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"
29 /* Allow use of deprecated function tevent_loop_allow_nesting() */
30 #define TEVENT_DEPRECATED
34 #include "lib/tdb_wrap/tdb_wrap.h"
35 #include "lib/util/dlinklist.h"
36 #include "lib/util/debug.h"
37 #include "lib/util/substitute.h"
38 #include "lib/util/time.h"
40 #include "ctdb_version.h"
41 #include "ctdb_private.h"
42 #include "ctdb_client.h"
44 #include "common/cmdline.h"
45 #include "common/rb_tree.h"
46 #include "common/system.h"
47 #include "common/common.h"
48 #include "common/logging.h"
50 #define ERR_TIMEOUT 20 /* timed out trying to reach node */
51 #define ERR_NONODE 21 /* node does not exist */
52 #define ERR_DISNODE 22 /* node is disconnected */
54 static void usage(void);
61 const char *machineseparator;
64 int printemptyrecords;
71 #define LONGTIMEOUT options.timelimit*10
73 #define TIMELIMIT() timeval_current_ofs(options.timelimit, 0)
74 #define LONGTIMELIMIT() timeval_current_ofs(LONGTIMEOUT, 0)
76 static double timeval_delta(struct timeval *tv2, struct timeval *tv)
78 return (tv2->tv_sec - tv->tv_sec) +
79 (tv2->tv_usec - tv->tv_usec)*1.0e-6;
82 static int control_version(struct ctdb_context *ctdb, int argc, const char **argv)
84 printf("CTDB version: %s\n", CTDB_VERSION_STRING);
88 /* Like printf(3) but substitute for separator in format */
89 static int printm(const char *format, ...) PRINTF_ATTRIBUTE(1,2);
90 static int printm(const char *format, ...)
94 size_t len = strlen(format);
95 char new_format[len+1];
97 strcpy(new_format, format);
99 if (options.machineseparator[0] != ':') {
100 all_string_sub(new_format,
101 ":", options.machineseparator, len + 1);
104 va_start(ap, format);
105 ret = vprintf(new_format, ap);
111 #define CTDB_NOMEM_ABORT(p) do { if (!(p)) { \
112 DEBUG(DEBUG_ALERT,("ctdb fatal error: %s\n", \
113 "Out of memory in " __location__ )); \
117 static uint32_t getpnn(struct ctdb_context *ctdb)
119 if ((options.pnn == CTDB_BROADCAST_ALL) ||
120 (options.pnn == CTDB_MULTICAST)) {
122 ("Cannot get PNN for node %u\n", options.pnn));
126 if (options.pnn == CTDB_CURRENT_NODE) {
127 return ctdb_get_pnn(ctdb);
133 static void assert_single_node_only(void)
135 if ((options.pnn == CTDB_BROADCAST_ALL) ||
136 (options.pnn == CTDB_MULTICAST)) {
138 ("This control can not be applied to multiple PNNs\n"));
143 static void assert_current_node_only(struct ctdb_context *ctdb)
145 if (options.pnn != ctdb_get_pnn(ctdb)) {
147 ("This control can only be applied to the current node\n"));
152 /* Pretty print the flags to a static buffer in human-readable format.
153 * This never returns NULL!
155 static const char *pretty_print_flags(uint32_t flags)
158 static const struct {
162 { NODE_FLAGS_DISCONNECTED, "DISCONNECTED" },
163 { NODE_FLAGS_PERMANENTLY_DISABLED, "DISABLED" },
164 { NODE_FLAGS_BANNED, "BANNED" },
165 { NODE_FLAGS_UNHEALTHY, "UNHEALTHY" },
166 { NODE_FLAGS_DELETED, "DELETED" },
167 { NODE_FLAGS_STOPPED, "STOPPED" },
168 { NODE_FLAGS_INACTIVE, "INACTIVE" },
170 static char flags_str[512]; /* Big enough to contain all flag names */
173 for (j=0;j<ARRAY_SIZE(flag_names);j++) {
174 if (flags & flag_names[j].flag) {
175 if (flags_str[0] == '\0') {
176 (void) strcpy(flags_str, flag_names[j].name);
178 (void) strncat(flags_str, "|", sizeof(flags_str)-1);
179 (void) strncat(flags_str, flag_names[j].name,
180 sizeof(flags_str)-1);
184 if (flags_str[0] == '\0') {
185 (void) strcpy(flags_str, "OK");
191 static int h2i(char h)
193 if (h >= 'a' && h <= 'f') return h - 'a' + 10;
194 if (h >= 'A' && h <= 'F') return h - 'f' + 10;
198 static TDB_DATA hextodata(TALLOC_CTX *mem_ctx, const char *str, size_t len)
201 TDB_DATA key = {NULL, 0};
204 DEBUG(DEBUG_ERR,("Key specified with odd number of hexadecimal digits\n"));
209 key.dptr = talloc_size(mem_ctx, key.dsize);
211 for (i=0; i < len/2; i++) {
212 key.dptr[i] = h2i(str[i*2]) << 4 | h2i(str[i*2+1]);
217 static TDB_DATA strtodata(TALLOC_CTX *mem_ctx, const char *str, size_t len)
221 if (!strncmp(str, "0x", 2)) {
222 key = hextodata(mem_ctx, str + 2, len - 2);
224 key.dptr = talloc_memdup(mem_ctx, str, len);
231 /* Parse a nodestring. Parameter dd_ok controls what happens to nodes
232 * that are disconnected or deleted. If dd_ok is true those nodes are
233 * included in the output list of nodes. If dd_ok is false, those
234 * nodes are filtered from the "all" case and cause an error if
235 * explicitly specified.
237 static bool parse_nodestring(struct ctdb_context *ctdb,
239 const char * nodestring,
240 uint32_t current_pnn,
245 TALLOC_CTX *tmp_ctx = talloc_new(mem_ctx);
248 struct ctdb_node_map_old *nodemap;
253 ret = ctdb_ctrl_getnodemap(ctdb, TIMELIMIT(), CTDB_CURRENT_NODE, tmp_ctx, &nodemap);
255 DEBUG(DEBUG_ERR, ("Unable to get nodemap from local node\n"));
256 talloc_free(tmp_ctx);
260 if (nodestring != NULL) {
261 *nodes = talloc_array(mem_ctx, uint32_t, 0);
262 if (*nodes == NULL) {
268 if (strcmp(nodestring, "all") == 0) {
269 *pnn_mode = CTDB_BROADCAST_ALL;
272 for (i = 0; i < nodemap->num; i++) {
273 if ((nodemap->nodes[i].flags &
274 (NODE_FLAGS_DISCONNECTED |
275 NODE_FLAGS_DELETED)) && !dd_ok) {
278 *nodes = talloc_realloc(mem_ctx, *nodes,
280 if (*nodes == NULL) {
290 ns = talloc_strdup(tmp_ctx, nodestring);
291 tok = strtok(ns, ",");
292 while (tok != NULL) {
295 i = (uint32_t)strtoul(tok, &endptr, 0);
296 if (i == 0 && tok == endptr) {
298 ("Invalid node %s\n", tok));
299 talloc_free(tmp_ctx);
302 if (i >= nodemap->num) {
303 DEBUG(DEBUG_ERR, ("Node %u does not exist\n", i));
304 talloc_free(tmp_ctx);
307 if ((nodemap->nodes[i].flags &
308 (NODE_FLAGS_DISCONNECTED |
309 NODE_FLAGS_DELETED)) && !dd_ok) {
310 DEBUG(DEBUG_ERR, ("Node %u has status %s\n", i, pretty_print_flags(nodemap->nodes[i].flags)));
311 talloc_free(tmp_ctx);
314 if ((pnn = ctdb_ctrl_getpnn(ctdb, TIMELIMIT(), i)) < 0) {
315 DEBUG(DEBUG_ERR, ("Can not access node %u. Node is not operational.\n", i));
316 talloc_free(tmp_ctx);
320 *nodes = talloc_realloc(mem_ctx, *nodes,
322 if (*nodes == NULL) {
329 tok = strtok(NULL, ",");
334 *pnn_mode = (*nodes)[0];
336 *pnn_mode = CTDB_MULTICAST;
340 /* default - no nodes specified */
341 *nodes = talloc_array(mem_ctx, uint32_t, 1);
342 if (*nodes == NULL) {
345 *pnn_mode = CTDB_CURRENT_NODE;
347 if (((*nodes)[0] = ctdb_ctrl_getpnn(ctdb, TIMELIMIT(), current_pnn)) < 0) {
352 talloc_free(tmp_ctx);
356 talloc_free(tmp_ctx);
361 check if a database exists
363 static bool db_exists(struct ctdb_context *ctdb, const char *dbarg,
364 uint32_t *dbid, const char **dbname, uint8_t *flags)
367 struct ctdb_dbid_map_old *dbmap=NULL;
368 bool dbid_given = false, found = false;
370 TALLOC_CTX *tmp_ctx = talloc_new(ctdb);
371 const char *name = NULL;
373 ret = ctdb_ctrl_getdbmap(ctdb, TIMELIMIT(), options.pnn, tmp_ctx, &dbmap);
375 DEBUG(DEBUG_ERR, ("Unable to get dbids from node %u\n", options.pnn));
379 if (strncmp(dbarg, "0x", 2) == 0) {
380 id = strtoul(dbarg, NULL, 0);
384 for(i=0; i<dbmap->num; i++) {
386 if (id == dbmap->dbs[i].db_id) {
391 ret = ctdb_ctrl_getdbname(ctdb, TIMELIMIT(), options.pnn, dbmap->dbs[i].db_id, tmp_ctx, &name);
393 DEBUG(DEBUG_ERR, ("Unable to get dbname from dbid %u\n", dbmap->dbs[i].db_id));
397 if (strcmp(name, dbarg) == 0) {
398 id = dbmap->dbs[i].db_id;
405 if (found && dbid_given && dbname != NULL) {
406 ret = ctdb_ctrl_getdbname(ctdb, TIMELIMIT(), options.pnn, dbmap->dbs[i].db_id, tmp_ctx, &name);
408 DEBUG(DEBUG_ERR, ("Unable to get dbname from dbid %u\n", dbmap->dbs[i].db_id));
415 if (dbid) *dbid = id;
416 if (dbname) *dbname = talloc_strdup(ctdb, name);
417 if (flags) *flags = dbmap->dbs[i].flags;
419 DEBUG(DEBUG_ERR,("No database matching '%s' found\n", dbarg));
423 talloc_free(tmp_ctx);
428 see if a process exists
430 static int control_process_exists(struct ctdb_context *ctdb, int argc, const char **argv)
438 if (sscanf(argv[0], "%u:%u", &pnn, &pid) != 2) {
439 DEBUG(DEBUG_ERR, ("Badly formed pnn:pid\n"));
443 ret = ctdb_ctrl_process_exists(ctdb, pnn, pid);
445 printf("%u:%u exists\n", pnn, pid);
447 printf("%u:%u does not exist\n", pnn, pid);
453 display statistics structure
455 static void show_statistics(struct ctdb_statistics *s, int show_header)
457 TALLOC_CTX *tmp_ctx = talloc_new(NULL);
459 const char *prefix=NULL;
461 int tmp, days, hours, minutes, seconds;
466 #define STATISTICS_FIELD(n) { #n, offsetof(struct ctdb_statistics, n) }
467 STATISTICS_FIELD(num_clients),
468 STATISTICS_FIELD(frozen),
469 STATISTICS_FIELD(recovering),
470 STATISTICS_FIELD(num_recoveries),
471 STATISTICS_FIELD(client_packets_sent),
472 STATISTICS_FIELD(client_packets_recv),
473 STATISTICS_FIELD(node_packets_sent),
474 STATISTICS_FIELD(node_packets_recv),
475 STATISTICS_FIELD(keepalive_packets_sent),
476 STATISTICS_FIELD(keepalive_packets_recv),
477 STATISTICS_FIELD(node.req_call),
478 STATISTICS_FIELD(node.reply_call),
479 STATISTICS_FIELD(node.req_dmaster),
480 STATISTICS_FIELD(node.reply_dmaster),
481 STATISTICS_FIELD(node.reply_error),
482 STATISTICS_FIELD(node.req_message),
483 STATISTICS_FIELD(node.req_control),
484 STATISTICS_FIELD(node.reply_control),
485 STATISTICS_FIELD(client.req_call),
486 STATISTICS_FIELD(client.req_message),
487 STATISTICS_FIELD(client.req_control),
488 STATISTICS_FIELD(timeouts.call),
489 STATISTICS_FIELD(timeouts.control),
490 STATISTICS_FIELD(timeouts.traverse),
491 STATISTICS_FIELD(locks.num_calls),
492 STATISTICS_FIELD(locks.num_current),
493 STATISTICS_FIELD(locks.num_pending),
494 STATISTICS_FIELD(locks.num_failed),
495 STATISTICS_FIELD(total_calls),
496 STATISTICS_FIELD(pending_calls),
497 STATISTICS_FIELD(childwrite_calls),
498 STATISTICS_FIELD(pending_childwrite_calls),
499 STATISTICS_FIELD(memory_used),
500 STATISTICS_FIELD(max_hop_count),
501 STATISTICS_FIELD(total_ro_delegations),
502 STATISTICS_FIELD(total_ro_revokes),
505 tmp = s->statistics_current_time.tv_sec - s->statistics_start_time.tv_sec;
514 if (options.machinereadable){
516 printm("CTDB version:");
517 printm("Current time of statistics:");
518 printm("Statistics collected since:");
519 for (i=0;i<ARRAY_SIZE(fields);i++) {
520 printm("%s:", fields[i].name);
522 printm("num_reclock_ctdbd_latency:");
523 printm("min_reclock_ctdbd_latency:");
524 printm("avg_reclock_ctdbd_latency:");
525 printm("max_reclock_ctdbd_latency:");
527 printm("num_reclock_recd_latency:");
528 printm("min_reclock_recd_latency:");
529 printm("avg_reclock_recd_latency:");
530 printm("max_reclock_recd_latency:");
532 printm("num_call_latency:");
533 printm("min_call_latency:");
534 printm("avg_call_latency:");
535 printm("max_call_latency:");
537 printm("num_lockwait_latency:");
538 printm("min_lockwait_latency:");
539 printm("avg_lockwait_latency:");
540 printm("max_lockwait_latency:");
542 printm("num_childwrite_latency:");
543 printm("min_childwrite_latency:");
544 printm("avg_childwrite_latency:");
545 printm("max_childwrite_latency:");
548 printm("%d:", CTDB_PROTOCOL);
549 printm("%d:", (int)s->statistics_current_time.tv_sec);
550 printm("%d:", (int)s->statistics_start_time.tv_sec);
551 for (i=0;i<ARRAY_SIZE(fields);i++) {
552 printm("%d:", *(uint32_t *)(fields[i].offset+(uint8_t *)s));
554 printm("%d:", s->reclock.ctdbd.num);
555 printm("%.6f:", s->reclock.ctdbd.min);
556 printm("%.6f:", s->reclock.ctdbd.num?s->reclock.ctdbd.total/s->reclock.ctdbd.num:0.0);
557 printm("%.6f:", s->reclock.ctdbd.max);
559 printm("%d:", s->reclock.recd.num);
560 printm("%.6f:", s->reclock.recd.min);
561 printm("%.6f:", s->reclock.recd.num?s->reclock.recd.total/s->reclock.recd.num:0.0);
562 printm("%.6f:", s->reclock.recd.max);
564 printm("%d:", s->call_latency.num);
565 printm("%.6f:", s->call_latency.min);
566 printm("%.6f:", s->call_latency.num?s->call_latency.total/s->call_latency.num:0.0);
567 printm("%.6f:", s->call_latency.max);
569 printm("%d:", s->childwrite_latency.num);
570 printm("%.6f:", s->childwrite_latency.min);
571 printm("%.6f:", s->childwrite_latency.num?s->childwrite_latency.total/s->childwrite_latency.num:0.0);
572 printm("%.6f:", s->childwrite_latency.max);
575 printf("CTDB version %u\n", CTDB_PROTOCOL);
576 printf("Current time of statistics : %s", ctime(&s->statistics_current_time.tv_sec));
577 printf("Statistics collected since : (%03d %02d:%02d:%02d) %s", days, hours, minutes, seconds, ctime(&s->statistics_start_time.tv_sec));
579 for (i=0;i<ARRAY_SIZE(fields);i++) {
580 if (strchr(fields[i].name, '.')) {
581 preflen = strcspn(fields[i].name, ".")+1;
582 if (!prefix || strncmp(prefix, fields[i].name, preflen) != 0) {
583 prefix = fields[i].name;
584 printf(" %*.*s\n", preflen-1, preflen-1, fields[i].name);
589 printf(" %*s%-22s%*s%10u\n",
591 fields[i].name+preflen,
593 *(uint32_t *)(fields[i].offset+(uint8_t *)s));
595 printf(" hop_count_buckets:");
596 for (i=0;i<MAX_COUNT_BUCKETS;i++) {
597 printf(" %d", s->hop_count_bucket[i]);
600 printf(" lock_buckets:");
601 for (i=0; i<MAX_COUNT_BUCKETS; i++) {
602 printf(" %d", s->locks.buckets[i]);
605 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);
607 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);
609 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);
611 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);
612 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);
615 talloc_free(tmp_ctx);
619 display remote ctdb statistics combined from all nodes
621 static int control_statistics_all(struct ctdb_context *ctdb)
624 struct ctdb_statistics statistics;
628 nodes = ctdb_get_connected_nodes(ctdb, TIMELIMIT(), ctdb, &num_nodes);
629 CTDB_NO_MEMORY(ctdb, nodes);
631 ZERO_STRUCT(statistics);
633 for (i=0;i<num_nodes;i++) {
634 struct ctdb_statistics s1;
636 uint32_t *v1 = (uint32_t *)&s1;
637 uint32_t *v2 = (uint32_t *)&statistics;
639 offsetof(struct ctdb_statistics, __last_counter) / sizeof(uint32_t);
640 ret = ctdb_ctrl_statistics(ctdb, nodes[i], &s1);
642 DEBUG(DEBUG_ERR, ("Unable to get statistics from node %u\n", nodes[i]));
645 for (j=0;j<num_ints;j++) {
648 statistics.max_hop_count =
649 MAX(statistics.max_hop_count, s1.max_hop_count);
650 statistics.call_latency.max =
651 MAX(statistics.call_latency.max, s1.call_latency.max);
654 printf("Gathered statistics for %u nodes\n", num_nodes);
655 show_statistics(&statistics, 1);
660 display remote ctdb statistics
662 static int control_statistics(struct ctdb_context *ctdb, int argc, const char **argv)
665 struct ctdb_statistics statistics;
667 if (options.pnn == CTDB_BROADCAST_ALL) {
668 return control_statistics_all(ctdb);
671 ret = ctdb_ctrl_statistics(ctdb, options.pnn, &statistics);
673 DEBUG(DEBUG_ERR, ("Unable to get statistics from node %u\n", options.pnn));
676 show_statistics(&statistics, 1);
682 reset remote ctdb statistics
684 static int control_statistics_reset(struct ctdb_context *ctdb, int argc, const char **argv)
688 ret = ctdb_statistics_reset(ctdb, options.pnn);
690 DEBUG(DEBUG_ERR, ("Unable to reset statistics on node %u\n", options.pnn));
698 display remote ctdb rolling statistics
700 static int control_stats(struct ctdb_context *ctdb, int argc, const char **argv)
703 struct ctdb_statistics_list_old *stats;
704 int i, num_records = -1;
706 assert_single_node_only();
709 num_records = atoi(argv[0]) - 1;
712 ret = ctdb_ctrl_getstathistory(ctdb, TIMELIMIT(), options.pnn, ctdb, &stats);
714 DEBUG(DEBUG_ERR, ("Unable to get rolling statistics from node %u\n", options.pnn));
717 for (i=0;i<stats->num;i++) {
718 if (stats->stats[i].statistics_start_time.tv_sec == 0) {
721 show_statistics(&stats->stats[i], i==0);
722 if (i == num_records) {
731 display remote ctdb db statistics
733 static int control_dbstatistics(struct ctdb_context *ctdb, int argc, const char **argv)
735 TALLOC_CTX *tmp_ctx = talloc_new(ctdb);
736 struct ctdb_db_statistics_old *dbstat;
746 if (!db_exists(ctdb, argv[0], &db_id, NULL, NULL)) {
750 ret = ctdb_ctrl_dbstatistics(ctdb, options.pnn, db_id, tmp_ctx, &dbstat);
752 DEBUG(DEBUG_ERR,("Failed to read db statistics from node\n"));
753 talloc_free(tmp_ctx);
757 printf("DB Statistics: %s\n", argv[0]);
758 printf(" %*s%-22s%*s%10u\n", 0, "", "ro_delegations", 4, "",
759 dbstat->db_ro_delegations);
760 printf(" %*s%-22s%*s%10u\n", 0, "", "ro_revokes", 4, "",
761 dbstat->db_ro_delegations);
762 printf(" %s\n", "locks");
763 printf(" %*s%-22s%*s%10u\n", 4, "", "total", 0, "",
764 dbstat->locks.num_calls);
765 printf(" %*s%-22s%*s%10u\n", 4, "", "failed", 0, "",
766 dbstat->locks.num_failed);
767 printf(" %*s%-22s%*s%10u\n", 4, "", "current", 0, "",
768 dbstat->locks.num_current);
769 printf(" %*s%-22s%*s%10u\n", 4, "", "pending", 0, "",
770 dbstat->locks.num_pending);
771 printf(" %s", "hop_count_buckets:");
772 for (i=0; i<MAX_COUNT_BUCKETS; i++) {
773 printf(" %d", dbstat->hop_count_bucket[i]);
776 printf(" %s", "lock_buckets:");
777 for (i=0; i<MAX_COUNT_BUCKETS; i++) {
778 printf(" %d", dbstat->locks.buckets[i]);
781 printf(" %-30s %.6f/%.6f/%.6f sec out of %d\n",
782 "locks_latency MIN/AVG/MAX",
783 dbstat->locks.latency.min,
784 (dbstat->locks.latency.num ?
