTDB_CONTROL_RELOAD_PUBLIC_IPS,
4561 nodes, 0, LONGTIMELIMIT(),
4563 NULL, NULL, NULL) != 0) {
4565 ("Unable to reload IPs on some nodes, try again.\n"));
4569 /* It isn't strictly necessary to wait until takeover runs are
4570 * re-enabled but doing so can't hurt.
4573 srvid_broadcast(ctdb, CTDB_SRVID_DISABLE_TAKEOVER_RUNS, &timeout,
4574 "Enable takeover runs", true);
4580 talloc_free(tmp_ctx);
4585 display a list of the databases on a remote ctdb
4587 static int control_getdbmap(struct ctdb_context *ctdb, int argc, const char **argv)
4590 struct ctdb_dbid_map *dbmap=NULL;
4592 ret = ctdb_ctrl_getdbmap(ctdb, TIMELIMIT(), options.pnn, ctdb, &dbmap);
4594 DEBUG(DEBUG_ERR, ("Unable to get dbids from node %u\n", options.pnn));
4598 if(options.machinereadable){
4599 printm(":ID:Name:Path:Persistent:Sticky:Unhealthy:ReadOnly:\n");
4600 for(i=0;i<dbmap->num;i++){
4601 const char *path = NULL;
4602 const char *name = NULL;
4603 const char *health = NULL;
4608 ctdb_ctrl_getdbpath(ctdb, TIMELIMIT(), options.pnn,
4609 dbmap->dbs[i].dbid, ctdb, &path);
4610 ctdb_ctrl_getdbname(ctdb, TIMELIMIT(), options.pnn,
4611 dbmap->dbs[i].dbid, ctdb, &name);
4612 ctdb_ctrl_getdbhealth(ctdb, TIMELIMIT(), options.pnn,
4613 dbmap->dbs[i].dbid, ctdb, &health);
4614 persistent = dbmap->dbs[i].flags & CTDB_DB_FLAGS_PERSISTENT;
4615 readonly = dbmap->dbs[i].flags & CTDB_DB_FLAGS_READONLY;
4616 sticky = dbmap->dbs[i].flags & CTDB_DB_FLAGS_STICKY;
4617 printm(":0x%08X:%s:%s:%d:%d:%d:%d:\n",
4618 dbmap->dbs[i].dbid, name, path,
4619 !!(persistent), !!(sticky),
4620 !!(health), !!(readonly));
4625 printf("Number of databases:%d\n", dbmap->num);
4626 for(i=0;i<dbmap->num;i++){
4627 const char *path = NULL;
4628 const char *name = NULL;
4629 const char *health = NULL;
4634 ctdb_ctrl_getdbpath(ctdb, TIMELIMIT(), options.pnn, dbmap->dbs[i].dbid, ctdb, &path);
4635 ctdb_ctrl_getdbname(ctdb, TIMELIMIT(), options.pnn, dbmap->dbs[i].dbid, ctdb, &name);
4636 ctdb_ctrl_getdbhealth(ctdb, TIMELIMIT(), options.pnn, dbmap->dbs[i].dbid, ctdb, &health);
4637 persistent = dbmap->dbs[i].flags & CTDB_DB_FLAGS_PERSISTENT;
4638 readonly = dbmap->dbs[i].flags & CTDB_DB_FLAGS_READONLY;
4639 sticky = dbmap->dbs[i].flags & CTDB_DB_FLAGS_STICKY;
4640 printf("dbid:0x%08x name:%s path:%s%s%s%s%s\n",
4641 dbmap->dbs[i].dbid, name, path,
4642 persistent?" PERSISTENT":"",
4643 sticky?" STICKY":"",
4644 readonly?" READONLY":"",
4645 health?" UNHEALTHY":"");
4652 display the status of a database on a remote ctdb
4654 static int control_getdbstatus(struct ctdb_context *ctdb, int argc, const char **argv)
4656 const char *db_name;
4659 const char *path = NULL;
4660 const char *health = NULL;
4666 if (!db_exists(ctdb, argv[0], &db_id, &db_name, &flags)) {
4670 ctdb_ctrl_getdbpath(ctdb, TIMELIMIT(), options.pnn, db_id, ctdb, &path);
4671 ctdb_ctrl_getdbhealth(ctdb, TIMELIMIT(), options.pnn, db_id, ctdb, &health);
4672 printf("dbid: 0x%08x\nname: %s\npath: %s\nPERSISTENT: %s\nSTICKY: %s\nREADONLY: %s\nHEALTH: %s\n",
4673 db_id, db_name, path,
4674 (flags & CTDB_DB_FLAGS_PERSISTENT ? "yes" : "no"),
4675 (flags & CTDB_DB_FLAGS_STICKY ? "yes" : "no"),
4676 (flags & CTDB_DB_FLAGS_READONLY ? "yes" : "no"),
4677 (health ? health : "OK"));
4683 check if the local node is recmaster or not
4684 it will return 1 if this node is the recmaster and 0 if it is not
4685 or if the local ctdb daemon could not be contacted
4687 static int control_isnotrecmaster(struct ctdb_context *ctdb, int argc, const char **argv)
4689 uint32_t mypnn, recmaster;
4692 assert_single_node_only();
4694 mypnn = getpnn(ctdb);
4696 ret = ctdb_ctrl_getrecmaster(ctdb, ctdb, TIMELIMIT(), options.pnn, &recmaster);
4698 printf("Failed to get the recmaster\n");
4702 if (recmaster != mypnn) {
4703 printf("this node is not the recmaster\n");
4707 printf("this node is the recmaster\n");
4714 static int control_ping(struct ctdb_context *ctdb, int argc, const char **argv)
4717 struct timeval tv = timeval_current();
4718 ret = ctdb_ctrl_ping(ctdb, options.pnn);
4720 printf("Unable to get ping response from node %u\n", options.pnn);
4723 printf("response from %u time=%.6f sec (%d clients)\n",
4724 options.pnn, timeval_elapsed(&tv), ret);
4731 get a node's runstate
4733 static int control_runstate(struct ctdb_context *ctdb, int argc, const char **argv)
4736 enum ctdb_runstate runstate;
4738 ret = ctdb_ctrl_get_runstate(ctdb, TIMELIMIT(), options.pnn, &runstate);
4740 printf("Unable to get runstate response from node %u\n",
4745 enum ctdb_runstate t;
4747 for (i=0; i<argc; i++) {
4749 t = runstate_from_string(argv[i]);
4750 if (t == CTDB_RUNSTATE_UNKNOWN) {
4751 printf("Invalid run state (%s)\n", argv[i]);
4755 if (t == runstate) {
4762 printf("CTDB not in required run state (got %s)\n",
4763 runstate_to_string((enum ctdb_runstate)runstate));
4768 printf("%s\n", runstate_to_string(runstate));
4776 static int control_getvar(struct ctdb_context *ctdb, int argc, const char **argv)
4787 ret = ctdb_ctrl_get_tunable(ctdb, TIMELIMIT(), options.pnn, name, &value);
4789 DEBUG(DEBUG_ERR, ("Unable to get tunable variable '%s'\n", name));
4793 printf("%-23s = %u\n", name, value);
4800 static int control_setvar(struct ctdb_context *ctdb, int argc, const char **argv)
4811 value = strtoul(argv[1], NULL, 0);
4813 ret = ctdb_ctrl_set_tunable(ctdb, TIMELIMIT(), options.pnn, name, value);
4815 DEBUG(DEBUG_ERR, ("Unable to set tunable variable '%s'\n", name));
4819 DEBUG(DEBUG_WARNING,
4820 ("Setting obsolete tunable variable '%s'\n",
4829 static int control_listvars(struct ctdb_context *ctdb, int argc, const char **argv)
4835 ret = ctdb_ctrl_list_tunables(ctdb, TIMELIMIT(), options.pnn, ctdb, &list, &count);
4837 DEBUG(DEBUG_ERR, ("Unable to list tunable variables\n"));
4841 for (i=0;i<count;i++) {
4842 control_getvar(ctdb, 1, &list[i]);
4851 display debug level on a node
4853 static int control_getdebug(struct ctdb_context *ctdb, int argc, const char **argv)
4858 ret = ctdb_ctrl_get_debuglevel(ctdb, options.pnn, &level);
4860 DEBUG(DEBUG_ERR, ("Unable to get debuglevel response from node %u\n", options.pnn));
4863 const char *desc = get_debug_by_level(level);
4865 /* This should never happen */
4868 if (options.machinereadable){
4869 printm(":Name:Level:\n");
4870 printm(":%s:%d:\n", desc, level);
4872 printf("Node %u is at debug level %s (%d)\n",
4873 options.pnn, desc, level);
4880 display reclock file of a node
4882 static int control_getreclock(struct ctdb_context *ctdb, int argc, const char **argv)
4885 const char *reclock;
4887 ret = ctdb_ctrl_getreclock(ctdb, TIMELIMIT(), options.pnn, ctdb, &reclock);
4889 DEBUG(DEBUG_ERR, ("Unable to get reclock file from node %u\n", options.pnn));
4892 if (options.machinereadable){