785 dbstat->locks.latency.total /dbstat->locks.latency.num :
787 dbstat->locks.latency.max,
788 dbstat->locks.latency.num);
789 printf(" %-30s %.6f/%.6f/%.6f sec out of %d\n",
790 "vacuum_latency MIN/AVG/MAX",
791 dbstat->vacuum.latency.min,
792 (dbstat->vacuum.latency.num ?
793 dbstat->vacuum.latency.total /dbstat->vacuum.latency.num :
795 dbstat->vacuum.latency.max,
796 dbstat->vacuum.latency.num);
798 for (i=0; i<dbstat->num_hot_keys; i++) {
799 if (dbstat->hot_keys[i].count > 0) {
803 dbstat->num_hot_keys = num_hot_keys;
805 printf(" Num Hot Keys: %d\n", dbstat->num_hot_keys);
806 for (i = 0; i < dbstat->num_hot_keys; i++) {
808 printf(" Count:%d Key:", dbstat->hot_keys[i].count);
809 for (j = 0; j < dbstat->hot_keys[i].key.dsize; j++) {
810 printf("%02x", dbstat->hot_keys[i].key.dptr[j]&0xff);
815 talloc_free(tmp_ctx);
820 display uptime of remote node
822 static int control_uptime(struct ctdb_context *ctdb, int argc, const char **argv)
825 struct ctdb_uptime *uptime = NULL;
826 int tmp, days, hours, minutes, seconds;
828 ret = ctdb_ctrl_uptime(ctdb, ctdb, TIMELIMIT(), options.pnn, &uptime);
830 DEBUG(DEBUG_ERR, ("Unable to get uptime from node %u\n", options.pnn));
834 if (options.machinereadable){
835 printm(":Current Node Time:Ctdb Start Time:Last Recovery/Failover Time:Last Recovery/IPFailover Duration:\n");
836 printm(":%u:%u:%u:%lf\n",
837 (unsigned int)uptime->current_time.tv_sec,
838 (unsigned int)uptime->ctdbd_start_time.tv_sec,
839 (unsigned int)uptime->last_recovery_finished.tv_sec,
840 timeval_delta(&uptime->last_recovery_finished,
841 &uptime->last_recovery_started)
846 printf("Current time of node : %s", ctime(&uptime->current_time.tv_sec));
848 tmp = uptime->current_time.tv_sec - uptime->ctdbd_start_time.tv_sec;
856 printf("Ctdbd start time : (%03d %02d:%02d:%02d) %s", days, hours, minutes, seconds, ctime(&uptime->ctdbd_start_time.tv_sec));
858 tmp = uptime->current_time.tv_sec - uptime->last_recovery_finished.tv_sec;
866 printf("Time of last recovery/failover: (%03d %02d:%02d:%02d) %s", days, hours, minutes, seconds, ctime(&uptime->last_recovery_finished.tv_sec));
868 printf("Duration of last recovery/failover: %lf seconds\n",
869 timeval_delta(&uptime->last_recovery_finished,
870 &uptime->last_recovery_started));
876 show the PNN of the current node
878 static int control_pnn(struct ctdb_context *ctdb, int argc, const char **argv)
882 mypnn = getpnn(ctdb);
884 printf("PNN:%d\n", mypnn);
889 static struct ctdb_node_map_old *read_nodes_file(TALLOC_CTX *mem_ctx)
891 const char *nodes_list;
893 /* read the nodes file */
894 nodes_list = getenv("CTDB_NODES");
895 if (nodes_list == NULL) {
896 nodes_list = talloc_asprintf(mem_ctx, "%s/nodes",
897 getenv("CTDB_BASE"));
898 if (nodes_list == NULL) {
899 DEBUG(DEBUG_ALERT,(__location__ " Out of memory\n"));
904 return ctdb_read_nodes_file(mem_ctx, nodes_list);
908 show the PNN of the current node
909 discover the pnn by loading the nodes file and try to bind to all
910 addresses one at a time until the ip address is found.
912 static int find_node_xpnn(void)
914 TALLOC_CTX *mem_ctx = talloc_new(NULL);
915 struct ctdb_node_map_old *node_map;
918 node_map = read_nodes_file(mem_ctx);
919 if (node_map == NULL) {
920 talloc_free(mem_ctx);
924 for (i = 0; i < node_map->num; i++) {
925 if (node_map->nodes[i].flags & NODE_FLAGS_DELETED) {
928 if (ctdb_sys_have_ip(&node_map->nodes[i].addr)) {
929 pnn = node_map->nodes[i].pnn;
930 talloc_free(mem_ctx);
935 printf("Failed to detect which PNN this node is\n");
936 talloc_free(mem_ctx);
940 static int control_xpnn(struct ctdb_context *ctdb, int argc, const char **argv)
944 assert_single_node_only();
946 pnn = find_node_xpnn();
951 printf("PNN:%d\n", pnn);
955 /* Helpers for ctdb status
957 static bool is_partially_online(struct ctdb_context *ctdb, struct ctdb_node_and_flags *node)
959 TALLOC_CTX *tmp_ctx = talloc_new(NULL);
963 if (node->flags == 0) {
964 struct ctdb_iface_list_old *ifaces;
966 if (ctdb_ctrl_get_ifaces(ctdb, TIMELIMIT(), node->pnn,
967 tmp_ctx, &ifaces) == 0) {
968 for (j=0; j < ifaces->num; j++) {
969 if (ifaces->ifaces[j].link_state != 0) {
977 talloc_free(tmp_ctx);
982 static void control_status_header_machine(void)
984 printm(":Node:IP:Disconnected:Banned:Disabled:Unhealthy:Stopped"
985 ":Inactive:PartiallyOnline:ThisNode:\n");
988 static int control_status_1_machine(struct ctdb_context *ctdb, int mypnn,
989 struct ctdb_node_and_flags *node)
991 printm(":%d:%s:%d:%d:%d:%d:%d:%d:%d:%c:\n", node->pnn,
992 ctdb_addr_to_str(&node->addr),
993 !!(node->flags&NODE_FLAGS_DISCONNECTED),
994 !!(node->flags&NODE_FLAGS_BANNED),
995 !!(node->flags&NODE_FLAGS_PERMANENTLY_DISABLED),
996 !!(node->flags&NODE_FLAGS_UNHEALTHY),
997 !!(node->flags&NODE_FLAGS_STOPPED),
998 !!(node->flags&NODE_FLAGS_INACTIVE),
999 is_partially_online(ctdb, node) ? 1 : 0,
1000 (node->pnn == mypnn)?'Y':'N');
1005 static int control_status_1_human(struct ctdb_context *ctdb, int mypnn,
1006 struct ctdb_node_and_flags *node)
1008 printf("pnn:%d %-16s %s%s\n", node->pnn,
1009 ctdb_addr_to_str(&node->addr),
1010 is_partially_online(ctdb, node) ? "PARTIALLYONLINE" : pretty_print_flags(node->flags),
1011 node->pnn == mypnn?" (THIS NODE)":"");
1017 display remote ctdb status
1019 static int control_status(struct ctdb_context *ctdb, int argc, const char **argv)
1021 TALLOC_CTX *tmp_ctx = talloc_new(ctdb);
1023 struct ctdb_vnn_map *vnnmap=NULL;
1024 struct ctdb_node_map_old *nodemap=NULL;
1025 uint32_t recmode, recmaster, mypnn;
1026 int num_deleted_nodes = 0;
1029 mypnn = getpnn(ctdb);
1031 ret = ctdb_ctrl_getnodemap(ctdb, TIMELIMIT(), options.pnn, tmp_ctx, &nodemap);
1033 DEBUG(DEBUG_ERR, ("Unable to get nodemap from node %u\n", options.pnn));
1034 talloc_free(tmp_ctx);
1038 if (options.machinereadable) {
1039 control_status_header_machine();
1040 for (i=0;i<nodemap->num;i++) {
1041 if (nodemap->nodes[i].flags & NODE_FLAGS_DELETED) {
1044 (void) control_status_1_machine(ctdb, mypnn,
1045 &nodemap->nodes[i]);
1047 talloc_free(tmp_ctx);
1051 for (i=0; i<nodemap->num; i++) {
1052 if (nodemap->nodes[i].flags & NODE_FLAGS_DELETED) {
1053 num_deleted_nodes++;
1056 if (num_deleted_nodes == 0) {
1057 printf("Number of nodes:%d\n", nodemap->num);
1059 printf("Number of nodes:%d (including %d deleted nodes)\n",
1060 nodemap->num, num_deleted_nodes);
1062 for(i=0;i<nodemap->num;i++){
1063 if (nodemap->nodes[i].flags & NODE_FLAGS_DELETED) {
1066 (void) control_status_1_human(ctdb, mypnn, &nodemap->nodes[i]);
1069 ret = ctdb_ctrl_getvnnmap(ctdb, TIMELIMIT(), options.pnn, tmp_ctx, &vnnmap);
1071 DEBUG(DEBUG_ERR, ("Unable to get vnnmap from node %u\n", options.pnn));
1072 talloc_free(tmp_ctx);
1075 if (vnnmap->generation == INVALID_GENERATION) {
1076 printf("Generation:INVALID\n");
1078 printf("Generation:%d\n",vnnmap->generation);
1080 printf("Size:%d\n",vnnmap->size);
1081 for(i=0;i<vnnmap->size;i++){
1082 printf("hash:%d lmaster:%d\n", i, vnnmap->map[i]);
1085 ret = ctdb_ctrl_getrecmode(ctdb, tmp_ctx, TIMELIMIT(), options.pnn, &recmode);
1087 DEBUG(DEBUG_ERR, ("Unable to get recmode from node %u\n", options.pnn));
1088 talloc_free(tmp_ctx);
1091 printf("Recovery mode:%s (%d)\n",recmode==CTDB_RECOVERY_NORMAL?"NORMAL":"RECOVERY",recmode);
1093 ret = ctdb_ctrl_getrecmaster(ctdb, tmp_ctx, TIMELIMIT(), options.pnn, &recmaster);
1095 DEBUG(DEBUG_ERR, ("Unable to get recmaster from node %u\n", options.pnn));
1096 talloc_free(tmp_ctx);
1099 printf("Recovery master:%d\n",recmaster);
1101 talloc_free(tmp_ctx);
1105 static int control_nodestatus(struct ctdb_context *ctdb, int argc, const char **argv)
1107 TALLOC_CTX *tmp_ctx = talloc_new(ctdb);
1109 struct ctdb_node_map_old *nodemap=NULL;
1111 uint32_t pnn_mode, mypnn;
1117 if (!parse_nodestring(ctdb, tmp_ctx, argc == 1 ? argv[0] : NULL,
1118 options.pnn, true, &nodes, &pnn_mode)) {
1122 if (options.machinereadable) {
1123 control_status_header_machine();
1124 } else if (pnn_mode == CTDB_BROADCAST_ALL) {
1125 printf("Number of nodes:%d\n", (int) talloc_array_length(nodes));
1128 mypnn = getpnn(ctdb);
1130 ret = ctdb_ctrl_getnodemap(ctdb, TIMELIMIT(), options.pnn, tmp_ctx, &nodemap);
1132 DEBUG(DEBUG_ERR, ("Unable to get nodemap from node %u\n", options.pnn));
1133 talloc_free(tmp_ctx);
1139 for (i = 0; i < talloc_array_length(nodes); i++) {
1140 if (options.machinereadable) {
1141 ret |= control_status_1_machine(ctdb, mypnn,
1142 &nodemap->nodes[nodes[i]]);
1144 ret |= control_status_1_human(ctdb, mypnn,
1145 &nodemap->nodes[nodes[i]]);
1149 talloc_free(tmp_ctx);
1153 /* talloc off the existing nodemap... */
1154 static struct ctdb_node_map_old *talloc_nodemap(struct ctdb_node_map_old *nodemap)
1156 return talloc_zero_size(nodemap,
1157 offsetof(struct ctdb_node_map_old, nodes) +
1158 nodemap->num * sizeof(struct ctdb_node_and_flags));
1161 static struct ctdb_node_map_old *
1162 filter_nodemap_by_capabilities(struct ctdb_context *ctdb,
1163 struct ctdb_node_map_old *nodemap,
1164 uint32_t required_capabilities,
1168 uint32_t capabilities;
1169 struct ctdb_node_map_old *ret;
1171 ret = talloc_nodemap(nodemap);
1172 CTDB_NO_MEMORY_NULL(ctdb, ret);
1176 for (i = 0; i < nodemap->num; i++) {
1179 /* Disconnected nodes have no capabilities! */
1180 if (nodemap->nodes[i].flags & NODE_FLAGS_DISCONNECTED) {
1184 res = ctdb_ctrl_getcapabilities(ctdb, TIMELIMIT(),
1185 nodemap->nodes[i].pnn,
1188 DEBUG(DEBUG_ERR, ("Unable to get capabilities from node %u\n",
1189 nodemap->nodes[i].pnn));
1193 if (!(capabilities & required_capabilities)) {
1197 ret->nodes[ret->num] = nodemap->nodes[i];
1207 static struct ctdb_node_map_old *
1208 filter_nodemap_by_flags(struct ctdb_context *ctdb,
1209 struct ctdb_node_map_old *nodemap,
1210 uint32_t flags_mask)
1213 struct ctdb_node_map_old *ret;
1215 ret = talloc_nodemap(nodemap);
1216 CTDB_NO_MEMORY_NULL(ctdb, ret);
1220 for (i = 0; i < nodemap->num; i++) {
1221 if (nodemap->nodes[i].flags & flags_mask) {
1225 ret->nodes[ret->num] = nodemap->nodes[i];
1233 display the list of nodes belonging to this natgw configuration
1235 static int control_natgwlist(struct ctdb_context *ctdb, int argc, const char **argv)
1237 static char prog[PATH_MAX+1] = "";
1243 if (!ctdb_set_helper("NAT gateway helper", prog, sizeof(prog),
1244 "CTDB_NATGW_HELPER", CTDB_HELPER_BINDIR,
1246 DEBUG(DEBUG_ERR, ("Unable to set NAT gateway helper\n"));
1250 execl(prog, prog, "natgwlist", NULL);
1253 ("Unable to run NAT gateway helper %s\n", strerror(errno)));
1258 display the status of the scripts for monitoring (or other events)
1260 static int control_one_scriptstatus(struct ctdb_context *ctdb,
1261 enum ctdb_event type)
1263 struct ctdb_script_list_old *script_status;
1266 ret = ctdb_ctrl_getscriptstatus(ctdb, TIMELIMIT(), options.pnn, ctdb, type, &script_status);
1268 DEBUG(DEBUG_ERR, ("Unable to get script status from node %u\n", options.pnn));
1272 if (script_status == NULL) {
1273 if (!options.machinereadable) {
1274 printf("%s cycle never run\n",
1275 ctdb_eventscript_call_names[type]);
1280 if (!options.machinereadable) {
1282 for (i=0; i<script_status->num_scripts; i++) {
1283 if (script_status->scripts[i].status != -ENOEXEC) {
1287 printf("%d scripts were executed last %s cycle\n",
1289 ctdb_eventscript_call_names[type]);
1291 for (i=0; i<script_status->num_scripts; i++) {
1292 const char *status = NULL;
1294 /* The ETIME status is ignored for certain events.
1295 * In that case the status is 0, but endtime is not set.
1297 if (script_status->scripts[i].status == 0 &&
1298 timeval_is_zero(&script_status->scripts[i].finished)) {
1299 script_status->scripts[i].status = -ETIME;
1302 switch (script_status->scripts[i].status) {
1304 status = "TIMEDOUT";
1307 status = "DISABLED";
1313 if (script_status->scripts[i].status > 0)
1317 if (options.machinereadable) {
1318 printm(":%s:%s:%i:%s:%lu.%06lu:%lu.%06lu:%s:\n",
1319 ctdb_eventscript_call_names[type],
1320 script_status->scripts[i].name,
1321 script_status->scripts[i].status,
1323 (long)script_status->scripts[i].start.tv_sec,
1324 (long)script_status->scripts[i].start.tv_usec,
1325 (long)script_status->scripts[i].finished.tv_sec,
1326 (long)script_status->scripts[i].finished.tv_usec,
1327 script_status->scripts[i].output);
1331 printf("%-20s Status:%s ",
1332 script_status->scripts[i].name, status);
1334 /* Some other error, eg from stat. */
1335 printf("%-20s Status:CANNOT RUN (%s)",
1336 script_status->scripts[i].name,
1337 strerror(-script_status->scripts[i].status));
1339 if (script_status->scripts[i].status >= 0) {
1340 printf("Duration:%.3lf ",
1341 timeval_delta(&script_status->scripts[i].finished,
1342 &script_status->scripts[i].start));
1344 if (script_status->scripts[i].status != -ENOEXEC) {
1346 ctime(&script_status->scripts[i].start.tv_sec));
1347 if (script_status->scripts[i].status != 0) {
1348 printf(" OUTPUT:%s\n",
1349 script_status->scripts[i].output);
1359 static int control_scriptstatus(struct ctdb_context *ctdb,
1360 int argc, const char **argv)
1363 enum ctdb_event type, min, max;
1367 DEBUG(DEBUG_ERR, ("Unknown arguments to scriptstatus\n"));
1372 arg = ctdb_eventscript_call_names[CTDB_EVENT_MONITOR];
1376 for (type = 0; type < CTDB_EVENT_MAX; type++) {
1377 if (strcmp(arg, ctdb_eventscript_call_names[type]) == 0) {
1383 if (type == CTDB_EVENT_MAX) {
1384 if (strcmp(arg, "all") == 0) {
1386 max = CTDB_EVENT_MAX;
1388 DEBUG(DEBUG_ERR, ("Unknown event type %s\n", argv[0]));
1393 if (options.machinereadable) {
1394 printm(":Type:Name:Code:Status:Start:End:Error Output...:\n");
1397 for (type = min; type < max; type++) {
1398 ret = control_one_scriptstatus(ctdb, type);
1408 enable an eventscript
1410 static int control_enablescript(struct ctdb_context *ctdb, int argc, const char **argv)
1418 ret = ctdb_ctrl_enablescript(ctdb, TIMELIMIT(), options.pnn, argv[0]);
1420 DEBUG(DEBUG_ERR, ("Unable to enable script %s on node %u\n", argv[0], options.pnn));
1428 disable an eventscript
1430 static int control_disablescript(struct ctdb_context *ctdb, int argc, const char **argv)
1438 ret = ctdb_ctrl_disablescript(ctdb, TIMELIMIT(), options.pnn, argv[0]);
1440 DEBUG(DEBUG_ERR, ("Unable to disable script %s on node %u\n", argv[0], options.pnn));
1448 display the pnn of the recovery master
1450 static int control_recmaster(struct ctdb_context *ctdb, int argc, const char **argv)
1455 ret = ctdb_ctrl_getrecmaster(ctdb, ctdb, TIMELIMIT(), options.pnn, &recmaster);
1457 DEBUG(DEBUG_ERR, ("Unable to get recmaster from node %u\n", options.pnn));
1460 printf("%d\n",recmaster);
1466 add a tickle to a public address
1468 static int control_add_tickle(struct ctdb_context *ctdb, int argc, const char **argv)
1470 struct ctdb_connection t;
1474 assert_single_node_only();
1480 if (parse_ip_port(argv[0], &t.src) == 0) {
1481 DEBUG(DEBUG_ERR,("Wrongly formed ip address '%s'\n", argv[0]));
1484 if (parse_ip_port(argv[1], &t.dst) == 0) {
1485 DEBUG(DEBUG_ERR,("Wrongly formed ip address '%s'\n", argv[1]));
1489 data.dptr = (uint8_t *)&t;
1490 data.dsize = sizeof(t);
1492 /* tell all nodes about this tcp connection */
1493 ret = ctdb_control(ctdb, options.pnn, 0, CTDB_CONTROL_TCP_ADD_DELAYED_UPDATE,
1494 0, data, ctdb, NULL, NULL, NULL, NULL);
1496 DEBUG(DEBUG_ERR,("Failed to add tickle\n"));
1505 delete a tickle from a node
1507 static int control_del_tickle(struct ctdb_context *ctdb, int argc, const char **argv)
1509 struct ctdb_connection t;
1513 assert_single_node_only();
1519 if (parse_ip_port(argv[0], &t.src) == 0) {
1520 DEBUG(DEBUG_ERR,("Wrongly formed ip address '%s'\n", argv[0]));
1523 if (parse_ip_port(argv[1], &t.dst) == 0) {
1524 DEBUG(DEBUG_ERR,("Wrongly formed ip address '%s'\n", argv[1]));
1528 data.dptr = (uint8_t *)&t;
1529 data.dsize = sizeof(t);
1531 /* tell all nodes about this tcp connection */
1532 ret = ctdb_control(ctdb, options.pnn, 0, CTDB_CONTROL_TCP_REMOVE,
1533 0, data, ctdb, NULL, NULL, NULL, NULL);
1535 DEBUG(DEBUG_ERR,("Failed to remove tickle\n"));
1544 get a list of all tickles for this pnn
1546 static int control_get_tickles(struct ctdb_context *ctdb, int argc, const char **argv)
1548 struct ctdb_tickle_list_old *list;
1549 ctdb_sock_addr addr;
1553 assert_single_node_only();
1560 port = atoi(argv[1]);
1563 if (parse_ip(argv[0], NULL, 0, &addr) == 0) {
1564 DEBUG(DEBUG_ERR,("Wrongly formed ip address '%s'\n", argv[0]));
1568 ret = ctdb_ctrl_get_tcp_tickles(ctdb, TIMELIMIT(), options.pnn, ctdb, &addr, &list);
1570 DEBUG(DEBUG_ERR, ("Unable to list tickles\n"));
1574 if (options.machinereadable){
1575 printm(":source ip:port:destination ip:port:\n");
1576 for (i=0;i<list->num;i++) {
1577 if (port && port != ntohs(list->connections[i].dst.ip.sin_port)) {
1580 printm(":%s:%u", ctdb_addr_to_str(&list->connections[i].src), ntohs(list->connections[i].src.ip.sin_port));
1581 printm(":%s:%u:\n", ctdb_addr_to_str(&list->connections[i].dst), ntohs(list->connections[i].dst.ip.sin_port));
1584 printf("Tickles for ip:%s\n", ctdb_addr_to_str(&list->addr));
1585 printf("Num tickles:%u\n", list->num);
1586 for (i=0;i<list->num;i++) {
1587 if (port && port != ntohs(list->connections[i].dst.ip.sin_port)) {
1590 printf("SRC: %s:%u ", ctdb_addr_to_str(&list->connections[i].src), ntohs(list->connections[i].src.ip.sin_port));
1591 printf("DST: %s:%u\n", ctdb_addr_to_str(&list->connections[i].dst), ntohs(list->connections[i].dst.ip.sin_port));
1601 static int move_ip(struct ctdb_context *ctdb, ctdb_sock_addr *addr, uint32_t pnn)
1603 struct ctdb_public_ip_list_old *ips;
1604 struct ctdb_public_ip ip;
1607 uint32_t disable_time;
1609 struct ctdb_node_map_old *nodemap=NULL;
1610 TALLOC_CTX *tmp_ctx = talloc_new(ctdb);
1613 data.dptr = (uint8_t*)&disable_time;
1614 data.dsize = sizeof(disable_time);
1615 ret = ctdb_client_send_message(ctdb, CTDB_BROADCAST_CONNECTED, CTDB_SRVID_DISABLE_IP_CHECK, data);
1617 DEBUG(DEBUG_ERR,("Failed to send message to disable ipcheck\n"));
1623 /* read the public ip list from the node */
1624 ret = ctdb_ctrl_get_public_ips(ctdb, TIMELIMIT(), pnn, ctdb, &ips);
1626 DEBUG(DEBUG_ERR, ("Unable to get public ip list from node %u\n", pnn));
1627 talloc_free(tmp_ctx);
1631 for (i=0;i<ips->num;i++) {
1632 if (ctdb_same_ip(addr, &ips->ips[i].addr)) {
1637 DEBUG(DEBUG_ERR, ("Node %u can not host ip address '%s'\n",
1638 pnn, ctdb_addr_to_str(addr)));
1639 talloc_free(tmp_ctx);
1646 data.dptr = (uint8_t *)&ip;
1647 data.dsize = sizeof(ip);
1649 ret = ctdb_ctrl_getnodemap(ctdb, TIMELIMIT(), options.pnn, tmp_ctx, &nodemap);
1651 DEBUG(DEBUG_ERR, ("Unable to get nodemap from node %u\n", options.pnn));
1652 talloc_free(tmp_ctx);
1656 nodes = list_of_nodes(ctdb, nodemap, tmp_ctx, NODE_FLAGS_INACTIVE, pnn);
1657 ret = ctdb_client_async_control(ctdb, CTDB_CONTROL_RELEASE_IP,
1664 DEBUG(DEBUG_ERR,("Failed to release IP on nodes\n"));
1665 talloc_free(tmp_ctx);
1669 ret = ctdb_ctrl_takeover_ip(ctdb, LONGTIMELIMIT(), pnn, &ip);
1671 DEBUG(DEBUG_ERR,("Failed to take over IP on node %d\n", pnn));
1672 talloc_free(tmp_ctx);
1676 /* update the recovery daemon so it now knows to expect the new
1677 node assignment for this ip.
1679 ret = ctdb_client_send_message(ctdb, CTDB_BROADCAST_CONNECTED, CTDB_SRVID_RECD_UPDATE_IP, data);
1681 DEBUG(DEBUG_ERR,("Failed to send message to update the ip on the recovery master.\n"));
1685 talloc_free(tmp_ctx);
1691 * scans all other nodes and returns a pnn for another node that can host this
1695 find_other_host_for_public_ip(struct ctdb_context *ctdb, ctdb_sock_addr *addr)
1697 TALLOC_CTX *tmp_ctx = talloc_new(ctdb);
1698 struct ctdb_public_ip_list_old *ips;
1699 struct ctdb_node_map_old *nodemap=NULL;
1703 ret = ctdb_ctrl_getnodemap(ctdb, TIMELIMIT(), CTDB_CURRENT_NODE, tmp_ctx, &nodemap);
1705 DEBUG(DEBUG_ERR, ("Unable to get nodemap from node %u\n", options.pnn));
1706 talloc_free(tmp_ctx);
1710 for(i=0;i<nodemap->num;i++){
1711 if (nodemap->nodes[i].flags & NODE_FLAGS_INACTIVE) {
1714 if (nodemap->nodes[i].pnn == options.pnn) {
1718 /* read the public ip list from this node */
1719 ret = ctdb_ctrl_get_public_ips(ctdb, TIMELIMIT(), nodemap->nodes[i].pnn, tmp_ctx, &ips);
1721 DEBUG(DEBUG_ERR, ("Unable to get public ip list from node %u\n", nodemap->nodes[i].pnn));
1725 for (j=0;j<ips->num;j++) {
1726 if (ctdb_same_ip(addr, &ips->ips[j].addr)) {
1727 pnn = nodemap->nodes[i].pnn;
1728 talloc_free(tmp_ctx);
1735 talloc_free(tmp_ctx);
1739 /* If pnn is -1 then try to find a node to move IP to... */
1740 static bool try_moveip(struct ctdb_context *ctdb, ctdb_sock_addr *addr, uint32_t pnn)
1742 bool pnn_specified = (pnn == -1 ? false : true);
1745 while (retries < 5) {
1746 if (!pnn_specified) {
1747 pnn = find_other_host_for_public_ip(ctdb, addr);
1752 ("Trying to move public IP to node %u\n", pnn));
1755 if (move_ip(ctdb, addr, pnn) == 0) {
1768 move/failover an ip address to a specific node
1770 static int control_moveip(struct ctdb_context *ctdb, int argc, const char **argv)
1773 ctdb_sock_addr addr;
1775 assert_single_node_only();
1782 if (parse_ip(argv[0], NULL, 0, &addr) == 0) {
1783 DEBUG(DEBUG_ERR,("Wrongly formed ip address '%s'\n", argv[0]));
1788 if (sscanf(argv[1], "%u", &pnn) != 1) {
1789 DEBUG(DEBUG_ERR, ("Badly formed pnn\n"));
1793 if (!try_moveip(ctdb, &addr, pnn)) {
1794 DEBUG(DEBUG_ERR,("Failed to move IP to node %d.\n", pnn));
1801 static int rebalance_node(struct ctdb_context *ctdb, uint32_t pnn)
1805 data.dptr = (uint8_t *)&pnn;
1806 data.dsize = sizeof(uint32_t);
1807 if (ctdb_client_send_message(ctdb, CTDB_BROADCAST_CONNECTED, CTDB_SRVID_REBALANCE_NODE, data) != 0) {
1809 ("Failed to send message to force node %u to be a rebalancing target\n",
1819 rebalance a node by setting it to allow failback and triggering a
1822 static int control_rebalancenode(struct ctdb_context *ctdb, int argc, const char **argv)
1824 TALLOC_CTX *tmp_ctx = talloc_new(ctdb);
1829 assert_single_node_only();
1835 /* Determine the nodes where IPs need to be reloaded */
1836 if (!parse_nodestring(ctdb, tmp_ctx, argc == 1 ? argv[0] : NULL,
1837 options.pnn, true, &nodes, &pnn_mode)) {
1842 for (i = 0; i < talloc_array_length(nodes); i++) {
1843 if (!rebalance_node(ctdb, nodes[i])) {
1849 talloc_free(tmp_ctx);
1853 static int rebalance_ip(struct ctdb_context *ctdb, ctdb_sock_addr *addr)
1855 struct ctdb_public_ip ip;
1858 uint32_t disable_time;
1860 struct ctdb_node_map_old *nodemap=NULL;
1861 TALLOC_CTX *tmp_ctx = talloc_new(ctdb);
1864 data.dptr = (uint8_t*)&disable_time;
1865 data.dsize = sizeof(disable_time);
1866 ret = ctdb_client_send_message(ctdb, CTDB_BROADCAST_CONNECTED, CTDB_SRVID_DISABLE_IP_CHECK, data);
1868 DEBUG(DEBUG_ERR,("Failed to send message to disable ipcheck\n"));
1875 data.dptr = (uint8_t *)&ip;
1876 data.dsize = sizeof(ip);
1878 ret = ctdb_ctrl_getnodemap(ctdb, TIMELIMIT(), options.pnn, tmp_ctx, &nodemap);
1880 DEBUG(DEBUG_ERR, ("Unable to get nodemap from node %u\n", options.pnn));
1881 talloc_free(tmp_ctx);
1885 nodes = list_of_active_nodes(ctdb, nodemap, tmp_ctx, true);
1886 ret = ctdb_client_async_control(ctdb, CTDB_CONTROL_RELEASE_IP,
1893 DEBUG(DEBUG_ERR,("Failed to release IP on nodes\n"));
1894 talloc_free(tmp_ctx);
1898 talloc_free(tmp_ctx);
1903 release an ip form all nodes and have it re-assigned by recd
1905 static int control_rebalanceip(struct ctdb_context *ctdb, int argc, const char **argv)
1907 ctdb_sock_addr addr;
1909 assert_single_node_only();
1916 if (parse_ip(argv[0], NULL, 0, &addr) == 0) {
1917 DEBUG(DEBUG_ERR,("Wrongly formed ip address '%s'\n", argv[0]));
1921 if (rebalance_ip(ctdb, &addr) != 0) {
1922 DEBUG(DEBUG_ERR,("Error when trying to reassign ip\n"));
1929 static int getips_store_callback(void *param, void *data)
1931 struct ctdb_public_ip *node_ip = (struct ctdb_public_ip *)data;
1932 struct ctdb_public_ip_list_old *ips = param;
1936 ips->ips[i].pnn = node_ip->pnn;
1937 ips->ips[i].addr = node_ip->addr;
1941 static int getips_count_callback(void *param, void *data)
1943 uint32_t *count = param;
1950 static uint32_t *ip_key(ctdb_sock_addr *ip)
1952 static uint32_t key[IP_KEYLEN];
1954 bzero(key, sizeof(key));
1956 switch (ip->sa.sa_family) {
1958 key[0] = ip->ip.sin_addr.s_addr;
1961 uint32_t *s6_a32 = (uint32_t *)&(ip->ip6.sin6_addr.s6_addr);
1969 DEBUG(DEBUG_ERR, (__location__ " ERROR, unknown family passed :%u\n", ip->sa.sa_family));
1976 static void *add_ip_callback(void *parm, void *data)
1982 control_get_all_public_ips(struct ctdb_context *ctdb, TALLOC_CTX *tmp_ctx, struct ctdb_public_ip_list_old **ips)
1984 struct ctdb_public_ip_list_old *tmp_ips;
1985 struct ctdb_node_map_old *nodemap=NULL;
1986 trbt_tree_t *ip_tree;
1990 ret = ctdb_ctrl_getnodemap(ctdb, TIMELIMIT(), CTDB_CURRENT_NODE, tmp_ctx, &nodemap);
1992 DEBUG(DEBUG_ERR, ("Unable to get nodemap from node %u\n", options.pnn));
1996 ip_tree = trbt_create(tmp_ctx, 0);
1998 for(i=0;i<nodemap->num;i++){
1999 if (nodemap->nodes[i].flags & NODE_FLAGS_DELETED) {
2002 if (nodemap->nodes[i].flags & NODE_FLAGS_DISCONNECTED) {
2006 /* read the public ip list from this node */
2007 ret = ctdb_ctrl_get_public_ips(ctdb, TIMELIMIT(), nodemap->nodes[i].pnn, tmp_ctx, &tmp_ips);
2009 DEBUG(DEBUG_ERR, ("Unable to get public ip list from node %u\n", nodemap->nodes[i].pnn));
2013 for (j=0; j<tmp_ips->num;j++) {
2014 struct ctdb_public_ip *node_ip;
2016 node_ip = talloc(tmp_ctx, struct ctdb_public_ip);
2017 node_ip->pnn = tmp_ips->ips[j].pnn;
2018 node_ip->addr = tmp_ips->ips[j].addr;
2020 trbt_insertarray32_callback(ip_tree,
2021 IP_KEYLEN, ip_key(&tmp_ips->ips[j].addr),
2025 talloc_free(tmp_ips);
2030 trbt_traversearray32(ip_tree, IP_KEYLEN, getips_count_callback, &count);
2032 len = offsetof(struct ctdb_public_ip_list_old, ips) +
2033 count*sizeof(struct ctdb_public_ip);
2034 tmp_ips = talloc_zero_size(tmp_ctx, len);
2035 trbt_traversearray32(ip_tree, IP_KEYLEN, getips_store_callback, tmp_ips);
2043 static void ctdb_every_second(struct tevent_context *ev,
2044 struct tevent_timer *te,
2045 struct timeval t, void *p)
2047 struct ctdb_context *ctdb = talloc_get_type(p, struct ctdb_context);
2049 tevent_add_timer(ctdb->ev, ctdb, timeval_current_ofs(1, 0),
2050 ctdb_every_second, ctdb);
2053 struct srvid_reply_handler_data {
2057 const char *srvid_str;
2060 static void srvid_broadcast_reply_handler(uint64_t srvid, TDB_DATA data,
2063 struct srvid_reply_handler_data *d =
2064 (struct srvid_reply_handler_data *)private_data;
2068 if (data.dsize != sizeof(ret)) {
2069 DEBUG(DEBUG_ERR, (__location__ " Wrong reply size\n"));
2073 /* ret will be a PNN (i.e. >=0) on success, or negative on error */
2074 ret = *(int32_t *)data.dptr;
2077 ("%s failed with result %d\n", d->srvid_str, ret));
2081 if (!d->wait_for_all) {
2086 /* Wait for all replies */
2088 for (i = 0; i < talloc_array_length(d->nodes); i++) {
2089 if (d->nodes[i] == ret) {
2091 ("%s reply received from node %u\n",
2092 d->srvid_str, ret));
2095 if (d->nodes[i] != -1) {
2096 /* Found a node that hasn't yet replied */
2102 /* Broadcast the given SRVID to all connected nodes. Wait for 1 reply
2103 * or replies from all connected nodes. arg is the data argument to
2104 * pass in the srvid_request structure - pass 0 if this isn't needed.