4893 if (reclock != NULL) {
4894 printm("%s", reclock);
4897 if (reclock == NULL) {
4898 printf("No reclock file used.\n");
4900 printf("Reclock file:%s\n", reclock);
4908 set the reclock file of a node
4910 static int control_setreclock(struct ctdb_context *ctdb, int argc, const char **argv)
4913 const char *reclock;
4917 } else if (argc == 1) {
4923 ret = ctdb_ctrl_setreclock(ctdb, TIMELIMIT(), options.pnn, reclock);
4925 DEBUG(DEBUG_ERR, ("Unable to get reclock file from node %u\n", options.pnn));
4932 set the natgw state on/off
4934 static int control_setnatgwstate(struct ctdb_context *ctdb, int argc, const char **argv)
4937 uint32_t natgwstate;
4943 if (!strcmp(argv[0], "on")) {
4945 } else if (!strcmp(argv[0], "off")) {
4951 ret = ctdb_ctrl_setnatgwstate(ctdb, TIMELIMIT(), options.pnn, natgwstate);
4953 DEBUG(DEBUG_ERR, ("Unable to set the natgw state for node %u\n", options.pnn));
4961 set the lmaster role on/off
4963 static int control_setlmasterrole(struct ctdb_context *ctdb, int argc, const char **argv)
4966 uint32_t lmasterrole;
4972 if (!strcmp(argv[0], "on")) {
4974 } else if (!strcmp(argv[0], "off")) {
4980 ret = ctdb_ctrl_setlmasterrole(ctdb, TIMELIMIT(), options.pnn, lmasterrole);
4982 DEBUG(DEBUG_ERR, ("Unable to set the lmaster role for node %u\n", options.pnn));
4990 set the recmaster role on/off
4992 static int control_setrecmasterrole(struct ctdb_context *ctdb, int argc, const char **argv)
4995 uint32_t recmasterrole;
5001 if (!strcmp(argv[0], "on")) {
5003 } else if (!strcmp(argv[0], "off")) {
5009 ret = ctdb_ctrl_setrecmasterrole(ctdb, TIMELIMIT(), options.pnn, recmasterrole);
5011 DEBUG(DEBUG_ERR, ("Unable to set the recmaster role for node %u\n", options.pnn));
5019 set debug level on a node or all nodes
5021 static int control_setdebug(struct ctdb_context *ctdb, int argc, const char **argv)
5027 printf("You must specify the debug level. Valid levels are:\n");
5028 print_debug_levels(stdout);
5032 if (!parse_debug(argv[0], &level)) {
5033 printf("Invalid debug level, must be one of\n");
5034 print_debug_levels(stdout);
5038 ret = ctdb_ctrl_set_debuglevel(ctdb, options.pnn, level);
5040 DEBUG(DEBUG_ERR, ("Unable to set debug level on node %u\n", options.pnn));
5049 static int control_thaw(struct ctdb_context *ctdb, int argc, const char **argv)
5055 priority = strtol(argv[0], NULL, 0);
5059 DEBUG(DEBUG_ERR,("Thaw by priority %u\n", priority));
5061 ret = ctdb_ctrl_thaw_priority(ctdb, TIMELIMIT(), options.pnn, priority);
5063 DEBUG(DEBUG_ERR, ("Unable to thaw node %u\n", options.pnn));
5070 attach to a database
5072 static int control_attach(struct ctdb_context *ctdb, int argc, const char **argv)
5074 const char *db_name;
5075 struct ctdb_db_context *ctdb_db;
5076 bool persistent = false;
5086 if (strcmp(argv[1], "persistent") != 0) {
5092 ctdb_db = ctdb_attach(ctdb, TIMELIMIT(), db_name, persistent, 0);
5093 if (ctdb_db == NULL) {
5094 DEBUG(DEBUG_ERR,("Unable to attach to database '%s'\n", db_name));
5102 * detach from a database
5104 static int control_detach(struct ctdb_context *ctdb, int argc,
5109 int ret, i, status = 0;
5110 struct ctdb_node_map *nodemap = NULL;
5111 TALLOC_CTX *tmp_ctx = talloc_new(ctdb);
5118 assert_single_node_only();
5120 ret = ctdb_ctrl_getrecmode(ctdb, tmp_ctx, TIMELIMIT(), options.pnn,
5123 DEBUG(DEBUG_ERR, ("Database cannot be detached "
5124 "when recovery is active\n"));
5125 talloc_free(tmp_ctx);
5129 ret = ctdb_ctrl_getnodemap(ctdb, TIMELIMIT(), options.pnn, tmp_ctx,
5132 DEBUG(DEBUG_ERR, ("Unable to get nodemap from node %u\n",
5134 talloc_free(tmp_ctx);
5138 for (i=0; i<nodemap->num; i++) {
5141 if (nodemap->nodes[i].flags & NODE_FLAGS_DISCONNECTED) {
5145 if (nodemap->nodes[i].flags & NODE_FLAGS_DELETED) {
5149 if (nodemap->nodes[i].flags & NODE_FLAGS_INACTIVE) {
5150 DEBUG(DEBUG_ERR, ("Database cannot be detached on "
5151 "inactive (stopped or banned) node "
5152 "%u\n", nodemap->nodes[i].pnn));
5153 talloc_free(tmp_ctx);
5157 ret = ctdb_ctrl_get_tunable(ctdb, TIMELIMIT(),
5158 nodemap->nodes[i].pnn,
5159 "AllowClientDBAttach",
5162 DEBUG(DEBUG_ERR, ("Unable to get tunable "
5163 "AllowClientDBAttach from node %u\n",
5164 nodemap->nodes[i].pnn));
5165 talloc_free(tmp_ctx);
5170 DEBUG(DEBUG_ERR, ("Database access is still active on "
5171 "node %u. Set AllowClientDBAttach=0 "
5173 nodemap->nodes[i].pnn));
5174 talloc_free(tmp_ctx);
5179 talloc_free(tmp_ctx);
5181 for (i=0; i<argc; i++) {
5182 if (!db_exists(ctdb, argv[i], &db_id, NULL, &flags)) {
5186 if (flags & CTDB_DB_FLAGS_PERSISTENT) {
5187 DEBUG(DEBUG_ERR, ("Persistent database '%s' "
5188 "cannot be detached\n", argv[i]));
5193 ret = ctdb_detach(ctdb, db_id);
5195 DEBUG(DEBUG_ERR, ("Database '%s' detach failed\n",
5207 static int control_setdbprio(struct ctdb_context *ctdb, int argc, const char **argv)
5209 struct ctdb_db_priority db_prio;
5216 db_prio.db_id = strtoul(argv[0], NULL, 0);
5217 db_prio.priority = strtoul(argv[1], NULL, 0);
5219 ret = ctdb_ctrl_set_db_priority(ctdb, TIMELIMIT(), options.pnn, &db_prio);
5221 DEBUG(DEBUG_ERR,("Unable to set db prio\n"));
5231 static int control_getdbprio(struct ctdb_context *ctdb, int argc, const char **argv)
5233 uint32_t db_id, priority;
5240 if (!db_exists(ctdb, argv[0], &db_id, NULL, NULL)) {
5244 ret = ctdb_ctrl_get_db_priority(ctdb, TIMELIMIT(), options.pnn, db_id, &priority);
5246 DEBUG(DEBUG_ERR,("Unable to get db prio\n"));
5250 DEBUG(DEBUG_ERR,("Priority:%u\n", priority));
5256 set the sticky records capability for a database
5258 static int control_setdbsticky(struct ctdb_context *ctdb, int argc, const char **argv)
5260 TALLOC_CTX *tmp_ctx = talloc_new(ctdb);
5268 if (!db_exists(ctdb, argv[0], &db_id, NULL, NULL)) {
5272 ret = ctdb_ctrl_set_db_sticky(ctdb, options.pnn, db_id);
5274 DEBUG(DEBUG_ERR,("Unable to set db to support sticky records\n"));
5275 talloc_free(tmp_ctx);
5279 talloc_free(tmp_ctx);
5284 set the readonly capability for a database
5286 static int control_setdbreadonly(struct ctdb_context *ctdb, int argc, const char **argv)
5288 TALLOC_CTX *tmp_ctx = talloc_new(ctdb);
5296 if (!db_exists(ctdb, argv[0], &db_id, NULL, NULL)) {
5300 ret = ctdb_ctrl_set_db_readonly(ctdb, options.pnn, db_id);
5302 DEBUG(DEBUG_ERR,("Unable to set db to support readonly\n"));
5303 talloc_free(tmp_ctx);
5307 talloc_free(tmp_ctx);
5314 static int control_getdbseqnum(struct ctdb_context *ctdb, int argc, const char **argv)
5324 if (!db_exists(ctdb, argv[0], &db_id, NULL, NULL)) {
5328 ret = ctdb_ctrl_getdbseqnum(ctdb, TIMELIMIT(), options.pnn, db_id, &seqnum);
5330 DEBUG(DEBUG_ERR, ("Unable to get seqnum from node."));
5334 printf("Sequence number:%lld\n", (long long)seqnum);
5340 run an eventscript on a node
5342 static int control_eventscript(struct ctdb_context *ctdb, int argc, const char **argv)
5348 TALLOC_CTX *tmp_ctx = talloc_new(ctdb);
5351 DEBUG(DEBUG_ERR,("Invalid arguments\n"));
5355 data.dptr = (unsigned char *)discard_const(argv[0]);
5356 data.dsize = strlen((char *)data.dptr) + 1;
5358 DEBUG(DEBUG_ERR, ("Running eventscripts with arguments \"%s\" on node %u\n", data.dptr, options.pnn));
5360 ret = ctdb_control(ctdb, options.pnn, 0, CTDB_CONTROL_RUN_EVENTSCRIPTS,
5361 0, data, tmp_ctx, NULL, &res, NULL, &errmsg);
5362 if (ret != 0 || res != 0) {
5363 DEBUG(DEBUG_ERR,("Failed to run eventscripts - %s\n", errmsg));
5364 talloc_free(tmp_ctx);
5367 talloc_free(tmp_ctx);
5371 #define DB_VERSION 1
5372 #define MAX_DB_NAME 64
5373 struct db_file_header {
5374 unsigned long version;
5376 unsigned long persistent;
5378 const char name[MAX_DB_NAME];
5381 struct backup_data {
5382 struct ctdb_marshall_buffer *records;
5385 bool traverse_error;
5388 static int backup_traverse(struct tdb_context *tdb, TDB_DATA key, TDB_DATA data, void *private)
5390 struct backup_data *bd = talloc_get_type(private, struct backup_data);
5391 struct ctdb_rec_data *rec;
5393 /* add the record */
5394 rec = ctdb_marshall_record(bd->records, 0, key, NULL, data);
5396 bd->traverse_error = true;
5397 DEBUG(DEBUG_ERR,("Failed to marshall record\n"));
5400 bd->records = talloc_realloc_size(NULL, bd->records, rec->length + bd->len);
5401 if (bd->records == NULL) {
5402 DEBUG(DEBUG_ERR,("Failed to expand marshalling buffer\n"));
5403 bd->traverse_error = true;
5406 bd->records->count++;
5407 memcpy(bd->len+(uint8_t *)bd->records, rec, rec->length);
5408 bd->len += rec->length;
5416 * backup a database to a file
5418 static int control_backupdb(struct ctdb_context *ctdb, int argc, const char **argv)
5420 const char *db_name;
5422 TALLOC_CTX *tmp_ctx = talloc_new(ctdb);
5423 struct db_file_header dbhdr;
5424 struct ctdb_db_context *ctdb_db;
5425 struct backup_data *bd;
5428 const char *reason = NULL;
5432 assert_single_node_only();
5435 DEBUG(DEBUG_ERR,("Invalid arguments\n"));
5439 if (!db_exists(ctdb, argv[0], &db_id, &db_name, &flags)) {
5443 ret = ctdb_ctrl_getdbhealth(ctdb, TIMELIMIT(), options.pnn,
5444 db_id, tmp_ctx, &reason);
5446 DEBUG(DEBUG_ERR,("Unable to get dbhealth for database '%s'\n",
5448 talloc_free(tmp_ctx);
5452 uint32_t allow_unhealthy = 0;
5454 ctdb_ctrl_get_tunable(ctdb, TIMELIMIT(), options.pnn,
5455 "AllowUnhealthyDBRead",
5458 if (allow_unhealthy != 1) {
5459 DEBUG(DEBUG_ERR,("database '%s' is unhealthy: %s\n",
5462 DEBUG(DEBUG_ERR,("disallow backup : tunable AllowUnhealthyDBRead = %u\n",
5464 talloc_free(tmp_ctx);
5468 DEBUG(DEBUG_WARNING,("WARNING database '%s' is unhealthy - see 'ctdb getdbstatus %s'\n",
5470 DEBUG(DEBUG_WARNING,("WARNING! allow backup of unhealthy database: "
5471 "tunnable AllowUnhealthyDBRead = %u\n",
5475 ctdb_db = ctdb_attach(ctdb, TIMELIMIT(), db_name, flags & CTDB_DB_FLAGS_PERSISTENT, 0);
5476 if (ctdb_db == NULL) {
5477 DEBUG(DEBUG_ERR,("Unable to attach to database '%s'\n", argv[0]));
5478 talloc_free(tmp_ctx);
5483 ret = tdb_transaction_start(ctdb_db->ltdb->tdb);
5485 DEBUG(DEBUG_ERR,("Failed to start transaction\n"));
5486 talloc_free(tmp_ctx);
5491 bd = talloc_zero(tmp_ctx, struct backup_data);
5493 DEBUG(DEBUG_ERR,("Failed to allocate backup_data\n"));
5494 talloc_free(tmp_ctx);
5498 bd->records = talloc_zero(bd, struct ctdb_marshall_buffer);
5499 if (bd->records == NULL) {
5500 DEBUG(DEBUG_ERR,("Failed to allocate ctdb_marshall_buffer\n"));
5501 talloc_free(tmp_ctx);
5505 bd->len = offsetof(struct ctdb_marshall_buffer, data);
5506 bd->records->db_id = ctdb_db->db_id;
5507 /* traverse the database collecting all records */
5508 if (tdb_traverse_read(ctdb_db->ltdb->tdb, backup_traverse, bd) == -1 ||
5509 bd->traverse_error) {
5510 DEBUG(DEBUG_ERR,("Traverse error\n"));
5511 talloc_free(tmp_ctx);
5515 tdb_transaction_cancel(ctdb_db->ltdb->tdb);
5518 fh = open(argv[1], O_RDWR|O_CREAT, 0600);
5520 DEBUG(DEBUG_ERR,("Failed to open file '%s'\n", argv[1]));
5521 talloc_free(tmp_ctx);
5526 dbhdr.version = DB_VERSION;
5527 dbhdr.timestamp = time(NULL);
5528 dbhdr.persistent = flags & CTDB_DB_FLAGS_PERSISTENT;
5529 dbhdr.size = bd->len;
5530 if (strlen(argv[0]) >= MAX_DB_NAME) {
5531 DEBUG(DEBUG_ERR,("Too long dbname\n"));
5534 strncpy(discard_const(dbhdr.name), argv[0], MAX_DB_NAME-1);
5535 ret = sys_write(fh, &dbhdr, sizeof(dbhdr));
5537 DEBUG(DEBUG_ERR,("write failed: %s\n", strerror(errno)));
5540 ret = sys_write(fh, bd->records, bd->len);
5542 DEBUG(DEBUG_ERR,("write failed: %s\n", strerror(errno)));
5551 DEBUG(DEBUG_ERR,("close failed: %s\n", strerror(errno)));
5555 DEBUG(DEBUG_ERR,("Database backed up to %s\n", argv[1]));
5557 talloc_free(tmp_ctx);
5562 * restore a database from a file
5564 static int control_restoredb(struct ctdb_context *ctdb, int argc, const char **argv)
5567 TALLOC_CTX *tmp_ctx = talloc_new(ctdb);
5570 struct db_file_header dbhdr;
5571 struct ctdb_db_context *ctdb_db;
5572 struct ctdb_node_map *nodemap=NULL;
5573 struct ctdb_vnn_map *vnnmap=NULL;
5575 struct ctdb_control_wipe_database w;
5577 uint32_t generation;
5582 assert_single_node_only();
5584 if (argc < 1 || argc > 2) {
5585 DEBUG(DEBUG_ERR,("Invalid arguments\n"));
5589 fh = open(argv[0], O_RDONLY);
5591 DEBUG(DEBUG_ERR,("Failed to open file '%s'\n", argv[0]));
5592 talloc_free(tmp_ctx);
5596 sys_read(fh, &dbhdr, sizeof(dbhdr));
5597 if (dbhdr.version != DB_VERSION) {
5598 DEBUG(DEBUG_ERR,("Invalid version of database dump. File is version %lu but expected version was %u\n", dbhdr.version, DB_VERSION));
5600 talloc_free(tmp_ctx);
5604 dbname = discard_const(dbhdr.name);
5606 dbname = discard_const(argv[1]);
5609 outdata.dsize = dbhdr.size;
5610 outdata.dptr = talloc_size(tmp_ctx, outdata.dsize);
5611 if (outdata.dptr == NULL) {
5612 DEBUG(DEBUG_ERR,("Failed to allocate data of size '%lu'\n", dbhdr.size));
5614 talloc_free(tmp_ctx);
5617 sys_read(fh, outdata.dptr, outdata.dsize);
5620 tm = localtime(&dbhdr.timestamp);
5621 strftime(tbuf,sizeof(tbuf)-1,"%Y/%m/%d %H:%M:%S", tm);
5622 printf("Restoring database '%s' from backup @ %s\n",
5626 ctdb_db = ctdb_attach(ctdb, TIMELIMIT(), dbname, dbhdr.persistent, 0);
5627 if (ctdb_db == NULL) {
5628 DEBUG(DEBUG_ERR,("Unable to attach to database '%s'\n", dbname));
5629 talloc_free(tmp_ctx);
5633 ret = ctdb_ctrl_getnodemap(ctdb, TIMELIMIT(), options.pnn, ctdb, &nodemap);
5635 DEBUG(DEBUG_ERR, ("Unable to get nodemap from node %u\n", options.pnn));
5636 talloc_free(tmp_ctx);
5641 ret = ctdb_ctrl_getvnnmap(ctdb, TIMELIMIT(), options.pnn, tmp_ctx, &vnnmap);
5643 DEBUG(DEBUG_ERR, ("Unable to get vnnmap from node %u\n", options.pnn));