2106 static int srvid_broadcast(struct ctdb_context *ctdb,
2107 uint64_t srvid, uint32_t *arg,
2108 const char *srvid_str, bool wait_for_all)
2113 uint64_t reply_srvid;
2114 struct ctdb_srvid_message request;
2115 struct ctdb_disable_message request_data;
2116 struct srvid_reply_handler_data reply_data;
2119 ZERO_STRUCT(request);
2121 /* Time ticks to enable timeouts to be processed */
2122 tevent_add_timer(ctdb->ev, ctdb, timeval_current_ofs(1, 0),
2123 ctdb_every_second, ctdb);
2125 pnn = ctdb_get_pnn(ctdb);
2126 reply_srvid = getpid();
2130 request.srvid = reply_srvid;
2132 data.dptr = (uint8_t *)&request;
2133 data.dsize = sizeof(request);
2135 request_data.pnn = pnn;
2136 request_data.srvid = reply_srvid;
2137 request_data.timeout = *arg;
2139 data.dptr = (uint8_t *)&request_data;
2140 data.dsize = sizeof(request_data);
2143 /* Register message port for reply from recovery master */
2144 ctdb_client_set_message_handler(ctdb, reply_srvid,
2145 srvid_broadcast_reply_handler,
2148 reply_data.wait_for_all = wait_for_all;
2149 reply_data.nodes = NULL;
2150 reply_data.srvid_str = srvid_str;
2153 reply_data.done = false;
2156 struct ctdb_node_map_old *nodemap;
2158 ret = ctdb_ctrl_getnodemap(ctdb, TIMELIMIT(),
2159 CTDB_CURRENT_NODE, ctdb, &nodemap);
2162 ("Unable to get nodemap from current node, try again\n"));
2167 if (reply_data.nodes != NULL) {
2168 talloc_free(reply_data.nodes);
2170 reply_data.nodes = list_of_connected_nodes(ctdb, nodemap,
2173 talloc_free(nodemap);
2176 /* Send to all connected nodes. Only recmaster replies */
2177 ret = ctdb_client_send_message(ctdb, CTDB_BROADCAST_CONNECTED,
2180 /* This can only happen if the socket is closed and
2181 * there's no way to recover from that, so don't try
2185 ("Failed to send %s request to connected nodes\n",
2190 tv = timeval_current();
2191 /* This loop terminates the reply is received */
2192 while (timeval_elapsed(&tv) < 5.0 && !reply_data.done) {
2193 tevent_loop_once(ctdb->ev);
2196 if (!reply_data.done) {
2198 ("Still waiting for confirmation of %s\n", srvid_str));
2203 ctdb_client_remove_message_handler(ctdb, reply_srvid, &reply_data);
2205 talloc_free(reply_data.nodes);
2210 static int ipreallocate(struct ctdb_context *ctdb)
2212 return srvid_broadcast(ctdb, CTDB_SRVID_TAKEOVER_RUN, NULL,
2213 "IP reallocation", false);
2217 static int control_ipreallocate(struct ctdb_context *ctdb, int argc, const char **argv)
2219 return ipreallocate(ctdb);
2223 add a public ip address to a node
2225 static int control_addip(struct ctdb_context *ctdb, int argc, const char **argv)
2228 int len, retries = 0;
2230 ctdb_sock_addr addr;
2231 struct ctdb_addr_info_old *pub;
2232 TALLOC_CTX *tmp_ctx = talloc_new(ctdb);
2233 struct ctdb_public_ip_list_old *ips;
2237 talloc_free(tmp_ctx);
2241 if (!parse_ip_mask(argv[0], argv[1], &addr, &mask)) {
2242 DEBUG(DEBUG_ERR, ("Badly formed ip/mask : %s\n", argv[0]));
2243 talloc_free(tmp_ctx);
2247 /* read the public ip list from the node */
2248 ret = ctdb_ctrl_get_public_ips(ctdb, TIMELIMIT(), options.pnn, tmp_ctx, &ips);
2250 DEBUG(DEBUG_ERR, ("Unable to get public ip list from node %u\n", options.pnn));
2251 talloc_free(tmp_ctx);
2254 for (i=0;i<ips->num;i++) {
2255 if (ctdb_same_ip(&addr, &ips->ips[i].addr)) {
2256 DEBUG(DEBUG_ERR,("Can not add ip to node. Node already hosts this ip\n"));
2263 /* Dont timeout. This command waits for an ip reallocation
2264 which sometimes can take wuite a while if there has
2265 been a recent recovery
2269 len = offsetof(struct ctdb_addr_info_old, iface) + strlen(argv[1]) + 1;
2270 pub = talloc_size(tmp_ctx, len);
2271 CTDB_NO_MEMORY(ctdb, pub);
2275 pub->len = strlen(argv[1])+1;
2276 memcpy(&pub->iface[0], argv[1], strlen(argv[1])+1);
2279 ret = ctdb_ctrl_add_public_ip(ctdb, TIMELIMIT(), options.pnn, pub);
2281 DEBUG(DEBUG_ERR, ("Unable to add public ip to node %u. Wait 3 seconds and try again.\n", options.pnn));
2285 } while (retries < 5 && ret != 0);
2287 DEBUG(DEBUG_ERR, ("Unable to add public ip to node %u. Giving up.\n", options.pnn));
2288 talloc_free(tmp_ctx);
2292 if (rebalance_node(ctdb, options.pnn) != 0) {
2293 DEBUG(DEBUG_ERR,("Error when trying to rebalance node\n"));
2297 talloc_free(tmp_ctx);
2302 add a public ip address to a node
2304 static int control_ipiface(struct ctdb_context *ctdb, int argc, const char **argv)
2306 ctdb_sock_addr addr;
2313 if (!parse_ip(argv[0], NULL, 0, &addr)) {
2314 printf("Badly formed ip : %s\n", argv[0]);
2318 iface = ctdb_sys_find_ifname(&addr);
2319 if (iface == NULL) {
2320 printf("Failed to get interface name for ip: %s", argv[0]);
2324 printf("IP on interface %s\n", iface);
2331 static int control_delip(struct ctdb_context *ctdb, int argc, const char **argv);
2333 static int control_delip_all(struct ctdb_context *ctdb, int argc, const char **argv, ctdb_sock_addr *addr)
2335 TALLOC_CTX *tmp_ctx = talloc_new(ctdb);
2336 struct ctdb_node_map_old *nodemap=NULL;
2337 struct ctdb_public_ip_list_old *ips;
2340 ret = ctdb_ctrl_getnodemap(ctdb, TIMELIMIT(), CTDB_CURRENT_NODE, tmp_ctx, &nodemap);
2342 DEBUG(DEBUG_ERR, ("Unable to get nodemap from current node\n"));
2346 /* remove it from the nodes that are not hosting the ip currently */
2347 for(i=0;i<nodemap->num;i++){
2348 if (nodemap->nodes[i].flags & NODE_FLAGS_INACTIVE) {
2351 ret = ctdb_ctrl_get_public_ips(ctdb, TIMELIMIT(), nodemap->nodes[i].pnn, tmp_ctx, &ips);
2353 DEBUG(DEBUG_ERR, ("Unable to get public ip list from node %d\n", nodemap->nodes[i].pnn));
2357 for (j=0;j<ips->num;j++) {
2358 if (ctdb_same_ip(addr, &ips->ips[j].addr)) {
2366 if (ips->ips[j].pnn == nodemap->nodes[i].pnn) {
2370 options.pnn = nodemap->nodes[i].pnn;
2371 control_delip(ctdb, argc, argv);
2375 /* remove it from every node (also the one hosting it) */
2376 for(i=0;i<nodemap->num;i++){
2377 if (nodemap->nodes[i].flags & NODE_FLAGS_INACTIVE) {
2380 ret = ctdb_ctrl_get_public_ips(ctdb, TIMELIMIT(), nodemap->nodes[i].pnn, tmp_ctx, &ips);
2382 DEBUG(DEBUG_ERR, ("Unable to get public ip list from node %d\n", nodemap->nodes[i].pnn));
2386 for (j=0;j<ips->num;j++) {
2387 if (ctdb_same_ip(addr, &ips->ips[j].addr)) {
2395 options.pnn = nodemap->nodes[i].pnn;
2396 control_delip(ctdb, argc, argv);
2399 talloc_free(tmp_ctx);
2404 delete a public ip address from a node
2406 static int control_delip(struct ctdb_context *ctdb, int argc, const char **argv)
2409 ctdb_sock_addr addr;
2410 struct ctdb_addr_info_old pub;
2411 TALLOC_CTX *tmp_ctx = talloc_new(ctdb);
2412 struct ctdb_public_ip_list_old *ips;
2415 talloc_free(tmp_ctx);
2419 if (parse_ip(argv[0], NULL, 0, &addr) == 0) {
2420 DEBUG(DEBUG_ERR,("Wrongly formed ip address '%s'\n", argv[0]));
2424 if (options.pnn == CTDB_BROADCAST_ALL) {
2425 return control_delip_all(ctdb, argc, argv, &addr);
2432 ret = ctdb_ctrl_get_public_ips(ctdb, TIMELIMIT(), options.pnn, tmp_ctx, &ips);
2434 DEBUG(DEBUG_ERR, ("Unable to get public ip list from cluster\n"));
2435 talloc_free(tmp_ctx);
2439 for (i=0;i<ips->num;i++) {
2440 if (ctdb_same_ip(&addr, &ips->ips[i].addr)) {
2446 DEBUG(DEBUG_ERR, ("This node does not support this public address '%s'\n",
2447 ctdb_addr_to_str(&addr)));
2448 talloc_free(tmp_ctx);
2452 /* This is an optimisation. If this node is hosting the IP
2453 * then try to move it somewhere else without invoking a full
2454 * takeover run. We don't care if this doesn't work!
2456 if (ips->ips[i].pnn == options.pnn) {
2457 (void) try_moveip(ctdb, &addr, -1);
2460 ret = ctdb_ctrl_del_public_ip(ctdb, TIMELIMIT(), options.pnn, &pub);
2462 DEBUG(DEBUG_ERR, ("Unable to del public ip from node %u\n", options.pnn));
2463 talloc_free(tmp_ctx);
2467 talloc_free(tmp_ctx);
2471 static int kill_tcp_from_file(struct ctdb_context *ctdb,
2472 int argc, const char **argv)
2474 struct ctdb_connection *killtcp;
2475 int max_entries, current, i;
2476 struct timeval timeout;
2477 char line[128], src[128], dst[128];
2480 struct client_async_data *async_data;
2481 struct ctdb_client_control_state *state;
2491 while (!feof(stdin)) {
2492 if (fgets(line, sizeof(line), stdin) == NULL) {
2496 /* Silently skip empty lines */
2497 if (line[0] == '\n') {
2501 if (sscanf(line, "%s %s\n", src, dst) != 2) {
2502 DEBUG(DEBUG_ERR, ("Bad line [%d]: '%s'\n",
2504 talloc_free(killtcp);
2508 if (current >= max_entries) {
2509 max_entries += 1024;
2510 killtcp = talloc_realloc(ctdb, killtcp,
2511 struct ctdb_connection,
2513 CTDB_NO_MEMORY(ctdb, killtcp);
2516 if (!parse_ip_port(src, &killtcp[current].src)) {
2517 DEBUG(DEBUG_ERR, ("Bad IP:port on line [%d]: '%s'\n",
2519 talloc_free(killtcp);
2523 if (!parse_ip_port(dst, &killtcp[current].dst)) {
2524 DEBUG(DEBUG_ERR, ("Bad IP:port on line [%d]: '%s'\n",
2526 talloc_free(killtcp);
2533 async_data = talloc_zero(ctdb, struct client_async_data);
2534 if (async_data == NULL) {
2535 talloc_free(killtcp);
2539 for (i = 0; i < current; i++) {
2541 data.dsize = sizeof(struct ctdb_connection);
2542 data.dptr = (unsigned char *)&killtcp[i];
2544 timeout = TIMELIMIT();
2545 state = ctdb_control_send(ctdb, options.pnn, 0,
2546 CTDB_CONTROL_KILL_TCP, 0, data,
2547 async_data, &timeout, NULL);
2549 if (state == NULL) {
2551 ("Failed to call async killtcp control to node %u\n",
2553 talloc_free(killtcp);
2557 ctdb_client_async_add(async_data, state);
2560 if (ctdb_client_async_wait(ctdb, async_data) != 0) {
2561 DEBUG(DEBUG_ERR,("killtcp failed\n"));
2562 talloc_free(killtcp);
2566 talloc_free(killtcp);
2572 kill a tcp connection
2574 static int kill_tcp(struct ctdb_context *ctdb, int argc, const char **argv)
2577 struct ctdb_connection killtcp;
2579 assert_single_node_only();
2582 return kill_tcp_from_file(ctdb, argc, argv);
2589 if (!parse_ip_port(argv[0], &killtcp.src)) {
2590 DEBUG(DEBUG_ERR, ("Bad IP:port '%s'\n", argv[0]));
2594 if (!parse_ip_port(argv[1], &killtcp.dst)) {
2595 DEBUG(DEBUG_ERR, ("Bad IP:port '%s'\n", argv[1]));
2599 ret = ctdb_ctrl_killtcp(ctdb, TIMELIMIT(), options.pnn, &killtcp);
2601 DEBUG(DEBUG_ERR, ("Unable to killtcp from node %u\n", options.pnn));
2612 static int control_gratious_arp(struct ctdb_context *ctdb, int argc, const char **argv)
2615 ctdb_sock_addr addr;
2617 assert_single_node_only();
2623 if (!parse_ip(argv[0], NULL, 0, &addr)) {
2624 DEBUG(DEBUG_ERR, ("Bad IP '%s'\n", argv[0]));
2628 ret = ctdb_ctrl_gratious_arp(ctdb, TIMELIMIT(), options.pnn, &addr, argv[1]);
2630 DEBUG(DEBUG_ERR, ("Unable to send gratious_arp from node %u\n", options.pnn));
2638 register a server id
2640 static int regsrvid(struct ctdb_context *ctdb, int argc, const char **argv)
2643 struct ctdb_client_id server_id;
2649 server_id.pnn = strtoul(argv[0], NULL, 0);
2650 server_id.type = strtoul(argv[1], NULL, 0);
2651 server_id.server_id = strtoul(argv[2], NULL, 0);
2653 ret = ctdb_ctrl_register_server_id(ctdb, TIMELIMIT(), &server_id);
2655 DEBUG(DEBUG_ERR, ("Unable to register server_id from node %u\n", options.pnn));
2658 DEBUG(DEBUG_ERR,("Srvid registered. Sleeping for 999 seconds\n"));
2664 unregister a server id
2666 static int unregsrvid(struct ctdb_context *ctdb, int argc, const char **argv)
2669 struct ctdb_client_id server_id;
2675 server_id.pnn = strtoul(argv[0], NULL, 0);
2676 server_id.type = strtoul(argv[1], NULL, 0);
2677 server_id.server_id = strtoul(argv[2], NULL, 0);
2679 ret = ctdb_ctrl_unregister_server_id(ctdb, TIMELIMIT(), &server_id);
2681 DEBUG(DEBUG_ERR, ("Unable to unregister server_id from node %u\n", options.pnn));
2688 check if a server id exists
2690 static int chksrvid(struct ctdb_context *ctdb, int argc, const char **argv)
2692 uint32_t status = 0;
2694 struct ctdb_client_id server_id;
2700 server_id.pnn = strtoul(argv[0], NULL, 0);
2701 server_id.type = strtoul(argv[1], NULL, 0);
2702 server_id.server_id = strtoul(argv[2], NULL, 0);
2704 ret = ctdb_ctrl_check_server_id(ctdb, TIMELIMIT(), options.pnn, &server_id, &status);
2706 DEBUG(DEBUG_ERR, ("Unable to check server_id from node %u\n", options.pnn));
2711 printf("Server id %d:%d:%d EXISTS\n", server_id.pnn, server_id.type, server_id.server_id);
2713 printf("Server id %d:%d:%d does NOT exist\n", server_id.pnn, server_id.type, server_id.server_id);
2719 get a list of all server ids that are registered on a node
2721 static int getsrvids(struct ctdb_context *ctdb, int argc, const char **argv)
2724 struct ctdb_client_id_list_old *server_ids;
2726 ret = ctdb_ctrl_get_server_id_list(ctdb, ctdb, TIMELIMIT(), options.pnn, &server_ids);
2728 DEBUG(DEBUG_ERR, ("Unable to get server_id list from node %u\n", options.pnn));
2732 for (i=0; i<server_ids->num; i++) {
2733 printf("Server id %d:%d:%d\n",
2734 server_ids->server_ids[i].pnn,
2735 server_ids->server_ids[i].type,
2736 server_ids->server_ids[i].server_id);
2743 check if a server id exists
2745 static int check_srvids(struct ctdb_context *ctdb, int argc, const char **argv)
2747 TALLOC_CTX *tmp_ctx = talloc_new(NULL);
2753 talloc_free(tmp_ctx);
2757 ids = talloc_array(tmp_ctx, uint64_t, argc);
2758 result = talloc_array(tmp_ctx, uint8_t, argc);
2760 for (i = 0; i < argc; i++) {
2761 ids[i] = strtoull(argv[i], NULL, 0);
2764 if (!ctdb_client_check_message_handlers(ctdb, ids, argc, result)) {
2765 DEBUG(DEBUG_ERR, ("Unable to check server_id from node %u\n",
2767 talloc_free(tmp_ctx);
2771 for (i=0; i < argc; i++) {
2772 printf("Server id %d:%llu %s\n", options.pnn, (long long)ids[i],
2773 result[i] ? "exists" : "does not exist");
2776 talloc_free(tmp_ctx);
2781 send a tcp tickle ack
2783 static int tickle_tcp(struct ctdb_context *ctdb, int argc, const char **argv)
2786 ctdb_sock_addr src, dst;
2792 if (!parse_ip_port(argv[0], &src)) {
2793 DEBUG(DEBUG_ERR, ("Bad IP:port '%s'\n", argv[0]));
2797 if (!parse_ip_port(argv[1], &dst)) {
2798 DEBUG(DEBUG_ERR, ("Bad IP:port '%s'\n", argv[1]));
2802 ret = ctdb_sys_send_tcp(&src, &dst, 0, 0, 0);
2806 DEBUG(DEBUG_ERR, ("Error while sending tickle ack\n"));
2813 display public ip status
2815 static int control_ip(struct ctdb_context *ctdb, int argc, const char **argv)
2818 TALLOC_CTX *tmp_ctx = talloc_new(ctdb);
2819 struct ctdb_public_ip_list_old *ips;
2821 if (argc == 1 && strcmp(argv[0], "all") == 0) {
2822 options.pnn = CTDB_BROADCAST_ALL;
2825 if (options.pnn == CTDB_BROADCAST_ALL) {
2826 /* read the list of public ips from all nodes */
2827 ret = control_get_all_public_ips(ctdb, tmp_ctx, &ips);
2829 /* read the public ip list from this node */
2830 ret = ctdb_ctrl_get_public_ips(ctdb, TIMELIMIT(), options.pnn, tmp_ctx, &ips);
2833 DEBUG(DEBUG_ERR, ("Unable to get public ips from node %u\n", options.pnn));
2834 talloc_free(tmp_ctx);
2838 if (options.machinereadable){
2839 printm(":Public IP:Node:");
2840 if (options.verbose){
2841 printm("ActiveInterface:AvailableInterfaces:ConfiguredInterfaces:");
2845 if (options.pnn == CTDB_BROADCAST_ALL) {
2846 printf("Public IPs on ALL nodes\n");
2848 printf("Public IPs on node %u\n", options.pnn);
2852 for (i=1;i<=ips->num;i++) {
2853 struct ctdb_public_ip_info_old *info = NULL;
2855 char *aciface = NULL;
2856 char *avifaces = NULL;
2857 char *cifaces = NULL;
2859 if (options.pnn == CTDB_BROADCAST_ALL) {
2860 pnn = ips->ips[ips->num-i].pnn;
2866 ret = ctdb_ctrl_get_public_ip_info(ctdb, TIMELIMIT(), pnn, ctdb,