5644 talloc_free(tmp_ctx);
5648 /* freeze all nodes */
5649 nodes = list_of_active_nodes(ctdb, nodemap, tmp_ctx, true);
5650 for (i=1; i<=NUM_DB_PRIORITIES; i++) {
5651 if (ctdb_client_async_control(ctdb, CTDB_CONTROL_FREEZE,
5657 DEBUG(DEBUG_ERR, ("Unable to freeze nodes.\n"));
5658 ctdb_ctrl_setrecmode(ctdb, TIMELIMIT(), options.pnn, CTDB_RECOVERY_ACTIVE);
5659 talloc_free(tmp_ctx);
5664 generation = vnnmap->generation;
5665 data.dptr = (void *)&generation;
5666 data.dsize = sizeof(generation);
5668 /* start a cluster wide transaction */
5669 nodes = list_of_active_nodes(ctdb, nodemap, tmp_ctx, true);
5670 if (ctdb_client_async_control(ctdb, CTDB_CONTROL_TRANSACTION_START,
5672 TIMELIMIT(), false, data,
5675 DEBUG(DEBUG_ERR, ("Unable to start cluster wide transactions.\n"));
5680 w.db_id = ctdb_db->db_id;
5681 w.transaction_id = generation;
5683 data.dptr = (void *)&w;
5684 data.dsize = sizeof(w);
5686 /* wipe all the remote databases. */
5687 nodes = list_of_active_nodes(ctdb, nodemap, tmp_ctx, true);
5688 if (ctdb_client_async_control(ctdb, CTDB_CONTROL_WIPE_DATABASE,
5690 TIMELIMIT(), false, data,
5693 DEBUG(DEBUG_ERR, ("Unable to wipe database.\n"));
5694 ctdb_ctrl_setrecmode(ctdb, TIMELIMIT(), options.pnn, CTDB_RECOVERY_ACTIVE);
5695 talloc_free(tmp_ctx);
5699 /* push the database */
5700 nodes = list_of_active_nodes(ctdb, nodemap, tmp_ctx, true);
5701 if (ctdb_client_async_control(ctdb, CTDB_CONTROL_PUSH_DB,
5703 TIMELIMIT(), false, outdata,
5706 DEBUG(DEBUG_ERR, ("Failed to push database.\n"));
5707 ctdb_ctrl_setrecmode(ctdb, TIMELIMIT(), options.pnn, CTDB_RECOVERY_ACTIVE);
5708 talloc_free(tmp_ctx);
5712 data.dptr = (void *)&ctdb_db->db_id;
5713 data.dsize = sizeof(ctdb_db->db_id);
5715 /* mark the database as healthy */
5716 nodes = list_of_active_nodes(ctdb, nodemap, tmp_ctx, true);
5717 if (ctdb_client_async_control(ctdb, CTDB_CONTROL_DB_SET_HEALTHY,
5719 TIMELIMIT(), false, data,
5722 DEBUG(DEBUG_ERR, ("Failed to mark database as healthy.\n"));
5723 ctdb_ctrl_setrecmode(ctdb, TIMELIMIT(), options.pnn, CTDB_RECOVERY_ACTIVE);
5724 talloc_free(tmp_ctx);
5728 data.dptr = (void *)&generation;
5729 data.dsize = sizeof(generation);
5731 /* commit all the changes */
5732 if (ctdb_client_async_control(ctdb, CTDB_CONTROL_TRANSACTION_COMMIT,
5734 TIMELIMIT(), false, data,
5737 DEBUG(DEBUG_ERR, ("Unable to commit databases.\n"));
5738 ctdb_ctrl_setrecmode(ctdb, TIMELIMIT(), options.pnn, CTDB_RECOVERY_ACTIVE);
5739 talloc_free(tmp_ctx);
5744 /* thaw all nodes */
5745 nodes = list_of_active_nodes(ctdb, nodemap, tmp_ctx, true);
5746 if (ctdb_client_async_control(ctdb, CTDB_CONTROL_THAW,
5752 DEBUG(DEBUG_ERR, ("Unable to thaw nodes.\n"));
5753 ctdb_ctrl_setrecmode(ctdb, TIMELIMIT(), options.pnn, CTDB_RECOVERY_ACTIVE);
5754 talloc_free(tmp_ctx);
5759 talloc_free(tmp_ctx);
5764 * dump a database backup from a file
5766 static int control_dumpdbbackup(struct ctdb_context *ctdb, int argc, const char **argv)
5768 TALLOC_CTX *tmp_ctx = talloc_new(ctdb);
5770 struct db_file_header dbhdr;
5774 struct ctdb_rec_data *rec = NULL;
5775 struct ctdb_marshall_buffer *m;
5776 struct ctdb_dump_db_context c;
5778 assert_single_node_only();
5781 DEBUG(DEBUG_ERR,("Invalid arguments\n"));
5785 fh = open(argv[0], O_RDONLY);
5787 DEBUG(DEBUG_ERR,("Failed to open file '%s'\n", argv[0]));
5788 talloc_free(tmp_ctx);
5792 sys_read(fh, &dbhdr, sizeof(dbhdr));
5793 if (dbhdr.version != DB_VERSION) {
5794 DEBUG(DEBUG_ERR,("Invalid version of database dump. File is version %lu but expected version was %u\n", dbhdr.version, DB_VERSION));
5796 talloc_free(tmp_ctx);
5800 outdata.dsize = dbhdr.size;
5801 outdata.dptr = talloc_size(tmp_ctx, outdata.dsize);
5802 if (outdata.dptr == NULL) {
5803 DEBUG(DEBUG_ERR,("Failed to allocate data of size '%lu'\n", dbhdr.size));
5805 talloc_free(tmp_ctx);
5808 sys_read(fh, outdata.dptr, outdata.dsize);
5810 m = (struct ctdb_marshall_buffer *)outdata.dptr;
5812 tm = localtime(&dbhdr.timestamp);
5813 strftime(tbuf,sizeof(tbuf)-1,"%Y/%m/%d %H:%M:%S", tm);
5814 printf("Backup of database name:'%s' dbid:0x%x08x from @ %s\n",
5815 dbhdr.name, m->db_id, tbuf);
5819 c.printemptyrecords = (bool)options.printemptyrecords;
5820 c.printdatasize = (bool)options.printdatasize;
5821 c.printlmaster = false;
5822 c.printhash = (bool)options.printhash;
5823 c.printrecordflags = (bool)options.printrecordflags;
5825 for (i=0; i < m->count; i++) {
5829 /* we do not want the header splitted, so we pass NULL*/
5830 rec = ctdb_marshall_loop_next(m, rec, &reqid,
5833 ctdb_dumpdb_record(ctdb, key, data, &c);
5836 printf("Dumped %d records\n", i);
5837 talloc_free(tmp_ctx);
5842 * wipe a database from a file
5844 static int control_wipedb(struct ctdb_context *ctdb, int argc,
5847 const char *db_name;
5849 TALLOC_CTX *tmp_ctx = talloc_new(ctdb);
5851 struct ctdb_db_context *ctdb_db;
5852 struct ctdb_node_map *nodemap = NULL;
5853 struct ctdb_vnn_map *vnnmap = NULL;
5855 struct ctdb_control_wipe_database w;
5857 uint32_t generation;
5860 assert_single_node_only();
5863 DEBUG(DEBUG_ERR,("Invalid arguments\n"));
5867 if (!db_exists(ctdb, argv[0], NULL, &db_name, &flags)) {
5871 ctdb_db = ctdb_attach(ctdb, TIMELIMIT(), db_name, flags & CTDB_DB_FLAGS_PERSISTENT, 0);
5872 if (ctdb_db == NULL) {
5873 DEBUG(DEBUG_ERR, ("Unable to attach to database '%s'\n",
5875 talloc_free(tmp_ctx);
5879 ret = ctdb_ctrl_getnodemap(ctdb, TIMELIMIT(), options.pnn, ctdb,
5882 DEBUG(DEBUG_ERR, ("Unable to get nodemap from node %u\n",
5884 talloc_free(tmp_ctx);
5888 ret = ctdb_ctrl_getvnnmap(ctdb, TIMELIMIT(), options.pnn, tmp_ctx,
5891 DEBUG(DEBUG_ERR, ("Unable to get vnnmap from node %u\n",
5893 talloc_free(tmp_ctx);
5897 /* freeze all nodes */
5898 nodes = list_of_active_nodes(ctdb, nodemap, tmp_ctx, true);
5899 for (i=1; i<=NUM_DB_PRIORITIES; i++) {
5900 ret = ctdb_client_async_control(ctdb, CTDB_CONTROL_FREEZE,
5907 DEBUG(DEBUG_ERR, ("Unable to freeze nodes.\n"));
5908 ctdb_ctrl_setrecmode(ctdb, TIMELIMIT(), options.pnn,
5909 CTDB_RECOVERY_ACTIVE);
5910 talloc_free(tmp_ctx);
5915 generation = vnnmap->generation;
5916 data.dptr = (void *)&generation;
5917 data.dsize = sizeof(generation);
5919 /* start a cluster wide transaction */
5920 nodes = list_of_active_nodes(ctdb, nodemap, tmp_ctx, true);
5921 ret = ctdb_client_async_control(ctdb, CTDB_CONTROL_TRANSACTION_START,
5923 TIMELIMIT(), false, data,
5927 DEBUG(DEBUG_ERR, ("Unable to start cluster wide "
5928 "transactions.\n"));
5932 w.db_id = ctdb_db->db_id;
5933 w.transaction_id = generation;
5935 data.dptr = (void *)&w;
5936 data.dsize = sizeof(w);
5938 /* wipe all the remote databases. */
5939 nodes = list_of_active_nodes(ctdb, nodemap, tmp_ctx, true);
5940 if (ctdb_client_async_control(ctdb, CTDB_CONTROL_WIPE_DATABASE,
5942 TIMELIMIT(), false, data,
5945 DEBUG(DEBUG_ERR, ("Unable to wipe database.\n"));
5946 ctdb_ctrl_setrecmode(ctdb, TIMELIMIT(), options.pnn, CTDB_RECOVERY_ACTIVE);
5947 talloc_free(tmp_ctx);
5951 data.dptr = (void *)&ctdb_db->db_id;
5952 data.dsize = sizeof(ctdb_db->db_id);
5954 /* mark the database as healthy */
5955 nodes = list_of_active_nodes(ctdb, nodemap, tmp_ctx, true);
5956 if (ctdb_client_async_control(ctdb, CTDB_CONTROL_DB_SET_HEALTHY,
5958 TIMELIMIT(), false, data,
5961 DEBUG(DEBUG_ERR, ("Failed to mark database as healthy.\n"));
5962 ctdb_ctrl_setrecmode(ctdb, TIMELIMIT(), options.pnn, CTDB_RECOVERY_ACTIVE);
5963 talloc_free(tmp_ctx);
5967 data.dptr = (void *)&generation;
5968 data.dsize = sizeof(generation);
5970 /* commit all the changes */
5971 if (ctdb_client_async_control(ctdb, CTDB_CONTROL_TRANSACTION_COMMIT,
5973 TIMELIMIT(), false, data,
5976 DEBUG(DEBUG_ERR, ("Unable to commit databases.\n"));
5977 ctdb_ctrl_setrecmode(ctdb, TIMELIMIT(), options.pnn, CTDB_RECOVERY_ACTIVE);
5978 talloc_free(tmp_ctx);
5982 /* thaw all nodes */
5983 nodes = list_of_active_nodes(ctdb, nodemap, tmp_ctx, true);
5984 if (ctdb_client_async_control(ctdb, CTDB_CONTROL_THAW,
5990 DEBUG(DEBUG_ERR, ("Unable to thaw nodes.\n"));
5991 ctdb_ctrl_setrecmode(ctdb, TIMELIMIT(), options.pnn, CTDB_RECOVERY_ACTIVE);
5992 talloc_free(tmp_ctx);
5996 DEBUG(DEBUG_ERR, ("Database wiped.\n"));
5998 talloc_free(tmp_ctx);
6005 static int control_dumpmemory(struct ctdb_context *ctdb, int argc, const char **argv)
6011 TALLOC_CTX *tmp_ctx = talloc_new(ctdb);
6012 ret = ctdb_control(ctdb, options.pnn, 0, CTDB_CONTROL_DUMP_MEMORY,
6013 0, tdb_null, tmp_ctx, &data, &res, NULL, &errmsg);
6014 if (ret != 0 || res != 0) {
6015 DEBUG(DEBUG_ERR,("Failed to dump memory - %s\n", errmsg));
6016 talloc_free(tmp_ctx);
6019 sys_write(1, data.dptr, data.dsize);
6020 talloc_free(tmp_ctx);
6025 handler for memory dumps
6027 static void mem_dump_handler(struct ctdb_context *ctdb, uint64_t srvid,
6028 TDB_DATA data, void *private_data)
6030 sys_write(1, data.dptr, data.dsize);
6035 dump memory usage on the recovery daemon
6037 static int control_rddumpmemory(struct ctdb_context *ctdb, int argc, const char **argv)
6041 struct srvid_request rd;
6043 rd.pnn = ctdb_get_pnn(ctdb);
6044 rd.srvid = getpid();
6046 /* register a message port for receiveing the reply so that we
6047 can receive the reply
6049 ctdb_client_set_message_handler(ctdb, rd.srvid, mem_dump_handler, NULL);
6052 data.dptr = (uint8_t *)&rd;
6053 data.dsize = sizeof(rd);
6055 ret = ctdb_client_send_message(ctdb, options.pnn, CTDB_SRVID_MEM_DUMP, data);
6057 DEBUG(DEBUG_ERR,("Failed to send memdump request message to %u\n", options.pnn));
6061 /* this loop will terminate when we have received the reply */
6063 event_loop_once(ctdb->ev);
6070 send a message to a srvid
6072 static int control_msgsend(struct ctdb_context *ctdb, int argc, const char **argv)
6074 unsigned long srvid;
6082 srvid = strtoul(argv[0], NULL, 0);
6084 data.dptr = (uint8_t *)discard_const(argv[1]);
6085 data.dsize= strlen(argv[1]);
6087 ret = ctdb_client_send_message(ctdb, CTDB_BROADCAST_CONNECTED, srvid, data);
6089 DEBUG(DEBUG_ERR,("Failed to send memdump request message to %u\n", options.pnn));
6097 handler for msglisten
6099 static void msglisten_handler(struct ctdb_context *ctdb, uint64_t srvid,
6100 TDB_DATA data, void *private_data)
6104 printf("Message received: ");
6105 for (i=0;i<data.dsize;i++) {
6106 printf("%c", data.dptr[i]);
6112 listen for messages on a messageport
6114 static int control_msglisten(struct ctdb_context *ctdb, int argc, const char **argv)
6120 /* register a message port and listen for messages
6122 ctdb_client_set_message_handler(ctdb, srvid, msglisten_handler, NULL);
6123 printf("Listening for messages on srvid:%d\n", (int)srvid);
6126 event_loop_once(ctdb->ev);
6133 list all nodes in the cluster
6134 we parse the nodes file directly
6136 static int control_listnodes(struct ctdb_context *ctdb, int argc, const char **argv)
6138 TALLOC_CTX *mem_ctx = talloc_new(NULL);
6139 struct ctdb_node_map *node_map;
6142 assert_single_node_only();
6144 node_map = read_nodes_file(mem_ctx);
6145 if (node_map == NULL) {
6146 talloc_free(mem_ctx);
6150 for (i = 0; i < node_map->num; i++) {
6153 if (node_map->nodes[i].flags & NODE_FLAGS_DELETED) {
6156 addr = ctdb_addr_to_str(&node_map->nodes[i].addr);
6157 if (options.machinereadable){
6158 printm(":%d:%s:\n", node_map->nodes[i].pnn, addr);
6160 printf("%s\n", addr);
6163 talloc_free(mem_ctx);
6168 /**********************************************************************/
6169 /* reload the nodes file on all nodes */
6171 static void get_nodes_files_callback(struct ctdb_context *ctdb,
6172 uint32_t node_pnn, int32_t res,
6173 TDB_DATA outdata, void *callback_data)
6175 struct ctdb_node_map **maps =
6176 talloc_get_type(callback_data, struct ctdb_node_map *);
6178 if (outdata.dsize < offsetof(struct ctdb_node_map, nodes) ||
6179 outdata.dptr == NULL) {
6181 (__location__ " Invalid return data: %u %p\n",
6182 (unsigned)outdata.dsize, outdata.dptr));
6186 if (node_pnn >= talloc_array_length(maps)) {
6188 (__location__ " unexpected PNN %u\n", node_pnn));
6192 maps[node_pnn] = talloc_memdup(maps, outdata.dptr, outdata.dsize);
6195 static void get_nodes_files_fail_callback(struct ctdb_context *ctdb,
6196 uint32_t node_pnn, int32_t res,
6197 TDB_DATA outdata, void *callback_data)
6200 ("ERROR: Failed to get nodes file from node %u\n", node_pnn));
6203 static struct ctdb_node_map **
6204 ctdb_get_nodes_files(struct ctdb_context *ctdb,
6205 TALLOC_CTX *mem_ctx,
6206 struct timeval timeout,
6207 struct ctdb_node_map *nodemap)
6211 struct ctdb_node_map **maps;
6213 maps = talloc_zero_array(mem_ctx, struct ctdb_node_map *, nodemap->num);
6214 CTDB_NO_MEMORY_NULL(ctdb, maps);
6216 nodes = list_of_connected_nodes(ctdb, nodemap, mem_ctx, true);
6218 ret = ctdb_client_async_control(ctdb, CTDB_CONTROL_GET_NODES_FILE,
6219 nodes, 0, TIMELIMIT(),
6221 get_nodes_files_callback,
6222 get_nodes_files_fail_callback,
6232 static bool node_files_are_identical(struct ctdb_node_map *nm1,
6233 struct ctdb_node_map *nm2)
6237 if (nm1->num != nm2->num) {
6240 for (i = 0; i < nm1->num; i++) {
6241 if (memcmp(&nm1->nodes[i], &nm2->nodes[i],
6242 sizeof(struct ctdb_node_and_flags)) != 0) {
6250 static bool check_all_node_files_are_identical(struct ctdb_context *ctdb,
6251 TALLOC_CTX *mem_ctx,
6252 struct timeval timeout,
6253 struct ctdb_node_map *nodemap,
6254 struct ctdb_node_map *file_nodemap)
6256 static struct ctdb_node_map **maps;
6260 maps = ctdb_get_nodes_files(ctdb, mem_ctx, timeout, nodemap);