2867 &ips->ips[ips->num-i].addr, &info);
2874 for (j=0; j < info->num; j++) {
2875 if (cifaces == NULL) {
2876 cifaces = talloc_strdup(info,
2877 info->ifaces[j].name);
2879 cifaces = talloc_asprintf_append(cifaces,
2881 info->ifaces[j].name);
2884 if (info->active_idx == j) {
2885 aciface = info->ifaces[j].name;
2888 if (info->ifaces[j].link_state == 0) {
2892 if (avifaces == NULL) {
2893 avifaces = talloc_strdup(info, info->ifaces[j].name);
2895 avifaces = talloc_asprintf_append(avifaces,
2897 info->ifaces[j].name);
2902 if (options.machinereadable){
2904 ctdb_addr_to_str(&ips->ips[ips->num-i].addr),
2905 ips->ips[ips->num-i].pnn);
2906 if (options.verbose){
2909 avifaces?avifaces:"",
2910 cifaces?cifaces:"");
2914 if (options.verbose) {
2915 printf("%s node[%d] active[%s] available[%s] configured[%s]\n",
2916 ctdb_addr_to_str(&ips->ips[ips->num-i].addr),
2917 ips->ips[ips->num-i].pnn,
2919 avifaces?avifaces:"",
2920 cifaces?cifaces:"");
2923 ctdb_addr_to_str(&ips->ips[ips->num-i].addr),
2924 ips->ips[ips->num-i].pnn);
2930 talloc_free(tmp_ctx);
2937 static int control_ipinfo(struct ctdb_context *ctdb, int argc, const char **argv)
2940 ctdb_sock_addr addr;
2941 TALLOC_CTX *tmp_ctx = talloc_new(ctdb);
2942 struct ctdb_public_ip_info_old *info;
2945 talloc_free(tmp_ctx);
2949 if (parse_ip(argv[0], NULL, 0, &addr) == 0) {
2950 DEBUG(DEBUG_ERR,("Wrongly formed ip address '%s'\n", argv[0]));
2954 /* read the public ip info from this node */
2955 ret = ctdb_ctrl_get_public_ip_info(ctdb, TIMELIMIT(), options.pnn,
2956 tmp_ctx, &addr, &info);
2958 DEBUG(DEBUG_ERR, ("Unable to get public ip[%s]info from node %u\n",
2959 argv[0], options.pnn));
2960 talloc_free(tmp_ctx);
2964 printf("Public IP[%s] info on node %u\n",
2965 ctdb_addr_to_str(&info->ip.addr),
2968 printf("IP:%s\nCurrentNode:%d\nNumInterfaces:%u\n",
2969 ctdb_addr_to_str(&info->ip.addr),
2970 info->ip.pnn, info->num);
2972 for (i=0; i<info->num; i++) {
2973 info->ifaces[i].name[CTDB_IFACE_SIZE] = '\0';
2975 printf("Interface[%u]: Name:%s Link:%s References:%u%s\n",
2976 i+1, info->ifaces[i].name,
2977 info->ifaces[i].link_state?"up":"down",
2978 (unsigned int)info->ifaces[i].references,
2979 (i==info->active_idx)?" (active)":"");
2982 talloc_free(tmp_ctx);
2987 display interfaces status
2989 static int control_ifaces(struct ctdb_context *ctdb, int argc, const char **argv)
2991 TALLOC_CTX *tmp_ctx = talloc_new(ctdb);
2993 struct ctdb_iface_list_old *ifaces;
2996 /* read the public ip list from this node */
2997 ret = ctdb_ctrl_get_ifaces(ctdb, TIMELIMIT(), options.pnn, tmp_ctx, &ifaces);
2999 DEBUG(DEBUG_ERR, ("Unable to get interfaces from node %u\n",
3001 talloc_free(tmp_ctx);
3005 if (options.machinereadable){
3006 printm(":Name:LinkStatus:References:\n");
3008 printf("Interfaces on node %u\n", options.pnn);
3011 for (i=0; i<ifaces->num; i++) {
3012 if (options.machinereadable){
3013 printm(":%s:%s:%u:\n",
3014 ifaces->ifaces[i].name,
3015 ifaces->ifaces[i].link_state?"1":"0",
3016 (unsigned int)ifaces->ifaces[i].references);
3018 printf("name:%s link:%s references:%u\n",
3019 ifaces->ifaces[i].name,
3020 ifaces->ifaces[i].link_state?"up":"down",
3021 (unsigned int)ifaces->ifaces[i].references);
3025 talloc_free(tmp_ctx);
3031 set link status of an interface
3033 static int control_setifacelink(struct ctdb_context *ctdb, int argc, const char **argv)
3036 TALLOC_CTX *tmp_ctx = talloc_new(ctdb);
3037 struct ctdb_iface info;
3045 if (strlen(argv[0]) > CTDB_IFACE_SIZE) {
3046 DEBUG(DEBUG_ERR, ("interfaces name '%s' too long\n",
3048 talloc_free(tmp_ctx);
3051 strcpy(info.name, argv[0]);
3053 if (strcmp(argv[1], "up") == 0) {
3054 info.link_state = 1;
3055 } else if (strcmp(argv[1], "down") == 0) {
3056 info.link_state = 0;
3058 DEBUG(DEBUG_ERR, ("link state invalid '%s' should be 'up' or 'down'\n",
3060 talloc_free(tmp_ctx);
3064 /* read the public ip list from this node */
3065 ret = ctdb_ctrl_set_iface_link(ctdb, TIMELIMIT(), options.pnn,
3068 DEBUG(DEBUG_ERR, ("Unable to set link state for interfaces %s node %u\n",
3069 argv[0], options.pnn));
3070 talloc_free(tmp_ctx);
3074 talloc_free(tmp_ctx);
3079 display pid of a ctdb daemon
3081 static int control_getpid(struct ctdb_context *ctdb, int argc, const char **argv)
3086 ret = ctdb_ctrl_getpid(ctdb, TIMELIMIT(), options.pnn, &pid);
3088 DEBUG(DEBUG_ERR, ("Unable to get daemon pid from node %u\n", options.pnn));
3091 printf("Pid:%d\n", pid);
3096 typedef bool update_flags_handler_t(struct ctdb_context *ctdb, void *data);
3098 static int update_flags_and_ipreallocate(struct ctdb_context *ctdb,
3100 update_flags_handler_t handler,
3105 struct ctdb_node_map_old *nodemap = NULL;
3109 /* Check if the node is already in the desired state */
3110 ret = ctdb_ctrl_getnodemap(ctdb, TIMELIMIT(), CTDB_CURRENT_NODE, ctdb, &nodemap);
3112 DEBUG(DEBUG_ERR, ("Unable to get nodemap from local node\n"));
3115 flag_is_set = nodemap->nodes[options.pnn].flags & flag;
3116 if (set_flag == flag_is_set) {
3117 DEBUG(DEBUG_NOTICE, ("Node %d is %s %s\n", options.pnn,
3118 (set_flag ? "already" : "not"), desc));
3123 if (!handler(ctdb, data)) {
3124 DEBUG(DEBUG_WARNING,
3125 ("Failed to send control to set state %s on node %u, try again\n",
3126 desc, options.pnn));
3131 /* Read the nodemap and verify the change took effect.
3132 * Even if the above control/hanlder timed out then it
3133 * could still have worked!
3135 ret = ctdb_ctrl_getnodemap(ctdb, TIMELIMIT(), CTDB_CURRENT_NODE,
3138 DEBUG(DEBUG_WARNING,
3139 ("Unable to get nodemap from local node, try again\n"));
3141 flag_is_set = nodemap->nodes[options.pnn].flags & flag;
3142 } while (nodemap == NULL || (set_flag != flag_is_set));
3144 return ipreallocate(ctdb);
3147 /* Administratively disable a node */
3148 static bool update_flags_disabled(struct ctdb_context *ctdb, void *data)
3152 ret = ctdb_ctrl_modflags(ctdb, TIMELIMIT(), options.pnn,
3153 NODE_FLAGS_PERMANENTLY_DISABLED, 0);
3157 static int control_disable(struct ctdb_context *ctdb, int argc, const char **argv)
3159 return update_flags_and_ipreallocate(ctdb, NULL,
3160 update_flags_disabled,
3161 NODE_FLAGS_PERMANENTLY_DISABLED,
3163 true /* set_flag*/);
3166 /* Administratively re-enable a node */
3167 static bool update_flags_not_disabled(struct ctdb_context *ctdb, void *data)
3171 ret = ctdb_ctrl_modflags(ctdb, TIMELIMIT(), options.pnn,
3172 0, NODE_FLAGS_PERMANENTLY_DISABLED);
3176 static int control_enable(struct ctdb_context *ctdb, int argc, const char **argv)
3178 return update_flags_and_ipreallocate(ctdb, NULL,
3179 update_flags_not_disabled,
3180 NODE_FLAGS_PERMANENTLY_DISABLED,
3182 false /* set_flag*/);
3186 static bool update_flags_stopped(struct ctdb_context *ctdb, void *data)
3190 ret = ctdb_ctrl_stop_node(ctdb, TIMELIMIT(), options.pnn);
3195 static int control_stop(struct ctdb_context *ctdb, int argc, const char **argv)
3197 return update_flags_and_ipreallocate(ctdb, NULL,
3198 update_flags_stopped,
3201 true /* set_flag*/);
3204 /* Continue a stopped node */
3205 static bool update_flags_not_stopped(struct ctdb_context *ctdb, void *data)
3209 ret = ctdb_ctrl_continue_node(ctdb, TIMELIMIT(), options.pnn);
3214 static int control_continue(struct ctdb_context *ctdb, int argc, const char **argv)
3216 return update_flags_and_ipreallocate(ctdb, NULL,
3217 update_flags_not_stopped,
3220 false /* set_flag */);
3223 static uint32_t get_generation(struct ctdb_context *ctdb)
3225 TALLOC_CTX *tmp_ctx = talloc_new(ctdb);
3226 struct ctdb_vnn_map *vnnmap=NULL;
3228 uint32_t generation;
3230 /* wait until the recmaster is not in recovery mode */
3232 uint32_t recmode, recmaster;
3234 if (vnnmap != NULL) {
3235 talloc_free(vnnmap);
3239 /* get the recmaster */
3240 ret = ctdb_ctrl_getrecmaster(ctdb, tmp_ctx, TIMELIMIT(), CTDB_CURRENT_NODE, &recmaster);
3242 DEBUG(DEBUG_ERR, ("Unable to get recmaster from node %u\n", options.pnn));
3243 talloc_free(tmp_ctx);
3247 /* get recovery mode */
3248 ret = ctdb_ctrl_getrecmode(ctdb, tmp_ctx, TIMELIMIT(), recmaster, &recmode);
3250 DEBUG(DEBUG_ERR, ("Unable to get recmode from node %u\n", options.pnn));
3251 talloc_free(tmp_ctx);
3255 /* get the current generation number */
3256 ret = ctdb_ctrl_getvnnmap(ctdb, TIMELIMIT(), recmaster, tmp_ctx, &vnnmap);
3258 DEBUG(DEBUG_ERR, ("Unable to get vnnmap from recmaster (%u)\n", recmaster));
3259 talloc_free(tmp_ctx);
3263 if ((recmode == CTDB_RECOVERY_NORMAL) && (vnnmap->generation != 1)) {
3264 generation = vnnmap->generation;
3265 talloc_free(tmp_ctx);
3273 static bool update_state_banned(struct ctdb_context *ctdb, void *data)
3275 struct ctdb_ban_state *bantime = (struct ctdb_ban_state *)data;
3278 ret = ctdb_ctrl_set_ban(ctdb, TIMELIMIT(), options.pnn, bantime);
3283 static int control_ban(struct ctdb_context *ctdb, int argc, const char **argv)
3285 struct ctdb_ban_state bantime;
3291 bantime.pnn = options.pnn;
3292 bantime.time = strtoul(argv[0], NULL, 0);
3294 if (bantime.time == 0) {
3295 DEBUG(DEBUG_ERR, ("Invalid ban time specified - must be >0\n"));
3299 return update_flags_and_ipreallocate(ctdb, &bantime,
3300 update_state_banned,
3303 true /* set_flag*/);
3308 static int control_unban(struct ctdb_context *ctdb, int argc, const char **argv)
3310 struct ctdb_ban_state bantime;
3312 bantime.pnn = options.pnn;
3315 return update_flags_and_ipreallocate(ctdb, &bantime,
3316 update_state_banned,
3319 false /* set_flag*/);
3323 show ban information for a node
3325 static int control_showban(struct ctdb_context *ctdb, int argc, const char **argv)
3328 struct ctdb_node_map_old *nodemap=NULL;
3329 struct ctdb_ban_state *bantime;
3331 /* verify the node exists */
3332 ret = ctdb_ctrl_getnodemap(ctdb, TIMELIMIT(), CTDB_CURRENT_NODE, ctdb, &nodemap);
3334 DEBUG(DEBUG_ERR, ("Unable to get nodemap from local node\n"));
3338 ret = ctdb_ctrl_get_ban(ctdb, TIMELIMIT(), options.pnn, ctdb, &bantime);
3340 DEBUG(DEBUG_ERR,("Showing ban info for node %d failed.\n", options.pnn));
3344 if (bantime->time == 0) {
3345 printf("Node %u is not banned\n", bantime->pnn);
3347 printf("Node %u is banned, %d seconds remaining\n",
3348 bantime->pnn, bantime->time);
3357 static int control_shutdown(struct ctdb_context *ctdb, int argc, const char **argv)
3361 ret = ctdb_ctrl_shutdown(ctdb, TIMELIMIT(), options.pnn);
3363 DEBUG(DEBUG_ERR, ("Unable to shutdown node %u\n", options.pnn));
3373 static int control_recover(struct ctdb_context *ctdb, int argc, const char **argv)
3376 uint32_t generation, next_generation;
3378 /* record the current generation number */
3379 generation = get_generation(ctdb);
3381 ret = ctdb_ctrl_setrecmode(ctdb, TIMELIMIT(), options.pnn, CTDB_RECOVERY_ACTIVE);
3383 DEBUG(DEBUG_ERR, ("Unable to set recovery mode\n"));
3387 /* wait until we are in a new generation */
3389 next_generation = get_generation(ctdb);
3390 if (next_generation != generation) {
3401 display monitoring mode of a remote node
3403 static int control_getmonmode(struct ctdb_context *ctdb, int argc, const char **argv)
3408 ret = ctdb_ctrl_getmonmode(ctdb, TIMELIMIT(), options.pnn, &monmode);
3410 DEBUG(DEBUG_ERR, ("Unable to get monmode from node %u\n", options.pnn));
3413 if (!options.machinereadable){
3414 printf("Monitoring mode:%s (%d)\n",monmode==CTDB_MONITORING_ACTIVE?"ACTIVE":"DISABLED",monmode);
3417 printm(":%d:\n",monmode);
3424 display capabilities of a remote node
3426 static int control_getcapabilities(struct ctdb_context *ctdb, int argc, const char **argv)
3428 uint32_t capabilities;
3431 ret = ctdb_ctrl_getcapabilities(ctdb, TIMELIMIT(), options.pnn, &capabilities);
3433 DEBUG(DEBUG_ERR, ("Unable to get capabilities from node %u\n", options.pnn));
3437 if (!options.machinereadable){
3438 printf("RECMASTER: %s\n", (capabilities&CTDB_CAP_RECMASTER)?"YES":"NO");
3439 printf("LMASTER: %s\n", (capabilities&CTDB_CAP_LMASTER)?"YES":"NO");
3440 printf("LVS: %s\n", (capabilities&CTDB_CAP_LVS)?"YES":"NO");
3441 printf("NATGW: %s\n", (capabilities&CTDB_CAP_NATGW)?"YES":"NO");
3443 printm(":RECMASTER:LMASTER:LVS:NATGW:\n");
3444 printm(":%d:%d:%d:%d:\n",
3445 !!(capabilities&CTDB_CAP_RECMASTER),
3446 !!(capabilities&CTDB_CAP_LMASTER),
3447 !!(capabilities&CTDB_CAP_LVS),
3448 !!(capabilities&CTDB_CAP_NATGW));
3454 display lvs configuration
3457 static uint32_t lvs_exclude_flags[] = {
3458 /* Look for a nice healthy node */
3459 NODE_FLAGS_INACTIVE|NODE_FLAGS_DISABLED,
3460 /* If not found, an UNHEALTHY node will do */
3461 NODE_FLAGS_INACTIVE|NODE_FLAGS_PERMANENTLY_DISABLED,
3465 static int control_lvs(struct ctdb_context *ctdb, int argc, const char **argv)
3467 TALLOC_CTX *tmp_ctx = talloc_new(ctdb);
3468 struct ctdb_node_map_old *orig_nodemap=NULL;
3469 struct ctdb_node_map_old *nodemap;
3472 ret = ctdb_ctrl_getnodemap(ctdb, TIMELIMIT(), options.pnn,
3473 tmp_ctx, &orig_nodemap);
3475 DEBUG(DEBUG_ERR, ("Unable to get nodemap from node %u\n", options.pnn));
3476 talloc_free(tmp_ctx);
3480 nodemap = filter_nodemap_by_capabilities(ctdb, orig_nodemap,
3481 CTDB_CAP_LVS, false);
3482 if (nodemap == NULL) {
3490 for (i = 0; lvs_exclude_flags[i] != 0; i++) {
3491 struct ctdb_node_map_old *t =
3492 filter_nodemap_by_flags(ctdb, nodemap,
3493 lvs_exclude_flags[i]);
3500 /* At least 1 node without excluded flags */
3502 for (j = 0; j < t->num; j++) {
3503 printf("%d:%s\n", t->nodes[j].pnn,
3504 ctdb_addr_to_str(&t->nodes[j].addr));
3511 talloc_free(tmp_ctx);
3516 display who is the lvs master
3518 static int control_lvsmaster(struct ctdb_context *ctdb, int argc, const char **argv)
3520 TALLOC_CTX *tmp_ctx = talloc_new(ctdb);
3521 struct ctdb_node_map_old *nodemap=NULL;
3524 ret = ctdb_ctrl_getnodemap(ctdb, TIMELIMIT(), options.pnn,
3527 DEBUG(DEBUG_ERR, ("Unable to get nodemap from node %u\n", options.pnn));
3528 talloc_free(tmp_ctx);
3532 for (i = 0; lvs_exclude_flags[i] != 0; i++) {
3533 struct ctdb_node_map_old *t =
3534 filter_nodemap_by_flags(ctdb, nodemap,
3535 lvs_exclude_flags[i]);
3542 struct ctdb_node_map_old *n;
3543 n = filter_nodemap_by_capabilities(ctdb,
3554 if (options.machinereadable) {
3555 printm("%d\n", n->nodes[0].pnn);
3557 printf("Node %d is LVS master\n", n->nodes[0].pnn);
3565 printf("There is no LVS master\n");
3568 talloc_free(tmp_ctx);
3573 disable monitoring on a node
3575 static int control_disable_monmode(struct ctdb_context *ctdb, int argc, const char **argv)
3580 ret = ctdb_ctrl_disable_monmode(ctdb, TIMELIMIT(), options.pnn);
3582 DEBUG(DEBUG_ERR, ("Unable to disable monmode on node %u\n", options.pnn));
3585 printf("Monitoring mode:%s\n","DISABLED");
3591 enable monitoring on a node
3593 static int control_enable_monmode(struct ctdb_context *ctdb, int argc, const char **argv)
3598 ret = ctdb_ctrl_enable_monmode(ctdb, TIMELIMIT(), options.pnn);
3600 DEBUG(DEBUG_ERR, ("Unable to enable monmode on node %u\n", options.pnn));
3603 printf("Monitoring mode:%s\n","ACTIVE");
3609 display remote list of keys/data for a db
3611 static int control_catdb(struct ctdb_context *ctdb, int argc, const char **argv)
3613 const char *db_name;
3614 struct ctdb_db_context *ctdb_db;
3616 struct ctdb_dump_db_context c;
3623 if (!db_exists(ctdb, argv[0], NULL, &db_name, &flags)) {