6265 for (i = 0; i < talloc_array_length(maps); i++) {
6266 if (maps[i] == NULL) {
6269 if (!node_files_are_identical(file_nodemap, maps[i])) {
6271 ("ERROR: Node file on node %u differs from current node (%u)\n",
6272 i, ctdb_get_pnn(ctdb)));
6281 reload the nodes file on the local node
6283 static bool sanity_check_nodes_file_changes(TALLOC_CTX *mem_ctx,
6284 struct ctdb_node_map *nodemap,
6285 struct ctdb_node_map *file_nodemap)
6288 bool should_abort = false;
6289 bool have_changes = false;
6291 for (i=0; i<nodemap->num; i++) {
6292 if (i >= file_nodemap->num) {
6294 ("ERROR: Node %u (%s) missing from nodes file\n",
6295 nodemap->nodes[i].pnn,
6296 ctdb_addr_to_str(&nodemap->nodes[i].addr)));
6297 should_abort = true;
6300 if (nodemap->nodes[i].flags & NODE_FLAGS_DELETED &&
6301 file_nodemap->nodes[i].flags & NODE_FLAGS_DELETED) {
6302 /* Node remains deleted */
6304 ("Node %u is unchanged (DELETED)\n",
6305 nodemap->nodes[i].pnn));
6306 } else if (!(nodemap->nodes[i].flags & NODE_FLAGS_DELETED) &&
6307 !(file_nodemap->nodes[i].flags & NODE_FLAGS_DELETED)) {
6308 /* Node not newly nor previously deleted */
6309 if (!ctdb_same_ip(&nodemap->nodes[i].addr,
6310 &file_nodemap->nodes[i].addr)) {
6312 ("ERROR: Node %u has changed IP address (was %s, now %s)\n",
6313 nodemap->nodes[i].pnn,
6314 /* ctdb_addr_to_str() returns a static */
6315 talloc_strdup(mem_ctx,
6316 ctdb_addr_to_str(&nodemap->nodes[i].addr)),
6317 ctdb_addr_to_str(&file_nodemap->nodes[i].addr)));
6318 should_abort = true;
6321 ("Node %u is unchanged\n",
6322 nodemap->nodes[i].pnn));
6323 if (nodemap->nodes[i].flags & NODE_FLAGS_DISCONNECTED) {
6324 DEBUG(DEBUG_WARNING,
6325 ("WARNING: Node %u is disconnected. You MUST fix this node manually!\n",
6326 nodemap->nodes[i].pnn));
6329 } else if (file_nodemap->nodes[i].flags & NODE_FLAGS_DELETED) {
6330 /* Node is being deleted */
6332 ("Node %u is DELETED\n",
6333 nodemap->nodes[i].pnn));
6334 have_changes = true;
6335 if (!(nodemap->nodes[i].flags & NODE_FLAGS_DISCONNECTED)) {
6337 ("ERROR: Node %u is still connected\n",
6338 nodemap->nodes[i].pnn));
6339 should_abort = true;
6341 } else if (nodemap->nodes[i].flags & NODE_FLAGS_DELETED) {
6342 /* Node was previously deleted */
6344 ("Node %u is UNDELETED\n", nodemap->nodes[i].pnn));
6345 have_changes = true;
6351 ("ERROR: Nodes will not be reloaded due to previous error\n"));
6352 talloc_free(mem_ctx);
6356 /* Leftover nodes in file are NEW */
6357 for (; i < file_nodemap->num; i++) {
6358 DEBUG(DEBUG_NOTICE, ("Node %u is NEW\n",
6359 file_nodemap->nodes[i].pnn));
6360 have_changes = true;
6363 return have_changes;
6366 static void reload_nodes_fail_callback(struct ctdb_context *ctdb,
6367 uint32_t node_pnn, int32_t res,
6368 TDB_DATA outdata, void *callback_data)
6370 DEBUG(DEBUG_WARNING,
6371 ("WARNING: Node %u failed to reload nodes. You MUST fix this node manually!\n",
6375 static int control_reload_nodes_file(struct ctdb_context *ctdb, int argc, const char **argv)
6378 struct ctdb_node_map *nodemap=NULL;
6379 TALLOC_CTX *tmp_ctx = talloc_new(NULL);
6380 struct ctdb_node_map *file_nodemap;
6384 assert_current_node_only(ctdb);
6386 /* Load both the current nodemap and the contents of the local
6387 * nodes file. Compare and sanity check them before doing
6390 ret = ctdb_ctrl_getnodemap(ctdb, TIMELIMIT(), CTDB_CURRENT_NODE, ctdb, &nodemap);
6392 DEBUG(DEBUG_ERR, ("Unable to get nodemap from local node\n"));
6396 file_nodemap = read_nodes_file(tmp_ctx);
6397 if (file_nodemap == NULL) {
6398 DEBUG(DEBUG_ERR,("Failed to read nodes file\n"));
6399 talloc_free(tmp_ctx);
6403 if (!check_all_node_files_are_identical(ctdb, tmp_ctx, TIMELIMIT(),
6404 nodemap, file_nodemap)) {
6408 if (!sanity_check_nodes_file_changes(tmp_ctx, nodemap, file_nodemap)) {
6410 ("No change in nodes file, skipping unnecessary reload\n"));
6411 talloc_free(tmp_ctx);
6415 /* Now make the changes */
6416 conn = list_of_connected_nodes(ctdb, nodemap, tmp_ctx, true);
6417 for (i = 0; i < talloc_array_length(conn); i++) {
6418 DEBUG(DEBUG_NOTICE, ("Reloading nodes file on node %u\n",
6422 /* Another timeout could be used, such as ReRecoveryTimeout or
6423 * a new one for this purpose. However, this is the simplest
6425 timeout = options.timelimit;
6426 srvid_broadcast(ctdb, CTDB_SRVID_DISABLE_RECOVERIES, &timeout,
6427 "Disable recoveries", true);
6430 ret = ctdb_client_async_control(ctdb, CTDB_CONTROL_RELOAD_NODES_FILE,
6431 conn, 0, TIMELIMIT(),
6433 NULL, reload_nodes_fail_callback,
6437 srvid_broadcast(ctdb, CTDB_SRVID_DISABLE_RECOVERIES, &timeout,
6438 "Enable recoveries", true);
6440 talloc_free(tmp_ctx);
6446 static const struct {
6448 int (*fn)(struct ctdb_context *, int, const char **);
6450 bool without_daemon; /* can be run without daemon running ? */
6453 } ctdb_commands[] = {
6454 { "version", control_version, true, true, "show version of ctdb" },
6455 { "status", control_status, true, false, "show node status" },
6456 { "uptime", control_uptime, true, false, "show node uptime" },
6457 { "ping", control_ping, true, false, "ping all nodes" },
6458 { "runstate", control_runstate, true, false, "get/check runstate of a node", "[setup|first_recovery|startup|running]" },
6459 { "getvar", control_getvar, true, false, "get a tunable variable", "<name>"},
6460 { "setvar", control_setvar, true, false, "set a tunable variable", "<name> <value>"},
6461 { "listvars", control_listvars, true, false, "list tunable variables"},
6462 { "statistics", control_statistics, false, false, "show statistics" },
6463 { "statisticsreset", control_statistics_reset, true, false, "reset statistics"},
6464 { "stats", control_stats, false, false, "show rolling statistics", "[number of history records]" },
6465 { "ip", control_ip, false, false, "show which public ip's that ctdb manages" },
6466 { "ipinfo", control_ipinfo, true, false, "show details about a public ip that ctdb manages", "<ip>" },
6467 { "ifaces", control_ifaces, true, false, "show which interfaces that ctdb manages" },
6468 { "setifacelink", control_setifacelink, true, false, "set interface link status", "<iface> <status>" },
6469 { "process-exists", control_process_exists, true, false, "check if a process exists on a node", "<pid>"},
6470 { "getdbmap", control_getdbmap, true, false, "show the database map" },
6471 { "getdbstatus", control_getdbstatus, true, false, "show the status of a database", "<dbname|dbid>" },
6472 { "catdb", control_catdb, true, false, "dump a ctdb database" , "<dbname|dbid>"},
6473 { "cattdb", control_cattdb, true, false, "dump a local tdb database" , "<dbname|dbid>"},
6474 { "getmonmode", control_getmonmode, true, false, "show monitoring mode" },
6475 { "getcapabilities", control_getcapabilities, true, false, "show node capabilities" },
6476 { "pnn", control_pnn, true, false, "show the pnn of the currnet node" },
6477 { "lvs", control_lvs, true, false, "show lvs configuration" },
6478 { "lvsmaster", control_lvsmaster, true, false, "show which node is the lvs master" },
6479 { "disablemonitor", control_disable_monmode,true, false, "set monitoring mode to DISABLE" },
6480 { "enablemonitor", control_enable_monmode, true, false, "set monitoring mode to ACTIVE" },
6481 { "setdebug", control_setdebug, true, false, "set debug level", "<EMERG|ALERT|CRIT|ERR|WARNING|NOTICE|INFO|DEBUG>" },
6482 { "getdebug", control_getdebug, true, false, "get debug level" },
6483 { "attach", control_attach, true, false, "attach to a database", "<dbname> [persistent]" },
6484 { "detach", control_detach, false, false, "detach from a database", "<dbname|dbid> [<dbname|dbid> ...]" },
6485 { "dumpmemory", control_dumpmemory, true, false, "dump memory map to stdout" },
6486 { "rddumpmemory", control_rddumpmemory, true, false, "dump memory map from the recovery daemon to stdout" },
6487 { "getpid", control_getpid, true, false, "get ctdbd process ID" },
6488 { "disable", control_disable, true, false, "disable a nodes public IP" },
6489 { "enable", control_enable, true, false, "enable a nodes public IP" },
6490 { "stop", control_stop, true, false, "stop a node" },
6491 { "continue", control_continue, true, false, "re-start a stopped node" },
6492 { "ban", control_ban, true, false, "ban a node from the cluster", "<bantime>"},
6493 { "unban", control_unban, true, false, "unban a node" },
6494 { "showban", control_showban, true, false, "show ban information"},
6495 { "shutdown", control_shutdown, true, false, "shutdown ctdbd" },
6496 { "recover", control_recover, true, false, "force recovery" },
6497 { "sync", control_ipreallocate, false, false, "wait until ctdbd has synced all state changes" },
6498 { "ipreallocate", control_ipreallocate, false, false, "force the recovery daemon to perform a ip reallocation procedure" },
6499 { "thaw", control_thaw, true, false, "thaw databases", "[priority:1-3]" },
6500 { "isnotrecmaster", control_isnotrecmaster, false, false, "check if the local node is recmaster or not" },
6501 { "killtcp", kill_tcp, false, false, "kill a tcp connection.", "[<srcip:port> <dstip:port>]" },
6502 { "gratiousarp", control_gratious_arp, false, false, "send a gratious arp", "<ip> <interface>" },
6503 { "tickle", tickle_tcp, false, false, "send a tcp tickle ack", "<srcip:port> <dstip:port>" },
6504 { "gettickles", control_get_tickles, false, false, "get the list of tickles registered for this ip", "<ip> [<port>]" },
6505 { "addtickle", control_add_tickle, false, false, "add a tickle for this ip", "<ip>:<port> <ip>:<port>" },
6507 { "deltickle", control_del_tickle, false, false, "delete a tickle from this ip", "<ip>:<port> <ip>:<port>" },
6509 { "regsrvid", regsrvid, false, false, "register a server id", "<pnn> <type> <id>" },
6510 { "unregsrvid", unregsrvid, false, false, "unregister a server id", "<pnn> <type> <id>" },
6511 { "chksrvid", chksrvid, false, false, "check if a server id exists", "<pnn> <type> <id>" },
6512 { "getsrvids", getsrvids, false, false, "get a list of all server ids"},
6513 { "check_srvids", check_srvids, false, false, "check if a srvid exists", "<id>+" },
6514 { "repack", ctdb_repack, false, false, "repack all databases", "[max_freelist]"},
6515 { "listnodes", control_listnodes, false, true, "list all nodes in the cluster"},
6516 { "reloadnodes", control_reload_nodes_file, false, false, "reload the nodes file and restart the transport on all nodes"},
6517 { "moveip", control_moveip, false, false, "move/failover an ip address to another node", "<ip> <node>"},
6518 { "rebalanceip", control_rebalanceip, false, false, "release an ip from the node and let recd rebalance it", "<ip>"},
6519 { "addip", control_addip, true, false, "add a ip address to a node", "<ip/mask> <iface>"},
6520 { "delip", control_delip, false, false, "delete an ip address from a node", "<ip>"},
6521 { "eventscript", control_eventscript, true, false, "run the eventscript with the given parameters on a node", "<arguments>"},
6522 { "backupdb", control_backupdb, false, false, "backup the database into a file.", "<dbname|dbid> <file>"},
6523 { "restoredb", control_restoredb, false, false, "restore the database from a file.", "<file> [dbname]"},
6524 { "dumpdbbackup", control_dumpdbbackup, false, true, "dump database backup from a file.", "<file>"},
6525 { "wipedb", control_wipedb, false, false, "wipe the contents of a database.", "<dbname|dbid>"},
6526 { "recmaster", control_recmaster, true, false, "show the pnn for the recovery master."},
6527 { "scriptstatus", control_scriptstatus, true, false, "show the status of the monitoring scripts (or all scripts)", "[all]"},
6528 { "enablescript", control_enablescript, true, false, "enable an eventscript", "<script>"},
6529 { "disablescript", control_disablescript, true, false, "disable an eventscript", "<script>"},
6530 { "natgwlist", control_natgwlist, true, false, "show the nodes belonging to this natgw configuration"},
6531 { "xpnn", control_xpnn, false, true, "find the pnn of the local node without talking to the daemon (unreliable)" },
6532 { "getreclock", control_getreclock, true, false, "Show the reclock file of a node"},
6533 { "setreclock", control_setreclock, true, false, "Set/clear the reclock file of a node", "[filename]"},
6534 { "setnatgwstate", control_setnatgwstate, false, false, "Set NATGW state to on/off", "{on|off}"},
6535 { "setlmasterrole", control_setlmasterrole, false, false, "Set LMASTER role to on/off", "{on|off}"},
6536 { "setrecmasterrole", control_setrecmasterrole, false, false, "Set RECMASTER role to on/off", "{on|off}"},
6537 { "setdbprio", control_setdbprio, false, false, "Set DB priority", "<dbname|dbid> <prio:1-3>"},
6538 { "getdbprio", control_getdbprio, false, false, "Get DB priority", "<dbname|dbid>"},
6539 { "setdbreadonly", control_setdbreadonly, false, false, "Set DB readonly capable", "<dbname|dbid>"},
6540 { "setdbsticky", control_setdbsticky, false, false, "Set DB sticky-records capable", "<dbname|dbid>"},
6541 { "msglisten", control_msglisten, false, false, "Listen on a srvid port for messages", "<msg srvid>"},
6542 { "msgsend", control_msgsend, false, false, "Send a message to srvid", "<srvid> <message>"},
6543 { "pfetch", control_pfetch, false, false, "fetch a record from a persistent database", "<dbname|dbid> <key> [<file>]" },
6544 { "pstore", control_pstore, false, false, "write a record to a persistent database", "<dbname|dbid> <key> <file containing record>" },