3627 ctdb_db = ctdb_attach(ctdb, TIMELIMIT(), db_name, flags & CTDB_DB_FLAGS_PERSISTENT, 0);
3628 if (ctdb_db == NULL) {
3629 DEBUG(DEBUG_ERR,("Unable to attach to database '%s'\n", db_name));
3633 if (options.printlmaster) {
3634 ret = ctdb_ctrl_getvnnmap(ctdb, TIMELIMIT(), options.pnn,
3635 ctdb, &ctdb->vnn_map);
3637 DEBUG(DEBUG_ERR, ("Unable to get vnnmap from node %u\n",
3646 c.printemptyrecords = (bool)options.printemptyrecords;
3647 c.printdatasize = (bool)options.printdatasize;
3648 c.printlmaster = (bool)options.printlmaster;
3649 c.printhash = (bool)options.printhash;
3650 c.printrecordflags = (bool)options.printrecordflags;
3652 /* traverse and dump the cluster tdb */
3653 ret = ctdb_dump_db(ctdb_db, &c);
3655 DEBUG(DEBUG_ERR, ("Unable to dump database\n"));
3656 DEBUG(DEBUG_ERR, ("Maybe try 'ctdb getdbstatus %s'"
3657 " and 'ctdb getvar AllowUnhealthyDBRead'\n",
3661 talloc_free(ctdb_db);
3663 printf("Dumped %d records\n", ret);
3667 struct cattdb_data {
3668 struct ctdb_context *ctdb;
3672 static int cattdb_traverse(struct tdb_context *tdb, TDB_DATA key, TDB_DATA data, void *private_data)
3674 struct cattdb_data *d = private_data;
3675 struct ctdb_dump_db_context c;
3682 c.printemptyrecords = (bool)options.printemptyrecords;
3683 c.printdatasize = (bool)options.printdatasize;
3684 c.printlmaster = false;
3685 c.printhash = (bool)options.printhash;
3686 c.printrecordflags = true;
3688 return ctdb_dumpdb_record(key, data, &c);
3692 cat the local tdb database using same format as catdb
3694 static int control_cattdb(struct ctdb_context *ctdb, int argc, const char **argv)
3696 const char *db_name;
3697 struct ctdb_db_context *ctdb_db;
3698 struct cattdb_data d;
3705 if (!db_exists(ctdb, argv[0], NULL, &db_name, &flags)) {
3709 ctdb_db = ctdb_attach(ctdb, TIMELIMIT(), db_name, flags & CTDB_DB_FLAGS_PERSISTENT, 0);
3710 if (ctdb_db == NULL) {
3711 DEBUG(DEBUG_ERR,("Unable to attach to database '%s'\n", db_name));
3715 /* traverse the local tdb */
3718 if (tdb_traverse_read(ctdb_db->ltdb->tdb, cattdb_traverse, &d) == -1) {
3719 printf("Failed to cattdb data\n");
3722 talloc_free(ctdb_db);
3724 printf("Dumped %d records\n", d.count);
3729 display the content of a database key
3731 static int control_readkey(struct ctdb_context *ctdb, int argc, const char **argv)
3733 const char *db_name;
3734 struct ctdb_db_context *ctdb_db;
3735 struct ctdb_record_handle *h;
3736 TALLOC_CTX *tmp_ctx = talloc_new(ctdb);
3744 if (!db_exists(ctdb, argv[0], NULL, &db_name, &flags)) {
3748 ctdb_db = ctdb_attach(ctdb, TIMELIMIT(), db_name, flags & CTDB_DB_FLAGS_PERSISTENT, 0);
3749 if (ctdb_db == NULL) {
3750 DEBUG(DEBUG_ERR,("Unable to attach to database '%s'\n", db_name));
3754 key.dptr = discard_const(argv[1]);
3755 key.dsize = strlen((char *)key.dptr);
3757 h = ctdb_fetch_lock(ctdb_db, tmp_ctx, key, &data);
3759 printf("Failed to fetch record '%s' on node %d\n",
3760 (const char *)key.dptr, ctdb_get_pnn(ctdb));
3761 talloc_free(tmp_ctx);
3765 printf("Data: size:%d ptr:[%.*s]\n", (int)data.dsize, (int)data.dsize, data.dptr);
3767 talloc_free(tmp_ctx);
3768 talloc_free(ctdb_db);
3773 display the content of a database key
3775 static int control_writekey(struct ctdb_context *ctdb, int argc, const char **argv)
3777 const char *db_name;
3778 struct ctdb_db_context *ctdb_db;
3779 struct ctdb_record_handle *h;
3780 TALLOC_CTX *tmp_ctx = talloc_new(ctdb);
3788 if (!db_exists(ctdb, argv[0], NULL, &db_name, &flags)) {
3792 ctdb_db = ctdb_attach(ctdb, TIMELIMIT(), db_name, flags & CTDB_DB_FLAGS_PERSISTENT, 0);
3793 if (ctdb_db == NULL) {
3794 DEBUG(DEBUG_ERR,("Unable to attach to database '%s'\n", db_name));
3798 key.dptr = discard_const(argv[1]);
3799 key.dsize = strlen((char *)key.dptr);
3801 h = ctdb_fetch_lock(ctdb_db, tmp_ctx, key, &data);
3803 printf("Failed to fetch record '%s' on node %d\n",
3804 (const char *)key.dptr, ctdb_get_pnn(ctdb));
3805 talloc_free(tmp_ctx);
3809 data.dptr = discard_const(argv[2]);
3810 data.dsize = strlen((char *)data.dptr);
3812 if (ctdb_record_store(h, data) != 0) {
3813 printf("Failed to store record\n");
3817 talloc_free(tmp_ctx);
3818 talloc_free(ctdb_db);
3823 fetch a record from a persistent database
3825 static int control_pfetch(struct ctdb_context *ctdb, int argc, const char **argv)
3827 const char *db_name;
3828 struct ctdb_db_context *ctdb_db;
3829 TALLOC_CTX *tmp_ctx = talloc_new(ctdb);
3830 struct ctdb_transaction_handle *h;
3837 talloc_free(tmp_ctx);
3841 if (!db_exists(ctdb, argv[0], NULL, &db_name, &flags)) {
3842 talloc_free(tmp_ctx);
3846 persistent = flags & CTDB_DB_FLAGS_PERSISTENT;
3848 DEBUG(DEBUG_ERR,("Database '%s' is not persistent\n", db_name));
3849 talloc_free(tmp_ctx);
3853 ctdb_db = ctdb_attach(ctdb, TIMELIMIT(), db_name, persistent, 0);
3854 if (ctdb_db == NULL) {
3855 DEBUG(DEBUG_ERR,("Unable to attach to database '%s'\n", db_name));
3856 talloc_free(tmp_ctx);
3860 h = ctdb_transaction_start(ctdb_db, tmp_ctx);
3862 DEBUG(DEBUG_ERR,("Failed to start transaction on database %s\n", db_name));
3863 talloc_free(tmp_ctx);
3867 key.dptr = discard_const(argv[1]);
3868 key.dsize = strlen(argv[1]);
3869 ret = ctdb_transaction_fetch(h, tmp_ctx, key, &data);
3871 DEBUG(DEBUG_ERR,("Failed to fetch record\n"));
3872 talloc_free(tmp_ctx);
3876 if (data.dsize == 0 || data.dptr == NULL) {
3877 DEBUG(DEBUG_ERR,("Record is empty\n"));
3878 talloc_free(tmp_ctx);
3883 fd = open(argv[2], O_WRONLY|O_CREAT|O_TRUNC, 0600);
3885 DEBUG(DEBUG_ERR,("Failed to open output file %s\n", argv[2]));
3886 talloc_free(tmp_ctx);
3889 sys_write(fd, data.dptr, data.dsize);
3892 sys_write(1, data.dptr, data.dsize);
3895 /* abort the transaction */
3899 talloc_free(tmp_ctx);
3904 fetch a record from a tdb-file
3906 static int control_tfetch(struct ctdb_context *ctdb, int argc, const char **argv)
3908 const char *tdb_file;
3911 TALLOC_CTX *tmp_ctx = talloc_new(NULL);
3920 tdb = tdb_open(tdb_file, 0, 0, O_RDONLY, 0);
3922 printf("Failed to open TDB file %s\n", tdb_file);
3926 key = strtodata(tmp_ctx, argv[1], strlen(argv[1]));
3927 if (key.dptr == NULL) {
3928 printf("Failed to convert \"%s\" into a TDB_DATA\n", argv[1]);
3932 data = tdb_fetch(tdb, key);
3933 if (data.dptr == NULL || data.dsize < sizeof(struct ctdb_ltdb_header)) {
3934 printf("Failed to read record %s from tdb %s\n", argv[1], tdb_file);
3942 fd = open(argv[2], O_WRONLY|O_CREAT|O_TRUNC, 0600);
3944 printf("Failed to open output file %s\n", argv[2]);
3947 if (options.verbose){
3948 sys_write(fd, data.dptr, data.dsize);
3950 sys_write(fd, data.dptr+sizeof(struct ctdb_ltdb_header), data.dsize-sizeof(struct ctdb_ltdb_header));
3954 if (options.verbose){
3955 sys_write(1, data.dptr, data.dsize);
3957 sys_write(1, data.dptr+sizeof(struct ctdb_ltdb_header), data.dsize-sizeof(struct ctdb_ltdb_header));
3961 talloc_free(tmp_ctx);
3966 store a record and header to a tdb-file
3968 static int control_tstore(struct ctdb_context *ctdb, int argc, const char **argv)
3970 const char *tdb_file;
3972 TDB_DATA key, value, data;
3973 TALLOC_CTX *tmp_ctx = talloc_new(NULL);
3974 struct ctdb_ltdb_header header;
3982 tdb = tdb_open(tdb_file, 0, 0, O_RDWR, 0);
3984 printf("Failed to open TDB file %s\n", tdb_file);
3988 key = strtodata(tmp_ctx, argv[1], strlen(argv[1]));
3989 if (key.dptr == NULL) {
3990 printf("Failed to convert \"%s\" into a TDB_DATA\n", argv[1]);
3994 value = strtodata(tmp_ctx, argv[2], strlen(argv[2]));
3995 if (value.dptr == NULL) {
3996 printf("Failed to convert \"%s\" into a TDB_DATA\n", argv[2]);
4000 ZERO_STRUCT(header);
4002 header.rsn = atoll(argv[3]);
4005 header.dmaster = atoi(argv[4]);
4008 header.flags = atoi(argv[5]);
4011 data.dsize = sizeof(struct ctdb_ltdb_header) + value.dsize;
4012 data.dptr = talloc_size(tmp_ctx, data.dsize);
4013 if (data.dptr == NULL) {
4014 printf("Failed to allocate header+value\n");
4018 *(struct ctdb_ltdb_header *)data.dptr = header;
4019 memcpy(data.dptr + sizeof(struct ctdb_ltdb_header), value.dptr, value.dsize);
4021 if (tdb_store(tdb, key, data, TDB_REPLACE) != 0) {
4022 printf("Failed to write record %s to tdb %s\n", argv[1], tdb_file);
4029 talloc_free(tmp_ctx);
4034 write a record to a persistent database
4036 static int control_pstore(struct ctdb_context *ctdb, int argc, const char **argv)
4038 const char *db_name;
4039 struct ctdb_db_context *ctdb_db;
4040 TALLOC_CTX *tmp_ctx = talloc_new(ctdb);
4041 struct ctdb_transaction_handle *h;
4047 talloc_free(tmp_ctx);
4051 fd = open(argv[2], O_RDONLY);
4053 DEBUG(DEBUG_ERR,("Failed to open file containing record data : %s %s\n", argv[2], strerror(errno)));
4054 talloc_free(tmp_ctx);
4058 ret = fstat(fd, &st);
4060 DEBUG(DEBUG_ERR,("fstat of file %s failed: %s\n", argv[2], strerror(errno)));
4062 talloc_free(tmp_ctx);
4066 if (!S_ISREG(st.st_mode)) {
4067 DEBUG(DEBUG_ERR,("Not a regular file %s\n", argv[2]));
4069 talloc_free(tmp_ctx);
4073 data.dsize = st.st_size;
4074 if (data.dsize == 0) {
4077 data.dptr = talloc_size(tmp_ctx, data.dsize);
4078 if (data.dptr == NULL) {
4079 DEBUG(DEBUG_ERR,("Failed to talloc %d of memory to store record data\n", (int)data.dsize));
4081 talloc_free(tmp_ctx);
4084 ret = sys_read(fd, data.dptr, data.dsize);
4085 if (ret != data.dsize) {
4086 DEBUG(DEBUG_ERR,("Failed to read %d bytes of record data\n", (int)data.dsize));
4088 talloc_free(tmp_ctx);
4097 ctdb_db = ctdb_attach(ctdb, TIMELIMIT(), db_name, true, 0);
4098 if (ctdb_db == NULL) {
4099 DEBUG(DEBUG_ERR,("Unable to attach to database '%s'\n", db_name));
4100 talloc_free(tmp_ctx);
4104 h = ctdb_transaction_start(ctdb_db, tmp_ctx);
4106 DEBUG(DEBUG_ERR,("Failed to start transaction on database %s\n", db_name));
4107 talloc_free(tmp_ctx);
4111 key = strtodata(tmp_ctx, argv[1], strlen(argv[1]));
4112 if (key.dptr == NULL) {
4113 printf("Failed to convert \"%s\" into a TDB_DATA\n", argv[1]);
4117 ret = ctdb_transaction_store(h, key, data);
4119 DEBUG(DEBUG_ERR,("Failed to store record\n"));
4120 talloc_free(tmp_ctx);
4124 ret = ctdb_transaction_commit(h);
4126 DEBUG(DEBUG_ERR,("Failed to commit transaction\n"));
4127 talloc_free(tmp_ctx);
4132 talloc_free(tmp_ctx);
4137 * delete a record from a persistent database
4139 static int control_pdelete(struct ctdb_context *ctdb, int argc, const char **argv)
4141 const char *db_name;
4142 struct ctdb_db_context *ctdb_db;
4143 TALLOC_CTX *tmp_ctx = talloc_new(ctdb);
4144 struct ctdb_transaction_handle *h;
4151 talloc_free(tmp_ctx);
4155 if (!db_exists(ctdb, argv[0], NULL, &db_name, &flags)) {
4156 talloc_free(tmp_ctx);
4160 persistent = flags & CTDB_DB_FLAGS_PERSISTENT;
4162 DEBUG(DEBUG_ERR, ("Database '%s' is not persistent\n", db_name));
4163 talloc_free(tmp_ctx);
4167 ctdb_db = ctdb_attach(ctdb, TIMELIMIT(), db_name, persistent, 0);
4168 if (ctdb_db == NULL) {
4169 DEBUG(DEBUG_ERR, ("Unable to attach to database '%s'\n", db_name));
4170 talloc_free(tmp_ctx);
4174 h = ctdb_transaction_start(ctdb_db, tmp_ctx);
4176 DEBUG(DEBUG_ERR, ("Failed to start transaction on database %s\n", db_name));
4177 talloc_free(tmp_ctx);
4181 key = strtodata(tmp_ctx, argv[1], strlen(argv[1]));
4182 if (key.dptr == NULL) {
4183 printf("Failed to convert \"%s\" into a TDB_DATA\n", argv[1]);
4187 ret = ctdb_transaction_store(h, key, tdb_null);
4189 DEBUG(DEBUG_ERR, ("Failed to delete record\n"));
4190 talloc_free(tmp_ctx);
4194 ret = ctdb_transaction_commit(h);
4196 DEBUG(DEBUG_ERR, ("Failed to commit transaction\n"));
4197 talloc_free(tmp_ctx);
4201 talloc_free(tmp_ctx);
4205 static const char *ptrans_parse_string(TALLOC_CTX *mem_ctx, const char *s,
4210 const char *ret; /* Next byte after successfully parsed value */
4212 /* Error, unless someone says otherwise */
4214 /* Indicates no value to parse */
4217 /* Skip whitespace */
4218 n = strspn(s, " \t");
4222 /* Quoted ASCII string - no wide characters! */
4224 n = strcspn(t, "\"");
4227 *data = strtodata(mem_ctx, t, n);
4228 CTDB_NOMEM_ABORT(data->dptr);
4232 DEBUG(DEBUG_WARNING,("Unmatched \" in input %s\n", s));
4235 DEBUG(DEBUG_WARNING,("Unsupported input format in %s\n", s));
4241 static bool ptrans_get_key_value(TALLOC_CTX *mem_ctx, FILE *file,
4242 TDB_DATA *key, TDB_DATA *value)
4244 char line [1024]; /* FIXME: make this more flexible? */
4248 ptr = fgets(line, sizeof(line), file);
4255 t = ptrans_parse_string(mem_ctx, line, key);
4256 if (t == NULL || key->dptr == NULL) {
4257 /* Line Ignored but not EOF */
4262 t = ptrans_parse_string(mem_ctx, t, value);
4264 /* Line Ignored but not EOF */
4265 talloc_free(key->dptr);
4274 * Update a persistent database as per file/stdin
4276 static int control_ptrans(struct ctdb_context *ctdb,
4277 int argc, const char **argv)
4279 const char *db_name;
4280 struct ctdb_db_context *ctdb_db;
4281 TALLOC_CTX *tmp_ctx = talloc_new(ctdb);
4282 struct ctdb_transaction_handle *h;
4283 TDB_DATA key, value;
4288 talloc_free(tmp_ctx);
4294 file = fopen(argv[1], "r");
4296 DEBUG(DEBUG_ERR,("Unable to open file for reading '%s'\n", argv[1]));
4297 talloc_free(tmp_ctx);
4304 ctdb_db = ctdb_attach(ctdb, TIMELIMIT(), db_name, true, 0);
4305 if (ctdb_db == NULL) {
4306 DEBUG(DEBUG_ERR,("Unable to attach to database '%s'\n", db_name));
4310 h = ctdb_transaction_start(ctdb_db, tmp_ctx);
4312 DEBUG(DEBUG_ERR,("Failed to start transaction on database %s\n", db_name));
4316 while (ptrans_get_key_value(tmp_ctx, file, &key, &value)) {
4317 if (key.dsize != 0) {
4318 ret = ctdb_transaction_store(h, key, value);
4319 /* Minimise memory use */
4320 talloc_free(key.dptr);
4321 if (value.dptr != NULL) {
4322 talloc_free(value.dptr);
4325 DEBUG(DEBUG_ERR,("Failed to store record\n"));
4326 ctdb_transaction_cancel(h);
4332 ret = ctdb_transaction_commit(h);
4334 DEBUG(DEBUG_ERR,("Failed to commit transaction\n"));
4338 if (file != stdin) {
4341 talloc_free(tmp_ctx);
4345 if (file != stdin) {
4349 talloc_free(tmp_ctx);
4354 check if a service is bound to a port or not
4356 static int control_chktcpport(struct ctdb_context *ctdb, int argc, const char **argv)
4361 struct sockaddr_in sin;
4364 printf("Use: ctdb chktcport <port>\n");
4368 port = atoi(argv[0]);
4370 s = socket(PF_INET, SOCK_STREAM, IPPROTO_TCP);
4372 printf("Failed to open local socket\n");
4376 v = fcntl(s, F_GETFL, 0);
4377 if (v == -1 || fcntl(s, F_SETFL, v | O_NONBLOCK) != 0) {
4378 printf("Unable to set socket non-blocking: %s\n", strerror(errno));
4381 bzero(&sin, sizeof(sin));
4382 sin.sin_family = PF_INET;
4383 sin.sin_port = htons(port);
4384 ret = bind(s, (struct sockaddr *)&sin, sizeof(sin));
4387 printf("Failed to bind to local socket: %d %s\n", errno, strerror(errno));
4395 /* Reload public IPs on a specified nodes */
4396 static int control_reloadips(struct ctdb_context *ctdb, int argc, const char **argv)
4398 TALLOC_CTX *tmp_ctx = talloc_new(ctdb);
4404 assert_single_node_only();
4410 /* Determine the nodes where IPs need to be reloaded */
4411 if (!parse_nodestring(ctdb, tmp_ctx, argc == 1 ? argv[0] : NULL,
4412 options.pnn, true, &nodes, &pnn_mode)) {
4418 /* Disable takeover runs on all connected nodes. A reply
4419 * indicating success is needed from each node so all nodes
4420 * will need to be active. This will retry until maxruntime
4421 * is exceeded, hence no error handling.
4423 * A check could be added to not allow reloading of IPs when
4424 * there are disconnected nodes. However, this should
4425 * probably be left up to the administrator.
4427 timeout = LONGTIMEOUT;
4428 srvid_broadcast(ctdb, CTDB_SRVID_DISABLE_TAKEOVER_RUNS, &timeout,
4429 "Disable takeover runs", true);
4431 /* Now tell all the desired nodes to reload their public IPs.
4432 * Keep trying this until it succeeds. This assumes all
4433 * failures are transient, which might not be true...
4435 if (ctdb_client_async_control(ctdb, CTDB_CONTROL_RELOAD_PUBLIC_IPS,
4436 nodes, 0, LONGTIMELIMIT(),
4438 NULL, NULL, NULL) != 0) {
4440 ("Unable to reload IPs on some nodes, try again.\n"));
4444 /* It isn't strictly necessary to wait until takeover runs are
4445 * re-enabled but doing so can't hurt.
4448 srvid_broadcast(ctdb, CTDB_SRVID_DISABLE_TAKEOVER_RUNS, &timeout,
4449 "Enable takeover runs", true);
4455 talloc_free(tmp_ctx);
4460 display a list of the databases on a remote ctdb
4462 static int control_getdbmap(struct ctdb_context *ctdb, int argc, const char **argv)
4465 struct ctdb_dbid_map_old *dbmap=NULL;
4467 ret = ctdb_ctrl_getdbmap(ctdb, TIMELIMIT(), options.pnn, ctdb, &dbmap);
4469 DEBUG(DEBUG_ERR, ("Unable to get dbids from node %u\n", options.pnn));
4473 if(options.machinereadable){
4474 printm(":ID:Name:Path:Persistent:Sticky:Unhealthy:ReadOnly:\n");
4475 for(i=0;i<dbmap->num;i++){
4476 const char *path = NULL;
4477 const char *name = NULL;
4478 const char *health = NULL;
4483 ctdb_ctrl_getdbpath(ctdb, TIMELIMIT(), options.pnn,
4484 dbmap->dbs[i].db_id, ctdb, &path);
4485 ctdb_ctrl_getdbname(ctdb, TIMELIMIT(), options.pnn,
4486 dbmap->dbs[i].db_id, ctdb, &name);
4487 ctdb_ctrl_getdbhealth(ctdb, TIMELIMIT(), options.pnn,
4488 dbmap->dbs[i].db_id, ctdb, &health);
4489 persistent = dbmap->dbs[i].flags & CTDB_DB_FLAGS_PERSISTENT;
4490 readonly = dbmap->dbs[i].flags & CTDB_DB_FLAGS_READONLY;
4491 sticky = dbmap->dbs[i].flags & CTDB_DB_FLAGS_STICKY;
4492 printm(":0x%08X:%s:%s:%d:%d:%d:%d:\n",
4493 dbmap->dbs[i].db_id, name, path,
4494 !!(persistent), !!(sticky),
4495 !!(health), !!(readonly));
4500 printf("Number of databases:%d\n", dbmap->num);
4501 for(i=0;i<dbmap->num;i++){
4502 const char *path = NULL;
4503 const char *name = NULL;
4504 const char *health = NULL;
4509 ctdb_ctrl_getdbpath(ctdb, TIMELIMIT(), options.pnn, dbmap->dbs[i].db_id, ctdb, &path);
4510 ctdb_ctrl_getdbname(ctdb, TIMELIMIT(), options.pnn, dbmap->dbs[i].db_id, ctdb, &name);
4511 ctdb_ctrl_getdbhealth(ctdb, TIMELIMIT(), options.pnn, dbmap->dbs[i].db_id, ctdb, &health);
4512 persistent = dbmap->dbs[i].flags & CTDB_DB_FLAGS_PERSISTENT;
4513 readonly = dbmap->dbs[i].flags & CTDB_DB_FLAGS_READONLY;
4514 sticky = dbmap->dbs[i].flags & CTDB_DB_FLAGS_STICKY;
4515 printf("dbid:0x%08x name:%s path:%s%s%s%s%s\n",
4516 dbmap->dbs[i].db_id, name, path,
4517 persistent?" PERSISTENT":"",
4518 sticky?" STICKY":"",
4519 readonly?" READONLY":"",
4520 health?" UNHEALTHY":"");
4527 display the status of a database on a remote ctdb
4529 static int control_getdbstatus(struct ctdb_context *ctdb, int argc, const char **argv)
4531 const char *db_name;
4534 const char *path = NULL;
4535 const char *health = NULL;
4541 if (!db_exists(ctdb, argv[0], &db_id, &db_name, &flags)) {
4545 ctdb_ctrl_getdbpath(ctdb, TIMELIMIT(), options.pnn, db_id, ctdb, &path);
4546 ctdb_ctrl_getdbhealth(ctdb, TIMELIMIT(), options.pnn, db_id, ctdb, &health);
4547 printf("dbid: 0x%08x\nname: %s\npath: %s\nPERSISTENT: %s\nSTICKY: %s\nREADONLY: %s\nHEALTH: %s\n",
4548 db_id, db_name, path,
4549 (flags & CTDB_DB_FLAGS_PERSISTENT ? "yes" : "no"),
4550 (flags & CTDB_DB_FLAGS_STICKY ? "yes" : "no"),
4551 (flags & CTDB_DB_FLAGS_READONLY ? "yes" : "no"),
4552 (health ? health : "OK"));
4558 check if the local node is recmaster or not
4559 it will return 1 if this node is the recmaster and 0 if it is not
4560 or if the local ctdb daemon could not be contacted
4562 static int control_isnotrecmaster(struct ctdb_context *ctdb, int argc, const char **argv)
4564 uint32_t mypnn, recmaster;
4567 assert_single_node_only();
4569 mypnn = getpnn(ctdb);
4571 ret = ctdb_ctrl_getrecmaster(ctdb, ctdb, TIMELIMIT(), options.pnn, &recmaster);
4573 printf("Failed to get the recmaster\n");
4577 if (recmaster != mypnn) {
4578 printf("this node is not the recmaster\n");
4582 printf("this node is the recmaster\n");
4589 static int control_ping(struct ctdb_context *ctdb, int argc, const char **argv)
4592 struct timeval tv = timeval_current();
4593 ret = ctdb_ctrl_ping(ctdb, options.pnn);
4595 printf("Unable to get ping response from node %u\n", options.pnn);
4598 printf("response from %u time=%.6f sec (%d clients)\n",
4599 options.pnn, timeval_elapsed(&tv), ret);
4606 get a node's runstate
4608 static int control_runstate(struct ctdb_context *ctdb, int argc, const char **argv)
4611 enum ctdb_runstate runstate;
4613 ret = ctdb_ctrl_get_runstate(ctdb, TIMELIMIT(), options.pnn, &runstate);
4615 printf("Unable to get runstate response from node %u\n",
4620 enum ctdb_runstate t;
4622 for (i=0; i<argc; i++) {
4624 t = runstate_from_string(argv[i]);
4625 if (t == CTDB_RUNSTATE_UNKNOWN) {
4626 printf("Invalid run state (%s)\n", argv[i]);
4630 if (t == runstate) {
4637 printf("CTDB not in required run state (got %s)\n",
4638 runstate_to_string((enum ctdb_runstate)runstate));
4643 printf("%s\n", runstate_to_string(runstate));
4651 static int control_getvar(struct ctdb_context *ctdb, int argc, const char **argv)
4662 ret = ctdb_ctrl_get_tunable(ctdb, TIMELIMIT(), options.pnn, name, &value);
4664 DEBUG(DEBUG_ERR, ("Unable to get tunable variable '%s'\n", name));
4668 printf("%-23s = %u\n", name, value);
4675 static int control_setvar(struct ctdb_context *ctdb, int argc, const char **argv)
4686 value = strtoul(argv[1], NULL, 0);
4688 ret = ctdb_ctrl_set_tunable(ctdb, TIMELIMIT(), options.pnn, name, value);
4690 DEBUG(DEBUG_ERR, ("Unable to set tunable variable '%s'\n", name));
4694 DEBUG(DEBUG_WARNING,
4695 ("Setting obsolete tunable variable '%s'\n",
4704 static int control_listvars(struct ctdb_context *ctdb, int argc, const char **argv)
4710 ret = ctdb_ctrl_list_tunables(ctdb, TIMELIMIT(), options.pnn, ctdb, &list, &count);
4712 DEBUG(DEBUG_ERR, ("Unable to list tunable variables\n"));
4716 for (i=0;i<count;i++) {
4717 control_getvar(ctdb, 1, &list[i]);
4726 display debug level on a node
4728 static int control_getdebug(struct ctdb_context *ctdb, int argc, const char **argv)
4733 ret = ctdb_ctrl_get_debuglevel(ctdb, options.pnn, &level);
4735 DEBUG(DEBUG_ERR, ("Unable to get debuglevel response from node %u\n", options.pnn));
4738 enum debug_level log_level = debug_level_from_int(level);
4739 const char *desc = debug_level_to_string(log_level);
4741 /* This should never happen */
4744 if (options.machinereadable){
4745 printm(":Name:Level:\n");
4746 printm(":%s:%d:\n", desc, level);
4748 printf("Node %u is at debug level %s (%d)\n",
4749 options.pnn, desc, level);
4756 display reclock file of a node
4758 static int control_getreclock(struct ctdb_context *ctdb, int argc, const char **argv)
4761 const char *reclock;
4763 ret = ctdb_ctrl_getreclock(ctdb, TIMELIMIT(), options.pnn, ctdb, &reclock);
4765 DEBUG(DEBUG_ERR, ("Unable to get reclock file from node %u\n", options.pnn));
4768 if (options.machinereadable){
4769 if (reclock != NULL) {
4770 printm("%s", reclock);
4773 if (reclock == NULL) {
4774 printf("No reclock file used.\n");
4776 printf("Reclock file:%s\n", reclock);
4784 set the reclock file of a node
4786 static int control_setreclock(struct ctdb_context *ctdb, int argc, const char **argv)
4789 const char *reclock = NULL;
4793 } else if (argc == 1) {
4799 ret = ctdb_ctrl_setreclock(ctdb, TIMELIMIT(), options.pnn, reclock);
4801 DEBUG(DEBUG_ERR, ("Unable to get reclock file from node %u\n", options.pnn));
4808 set the natgw state on/off
4810 static int control_setnatgwstate(struct ctdb_context *ctdb, int argc, const char **argv)
4813 uint32_t natgwstate = 0;
4819 if (!strcmp(argv[0], "on")) {
4821 } else if (!strcmp(argv[0], "off")) {
4827 ret = ctdb_ctrl_setnatgwstate(ctdb, TIMELIMIT(), options.pnn, natgwstate);
4829 DEBUG(DEBUG_ERR, ("Unable to set the natgw state for node %u\n", options.pnn));
4837 set the lmaster role on/off
4839 static int control_setlmasterrole(struct ctdb_context *ctdb, int argc, const char **argv)
4842 uint32_t lmasterrole = 0;
4848 if (!strcmp(argv[0], "on")) {
4850 } else if (!strcmp(argv[0], "off")) {
4856 ret = ctdb_ctrl_setlmasterrole(ctdb, TIMELIMIT(), options.pnn, lmasterrole);
4858 DEBUG(DEBUG_ERR, ("Unable to set the lmaster role for node %u\n", options.pnn));
4866 set the recmaster role on/off
4868 static int control_setrecmasterrole(struct ctdb_context *ctdb, int argc, const char **argv)
4871 uint32_t recmasterrole = 0;
4877 if (!strcmp(argv[0], "on")) {
4879 } else if (!strcmp(argv[0], "off")) {
4885 ret = ctdb_ctrl_setrecmasterrole(ctdb, TIMELIMIT(), options.pnn, recmasterrole);
4887 DEBUG(DEBUG_ERR, ("Unable to set the recmaster role for node %u\n", options.pnn));
4895 set debug level on a node or all nodes
4897 static int control_setdebug(struct ctdb_context *ctdb, int argc, const char **argv)
4901 enum debug_level log_level;
4904 printf("You must specify the debug level. Valid levels are:\n");
4905 printf("\tERROR | WARNING | NOTICE | INFO | DEBUG\n");
4909 if (!debug_level_parse(argv[0], &log_level)) {
4910 printf("Invalid debug level, must be one of\n");
4911 printf("\tERROR | WARNING | NOTICE | INFO | DEBUG\n");
4915 level = debug_level_to_int(log_level);
4917 ret = ctdb_ctrl_set_debuglevel(ctdb, options.pnn, level);
4919 DEBUG(DEBUG_ERR, ("Unable to set debug level on node %u\n", options.pnn));
4928 static int control_thaw(struct ctdb_context *ctdb, int argc, const char **argv)
4934 priority = strtol(argv[0], NULL, 0);
4938 DEBUG(DEBUG_ERR,("Thaw by priority %u\n", priority));
4940 ret = ctdb_ctrl_thaw_priority(ctdb, TIMELIMIT(), options.pnn, priority);
4942 DEBUG(DEBUG_ERR, ("Unable to thaw node %u\n", options.pnn));
4949 attach to a database
4951 static int control_attach(struct ctdb_context *ctdb, int argc, const char **argv)
4953 const char *db_name;
4954 struct ctdb_db_context *ctdb_db;
4955 bool persistent = false;
4965 if (strcmp(argv[1], "persistent") != 0) {
4971 ctdb_db = ctdb_attach(ctdb, TIMELIMIT(), db_name, persistent, 0);
4972 if (ctdb_db == NULL) {
4973 DEBUG(DEBUG_ERR,("Unable to attach to database '%s'\n", db_name));
4981 * detach from a database
4983 static int control_detach(struct ctdb_context *ctdb, int argc,
4988 int ret, i, status = 0;
4989 struct ctdb_node_map_old *nodemap = NULL;
4990 TALLOC_CTX *tmp_ctx = talloc_new(ctdb);
4997 assert_single_node_only();
4999 ret = ctdb_ctrl_getrecmode(ctdb, tmp_ctx, TIMELIMIT(), options.pnn,
5002 DEBUG(DEBUG_ERR, ("Database cannot be detached "
5003 "when recovery is active\n"));
5004 talloc_free(tmp_ctx);
5008 ret = ctdb_ctrl_getnodemap(ctdb, TIMELIMIT(), options.pnn, tmp_ctx,
5011 DEBUG(DEBUG_ERR, ("Unable to get nodemap from node %u\n",
5013 talloc_free(tmp_ctx);
5017 for (i=0; i<nodemap->num; i++) {
5020 if (nodemap->nodes[i].flags & NODE_FLAGS_DISCONNECTED) {
5024 if (nodemap->nodes[i].flags & NODE_FLAGS_DELETED) {
5028 if (nodemap->nodes[i].flags & NODE_FLAGS_INACTIVE) {
5029 DEBUG(DEBUG_ERR, ("Database cannot be detached on "
5030 "inactive (stopped or banned) node "
5031 "%u\n", nodemap->nodes[i].pnn));
5032 talloc_free(tmp_ctx);
5036 ret = ctdb_ctrl_get_tunable(ctdb, TIMELIMIT(),
5037 nodemap->nodes[i].pnn,
5038 "AllowClientDBAttach",
5041 DEBUG(DEBUG_ERR, ("Unable to get tunable "
5042 "AllowClientDBAttach from node %u\n",
5043 nodemap->nodes[i].pnn));
5044 talloc_free(tmp_ctx);
5049 DEBUG(DEBUG_ERR, ("Database access is still active on "
5050 "node %u. Set AllowClientDBAttach=0 "
5052 nodemap->nodes[i].pnn));
5053 talloc_free(tmp_ctx);
5058 talloc_free(tmp_ctx);
5060 for (i=0; i<argc; i++) {
5061 if (!db_exists(ctdb, argv[i], &db_id, NULL, &flags)) {
5065 if (flags & CTDB_DB_FLAGS_PERSISTENT) {
5066 DEBUG(DEBUG_ERR, ("Persistent database '%s' "
5067 "cannot be detached\n", argv[i]));
5072 ret = ctdb_detach(ctdb, db_id);
5074 DEBUG(DEBUG_ERR, ("Database '%s' detach failed\n",
5086 static int control_setdbprio(struct ctdb_context *ctdb, int argc, const char **argv)
5088 struct ctdb_db_priority db_prio;
5095 db_prio.db_id = strtoul(argv[0], NULL, 0);
5096 db_prio.priority = strtoul(argv[1], NULL, 0);
5098 ret = ctdb_ctrl_set_db_priority(ctdb, TIMELIMIT(), options.pnn, &db_prio);
5100 DEBUG(DEBUG_ERR,("Unable to set db prio\n"));
5110 static int control_getdbprio(struct ctdb_context *ctdb, int argc, const char **argv)