6545 { "pdelete", control_pdelete, false, false, "delete a record from a persistent database", "<dbname|dbid> <key>" },
6546 { "ptrans", control_ptrans, false, false, "update a persistent database (from stdin)", "<dbname|dbid>" },
6547 { "tfetch", control_tfetch, false, true, "fetch a record from a [c]tdb-file [-v]", "<tdb-file> <key> [<file>]" },
6548 { "tstore", control_tstore, false, true, "store a record (including ltdb header)", "<tdb-file> <key> <data> [<rsn> <dmaster> <flags>]" },
6549 { "readkey", control_readkey, true, false, "read the content off a database key", "<dbname|dbid> <key>" },
6550 { "writekey", control_writekey, true, false, "write to a database key", "<dbname|dbid> <key> <value>" },
6551 { "checktcpport", control_chktcpport, false, true, "check if a service is bound to a specific tcp port or not", "<port>" },
6552 { "rebalancenode", control_rebalancenode, false, false, "mark nodes as forced IP rebalancing targets", "[<pnn-list>]"},
6553 { "getdbseqnum", control_getdbseqnum, false, false, "get the sequence number off a database", "<dbname|dbid>" },
6554 { "nodestatus", control_nodestatus, true, false, "show and return node status", "[<pnn-list>]" },
6555 { "dbstatistics", control_dbstatistics, false, false, "show db statistics", "<dbname|dbid>" },
6556 { "reloadips", control_reloadips, false, false, "reload the public addresses file on specified nodes" , "[<pnn-list>]" },
6557 { "ipiface", control_ipiface, false, true, "Find which interface an ip address is hosted on", "<ip>" },
6563 static void usage(void)
6567 "Usage: ctdb [options] <control>\n" \
6569 " -n <node> choose node number, or 'all' (defaults to local node)\n"
6570 " -Y generate machine readable output\n"
6571 " -x <char> specify delimiter for machine readable output\n"
6572 " -v generate verbose output\n"
6573 " -t <timelimit> set timelimit for control in seconds (default %u)\n", options.timelimit);
6574 printf("Controls:\n");
6575 for (i=0;i<ARRAY_SIZE(ctdb_commands);i++) {
6576 printf(" %-15s %-27s %s\n",
6577 ctdb_commands[i].name,
6578 ctdb_commands[i].args?ctdb_commands[i].args:"",
6579 ctdb_commands[i].msg);
6585 static void ctdb_alarm(int sig)
6587 printf("Maximum runtime exceeded - exiting\n");
6594 int main(int argc, const char *argv[])
6596 struct ctdb_context *ctdb;
6597 char *nodestring = NULL;
6598 int machineparsable = 0;
6599 struct poptOption popt_options[] = {
6602 { "timelimit", 't', POPT_ARG_INT, &options.timelimit, 0, "timelimit", "integer" },
6603 { "node", 'n', POPT_ARG_STRING, &nodestring, 0, "node", "integer|all" },
6604 { "machinereadable", 'Y', POPT_ARG_NONE, &options.machinereadable, 0, "enable machine readable output", NULL },
6605 { NULL, 'x', POPT_ARG_STRING, &options.machineseparator, 0, "specify separator for machine readable output", "char" },
6606 { NULL, 'X', POPT_ARG_NONE, &machineparsable, 0, "enable machine parsable output with separator |", NULL },
6607 { "verbose", 'v', POPT_ARG_NONE, &options.verbose, 0, "enable verbose output", NULL },
6608 { "maxruntime", 'T', POPT_ARG_INT, &options.maxruntime, 0, "die if runtime exceeds this limit (in seconds)", "integer" },
6609 { "print-emptyrecords", 0, POPT_ARG_NONE, &options.printemptyrecords, 0, "print the empty records when dumping databases (catdb, cattdb, dumpdbbackup)", NULL },
6610 { "print-datasize", 0, POPT_ARG_NONE, &options.printdatasize, 0, "do not print record data when dumping databases, only the data size", NULL },
6611 { "print-lmaster", 0, POPT_ARG_NONE, &options.printlmaster, 0, "print the record's lmaster in catdb", NULL },
6612 { "print-hash", 0, POPT_ARG_NONE, &options.printhash, 0, "print the record's hash when dumping databases", NULL },
6613 { "print-recordflags", 0, POPT_ARG_NONE, &options.printrecordflags, 0, "print the record flags in catdb and dumpdbbackup", NULL },
6617 const char **extra_argv;
6621 struct event_context *ev;
6622 const char *control;
6626 /* set some defaults */
6627 options.maxruntime = 0;
6628 options.timelimit = 10;
6629 options.pnn = CTDB_CURRENT_NODE;
6631 pc = poptGetContext(argv[0], argc, argv, popt_options, POPT_CONTEXT_KEEP_FIRST);
6633 while ((opt = poptGetNextOpt(pc)) != -1) {
6636 DEBUG(DEBUG_ERR, ("Invalid option %s: %s\n",
6637 poptBadOption(pc, 0), poptStrerror(opt)));
6642 /* setup the remaining options for the main program to use */
6643 extra_argv = poptGetArgs(pc);
6646 while (extra_argv[extra_argc]) extra_argc++;
6649 if (extra_argc < 1) {
6653 if (options.maxruntime == 0) {
6654 const char *ctdb_timeout;
6655 ctdb_timeout = getenv("CTDB_TIMEOUT");
6656 if (ctdb_timeout != NULL) {
6657 options.maxruntime = strtoul(ctdb_timeout, NULL, 0);
6659 /* default timeout is 120 seconds */
6660 options.maxruntime = 120;
6664 if (machineparsable) {
6665 options.machineseparator = "|";
6667 if (options.machineseparator != NULL) {
6668 if (strlen(options.machineseparator) != 1) {
6669 printf("Invalid separator \"%s\" - "
6670 "must be single character\n",
6671 options.machineseparator);
6676 options.machinereadable = true;
6677 } else if (options.machinereadable) {
6678 options.machineseparator = ":";
6681 signal(SIGALRM, ctdb_alarm);
6682 alarm(options.maxruntime);
6684 control = extra_argv[0];
6686 /* Default value for CTDB_BASE - don't override */
6687 setenv("CTDB_BASE", CTDB_ETCDIR, 0);
6689 ev = event_context_init(NULL);
6691 DEBUG(DEBUG_ERR, ("Failed to initialize event system\n"));
6695 for (i=0;i<ARRAY_SIZE(ctdb_commands);i++) {
6696 if (strcmp(control, ctdb_commands[i].name) == 0) {
6701 if (i == ARRAY_SIZE(ctdb_commands)) {
6702 DEBUG(DEBUG_ERR, ("Unknown control '%s'\n", control));
6706 if (ctdb_commands[i].without_daemon == true) {
6707 if (nodestring != NULL) {
6708 DEBUG(DEBUG_ERR, ("Can't specify node(s) with \"ctdb %s\"\n", control));
6711 return ctdb_commands[i].fn(NULL, extra_argc-1, extra_argv+1);
6714 /* initialise ctdb */
6715 ctdb = ctdb_cmdline_client(ev, TIMELIMIT());
6719 DEBUG(DEBUG_ERR, ("Failed to init ctdb\n"));
6721 pnn = find_node_xpnn();
6724 ("Is this node part of a CTDB cluster?\n"));
6729 /* setup the node number(s) to contact */
6730 if (!parse_nodestring(ctdb, ctdb, nodestring, CTDB_CURRENT_NODE, false,
6731 &options.nodes, &options.pnn)) {
6735 if (options.pnn == CTDB_CURRENT_NODE) {
6736 options.pnn = options.nodes[0];
6739 if (ctdb_commands[i].auto_all &&
6740 ((options.pnn == CTDB_BROADCAST_ALL) ||
6741 (options.pnn == CTDB_MULTICAST))) {
6745 for (j = 0; j < talloc_array_length(options.nodes); j++) {
6746 options.pnn = options.nodes[j];
6747 ret |= ctdb_commands[i].fn(ctdb, extra_argc-1, extra_argv+1);
6750 ret = ctdb_commands[i].fn(ctdb, extra_argc-1, extra_argv+1);
6755 (void)poptFreeContext(pc);