5112 uint32_t db_id, priority;
5119 if (!db_exists(ctdb, argv[0], &db_id, NULL, NULL)) {
5123 ret = ctdb_ctrl_get_db_priority(ctdb, TIMELIMIT(), options.pnn, db_id, &priority);
5125 DEBUG(DEBUG_ERR,("Unable to get db prio\n"));
5129 DEBUG(DEBUG_ERR,("Priority:%u\n", priority));
5135 set the sticky records capability for a database
5137 static int control_setdbsticky(struct ctdb_context *ctdb, int argc, const char **argv)
5139 TALLOC_CTX *tmp_ctx = talloc_new(ctdb);
5147 if (!db_exists(ctdb, argv[0], &db_id, NULL, NULL)) {
5151 ret = ctdb_ctrl_set_db_sticky(ctdb, options.pnn, db_id);
5153 DEBUG(DEBUG_ERR,("Unable to set db to support sticky records\n"));
5154 talloc_free(tmp_ctx);
5158 talloc_free(tmp_ctx);
5163 set the readonly capability for a database
5165 static int control_setdbreadonly(struct ctdb_context *ctdb, int argc, const char **argv)
5167 TALLOC_CTX *tmp_ctx = talloc_new(ctdb);
5175 if (!db_exists(ctdb, argv[0], &db_id, NULL, NULL)) {
5179 ret = ctdb_ctrl_set_db_readonly(ctdb, options.pnn, db_id);
5181 DEBUG(DEBUG_ERR,("Unable to set db to support readonly\n"));
5182 talloc_free(tmp_ctx);
5186 talloc_free(tmp_ctx);
5193 static int control_getdbseqnum(struct ctdb_context *ctdb, int argc, const char **argv)
5203 if (!db_exists(ctdb, argv[0], &db_id, NULL, NULL)) {
5207 ret = ctdb_ctrl_getdbseqnum(ctdb, TIMELIMIT(), options.pnn, db_id, &seqnum);
5209 DEBUG(DEBUG_ERR, ("Unable to get seqnum from node."));
5213 printf("Sequence number:%lld\n", (long long)seqnum);
5219 run an eventscript on a node
5221 static int control_eventscript(struct ctdb_context *ctdb, int argc, const char **argv)
5227 TALLOC_CTX *tmp_ctx = talloc_new(ctdb);
5230 DEBUG(DEBUG_ERR,("Invalid arguments\n"));
5234 data.dptr = (unsigned char *)discard_const(argv[0]);
5235 data.dsize = strlen((char *)data.dptr) + 1;
5237 DEBUG(DEBUG_ERR, ("Running eventscripts with arguments \"%s\" on node %u\n", data.dptr, options.pnn));
5239 ret = ctdb_control(ctdb, options.pnn, 0, CTDB_CONTROL_RUN_EVENTSCRIPTS,
5240 0, data, tmp_ctx, NULL, &res, NULL, &errmsg);
5241 if (ret != 0 || res != 0) {
5242 if (errmsg != NULL) {
5244 ("Failed to run eventscripts - %s\n", errmsg));
5246 DEBUG(DEBUG_ERR, ("Failed to run eventscripts\n"));
5248 talloc_free(tmp_ctx);
5251 talloc_free(tmp_ctx);
5255 #define DB_VERSION 1
5256 #define MAX_DB_NAME 64
5257 struct db_file_header {
5258 unsigned long version;
5260 unsigned long persistent;
5262 const char name[MAX_DB_NAME];
5265 struct backup_data {
5266 struct ctdb_marshall_buffer *records;
5269 bool traverse_error;
5272 static int backup_traverse(struct tdb_context *tdb, TDB_DATA key, TDB_DATA data, void *private)
5274 struct backup_data *bd = talloc_get_type(private, struct backup_data);
5275 struct ctdb_rec_data_old *rec;
5277 /* add the record */
5278 rec = ctdb_marshall_record(bd->records, 0, key, NULL, data);
5280 bd->traverse_error = true;
5281 DEBUG(DEBUG_ERR,("Failed to marshall record\n"));
5284 bd->records = talloc_realloc_size(NULL, bd->records, rec->length + bd->len);
5285 if (bd->records == NULL) {
5286 DEBUG(DEBUG_ERR,("Failed to expand marshalling buffer\n"));
5287 bd->traverse_error = true;
5290 bd->records->count++;
5291 memcpy(bd->len+(uint8_t *)bd->records, rec, rec->length);
5292 bd->len += rec->length;
5300 * backup a database to a file
5302 static int control_backupdb(struct ctdb_context *ctdb, int argc, const char **argv)
5304 const char *db_name;
5306 TALLOC_CTX *tmp_ctx = talloc_new(ctdb);
5307 struct db_file_header dbhdr;
5308 struct ctdb_db_context *ctdb_db;
5309 struct backup_data *bd;
5312 const char *reason = NULL;
5316 assert_single_node_only();
5319 DEBUG(DEBUG_ERR,("Invalid arguments\n"));
5323 if (!db_exists(ctdb, argv[0], &db_id, &db_name, &flags)) {
5327 ret = ctdb_ctrl_getdbhealth(ctdb, TIMELIMIT(), options.pnn,
5328 db_id, tmp_ctx, &reason);
5330 DEBUG(DEBUG_ERR,("Unable to get dbhealth for database '%s'\n",
5332 talloc_free(tmp_ctx);
5336 uint32_t allow_unhealthy = 0;
5338 ctdb_ctrl_get_tunable(ctdb, TIMELIMIT(), options.pnn,
5339 "AllowUnhealthyDBRead",
5342 if (allow_unhealthy != 1) {
5343 DEBUG(DEBUG_ERR,("database '%s' is unhealthy: %s\n",
5346 DEBUG(DEBUG_ERR,("disallow backup : tunable AllowUnhealthyDBRead = %u\n",
5348 talloc_free(tmp_ctx);
5352 DEBUG(DEBUG_WARNING,("WARNING database '%s' is unhealthy - see 'ctdb getdbstatus %s'\n",
5354 DEBUG(DEBUG_WARNING,("WARNING! allow backup of unhealthy database: "
5355 "tunnable AllowUnhealthyDBRead = %u\n",
5359 ctdb_db = ctdb_attach(ctdb, TIMELIMIT(), db_name, flags & CTDB_DB_FLAGS_PERSISTENT, 0);
5360 if (ctdb_db == NULL) {
5361 DEBUG(DEBUG_ERR,("Unable to attach to database '%s'\n", argv[0]));
5362 talloc_free(tmp_ctx);
5367 ret = tdb_transaction_start(ctdb_db->ltdb->tdb);
5369 DEBUG(DEBUG_ERR,("Failed to start transaction\n"));
5370 talloc_free(tmp_ctx);
5375 bd = talloc_zero(tmp_ctx, struct backup_data);
5377 DEBUG(DEBUG_ERR,("Failed to allocate backup_data\n"));
5378 talloc_free(tmp_ctx);
5382 bd->records = talloc_zero(bd, struct ctdb_marshall_buffer);
5383 if (bd->records == NULL) {
5384 DEBUG(DEBUG_ERR,("Failed to allocate ctdb_marshall_buffer\n"));
5385 talloc_free(tmp_ctx);
5389 bd->len = offsetof(struct ctdb_marshall_buffer, data);
5390 bd->records->db_id = ctdb_db->db_id;
5391 /* traverse the database collecting all records */
5392 if (tdb_traverse_read(ctdb_db->ltdb->tdb, backup_traverse, bd) == -1 ||
5393 bd->traverse_error) {
5394 DEBUG(DEBUG_ERR,("Traverse error\n"));
5395 talloc_free(tmp_ctx);
5399 tdb_transaction_cancel(ctdb_db->ltdb->tdb);
5402 fh = open(argv[1], O_RDWR|O_CREAT, 0600);
5404 DEBUG(DEBUG_ERR,("Failed to open file '%s'\n", argv[1]));
5405 talloc_free(tmp_ctx);
5410 dbhdr.version = DB_VERSION;
5411 dbhdr.timestamp = time(NULL);
5412 dbhdr.persistent = flags & CTDB_DB_FLAGS_PERSISTENT;
5413 dbhdr.size = bd->len;
5414 if (strlen(argv[0]) >= MAX_DB_NAME) {
5415 DEBUG(DEBUG_ERR,("Too long dbname\n"));
5418 strncpy(discard_const(dbhdr.name), argv[0], MAX_DB_NAME-1);
5419 ret = sys_write(fh, &dbhdr, sizeof(dbhdr));
5421 DEBUG(DEBUG_ERR,("write failed: %s\n", strerror(errno)));
5424 ret = sys_write(fh, bd->records, bd->len);
5426 DEBUG(DEBUG_ERR,("write failed: %s\n", strerror(errno)));
5435 DEBUG(DEBUG_ERR,("close failed: %s\n", strerror(errno)));
5439 DEBUG(DEBUG_ERR,("Database backed up to %s\n", argv[1]));
5441 talloc_free(tmp_ctx);
5446 * restore a database from a file
5448 static int control_restoredb(struct ctdb_context *ctdb, int argc, const char **argv)
5451 TALLOC_CTX *tmp_ctx = talloc_new(ctdb);
5454 struct db_file_header dbhdr;
5455 struct ctdb_db_context *ctdb_db;
5456 struct ctdb_node_map_old *nodemap=NULL;
5457 struct ctdb_vnn_map *vnnmap=NULL;
5459 struct ctdb_transdb w;
5461 uint32_t generation;
5466 assert_single_node_only();
5468 if (argc < 1 || argc > 2) {
5469 DEBUG(DEBUG_ERR,("Invalid arguments\n"));
5473 fh = open(argv[0], O_RDONLY);
5475 DEBUG(DEBUG_ERR,("Failed to open file '%s'\n", argv[0]));
5476 talloc_free(tmp_ctx);
5480 sys_read(fh, &dbhdr, sizeof(dbhdr));
5481 if (dbhdr.version != DB_VERSION) {
5482 DEBUG(DEBUG_ERR,("Invalid version of database dump. File is version %lu but expected version was %u\n", dbhdr.version, DB_VERSION));
5484 talloc_free(tmp_ctx);
5488 dbname = discard_const(dbhdr.name);
5490 dbname = discard_const(argv[1]);
5493 outdata.dsize = dbhdr.size;
5494 outdata.dptr = talloc_size(tmp_ctx, outdata.dsize);
5495 if (outdata.dptr == NULL) {
5496 DEBUG(DEBUG_ERR,("Failed to allocate data of size '%lu'\n", dbhdr.size));
5498 talloc_free(tmp_ctx);
5501 sys_read(fh, outdata.dptr, outdata.dsize);
5504 tm = localtime(&dbhdr.timestamp);
5505 strftime(tbuf,sizeof(tbuf)-1,"%Y/%m/%d %H:%M:%S", tm);
5506 printf("Restoring database '%s' from backup @ %s\n",
5510 ctdb_db = ctdb_attach(ctdb, TIMELIMIT(), dbname, dbhdr.persistent, 0);
5511 if (ctdb_db == NULL) {
5512 DEBUG(DEBUG_ERR,("Unable to attach to database '%s'\n", dbname));
5513 talloc_free(tmp_ctx);
5517 ret = ctdb_ctrl_getnodemap(ctdb, TIMELIMIT(), options.pnn, ctdb, &nodemap);
5519 DEBUG(DEBUG_ERR, ("Unable to get nodemap from node %u\n", options.pnn));
5520 talloc_free(tmp_ctx);
5525 ret = ctdb_ctrl_getvnnmap(ctdb, TIMELIMIT(), options.pnn, tmp_ctx, &vnnmap);
5527 DEBUG(DEBUG_ERR, ("Unable to get vnnmap from node %u\n", options.pnn));
5528 talloc_free(tmp_ctx);
5532 /* freeze all nodes */
5533 nodes = list_of_active_nodes(ctdb, nodemap, tmp_ctx, true);
5534 for (i=1; i<=NUM_DB_PRIORITIES; i++) {
5535 if (ctdb_client_async_control(ctdb, CTDB_CONTROL_FREEZE,
5541 DEBUG(DEBUG_ERR, ("Unable to freeze nodes.\n"));
5542 ctdb_ctrl_setrecmode(ctdb, TIMELIMIT(), options.pnn, CTDB_RECOVERY_ACTIVE);
5543 talloc_free(tmp_ctx);
5548 generation = vnnmap->generation;
5549 data.dptr = (void *)&generation;
5550 data.dsize = sizeof(generation);
5552 /* start a cluster wide transaction */
5553 nodes = list_of_active_nodes(ctdb, nodemap, tmp_ctx, true);
5554 if (ctdb_client_async_control(ctdb, CTDB_CONTROL_TRANSACTION_START,
5556 TIMELIMIT(), false, data,
5559 DEBUG(DEBUG_ERR, ("Unable to start cluster wide transactions.\n"));
5564 w.db_id = ctdb_db->db_id;
5567 data.dptr = (void *)&w;
5568 data.dsize = sizeof(w);
5570 /* wipe all the remote databases. */
5571 nodes = list_of_active_nodes(ctdb, nodemap, tmp_ctx, true);
5572 if (ctdb_client_async_control(ctdb, CTDB_CONTROL_WIPE_DATABASE,
5574 TIMELIMIT(), false, data,
5577 DEBUG(DEBUG_ERR, ("Unable to wipe database.\n"));
5578 ctdb_ctrl_setrecmode(ctdb, TIMELIMIT(), options.pnn, CTDB_RECOVERY_ACTIVE);
5579 talloc_free(tmp_ctx);
5583 /* push the database */
5584 nodes = list_of_active_nodes(ctdb, nodemap, tmp_ctx, true);
5585 if (ctdb_client_async_control(ctdb, CTDB_CONTROL_PUSH_DB,
5587 TIMELIMIT(), false, outdata,
5590 DEBUG(DEBUG_ERR, ("Failed to push database.\n"));
5591 ctdb_ctrl_setrecmode(ctdb, TIMELIMIT(), options.pnn, CTDB_RECOVERY_ACTIVE);
5592 talloc_free(tmp_ctx);
5596 data.dptr = (void *)&ctdb_db->db_id;
5597 data.dsize = sizeof(ctdb_db->db_id);
5599 /* mark the database as healthy */
5600 nodes = list_of_active_nodes(ctdb, nodemap, tmp_ctx, true);
5601 if (ctdb_client_async_control(ctdb, CTDB_CONTROL_DB_SET_HEALTHY,
5603 TIMELIMIT(), false, data,
5606 DEBUG(DEBUG_ERR, ("Failed to mark database as healthy.\n"));
5607 ctdb_ctrl_setrecmode(ctdb, TIMELIMIT(), options.pnn, CTDB_RECOVERY_ACTIVE);
5608 talloc_free(tmp_ctx);
5612 data.dptr = (void *)&generation;
5613 data.dsize = sizeof(generation);
5615 /* commit all the changes */
5616 if (ctdb_client_async_control(ctdb, CTDB_CONTROL_TRANSACTION_COMMIT,
5618 TIMELIMIT(), false, data,
5621 DEBUG(DEBUG_ERR, ("Unable to commit databases.\n"));
5622 ctdb_ctrl_setrecmode(ctdb, TIMELIMIT(), options.pnn, CTDB_RECOVERY_ACTIVE);
5623 talloc_free(tmp_ctx);
5628 /* thaw all nodes */
5629 nodes = list_of_active_nodes(ctdb, nodemap, tmp_ctx, true);
5630 if (ctdb_client_async_control(ctdb, CTDB_CONTROL_THAW,
5636 DEBUG(DEBUG_ERR, ("Unable to thaw nodes.\n"));
5637 ctdb_ctrl_setrecmode(ctdb, TIMELIMIT(), options.pnn, CTDB_RECOVERY_ACTIVE);
5638 talloc_free(tmp_ctx);
5643 talloc_free(tmp_ctx);
5648 * dump a database backup from a file
5650 static int control_dumpdbbackup(struct ctdb_context *ctdb, int argc, const char **argv)
5652 TALLOC_CTX *tmp_ctx = talloc_new(ctdb);
5654 struct db_file_header dbhdr;
5658 struct ctdb_rec_data_old *rec = NULL;
5659 struct ctdb_marshall_buffer *m;
5660 struct ctdb_dump_db_context c;
5662 assert_single_node_only();
5665 DEBUG(DEBUG_ERR,("Invalid arguments\n"));
5669 fh = open(argv[0], O_RDONLY);
5671 DEBUG(DEBUG_ERR,("Failed to open file '%s'\n", argv[0]));
5672 talloc_free(tmp_ctx);
5676 sys_read(fh, &dbhdr, sizeof(dbhdr));
5677 if (dbhdr.version != DB_VERSION) {
5678 DEBUG(DEBUG_ERR,("Invalid version of database dump. File is version %lu but expected version was %u\n", dbhdr.version, DB_VERSION));
5680 talloc_free(tmp_ctx);
5684 outdata.dsize = dbhdr.size;
5685 outdata.dptr = talloc_size(tmp_ctx, outdata.dsize);
5686 if (outdata.dptr == NULL) {
5687 DEBUG(DEBUG_ERR,("Failed to allocate data of size '%lu'\n", dbhdr.size));
5689 talloc_free(tmp_ctx);
5692 sys_read(fh, outdata.dptr, outdata.dsize);
5694 m = (struct ctdb_marshall_buffer *)outdata.dptr;
5696 tm = localtime(&dbhdr.timestamp);
5697 strftime(tbuf,sizeof(tbuf)-1,"%Y/%m/%d %H:%M:%S", tm);
5698 printf("Backup of database name:'%s' dbid:0x%x08x from @ %s\n",
5699 dbhdr.name, m->db_id, tbuf);
5704 c.printemptyrecords = (bool)options.printemptyrecords;
5705 c.printdatasize = (bool)options.printdatasize;
5706 c.printlmaster = false;
5707 c.printhash = (bool)options.printhash;
5708 c.printrecordflags = (bool)options.printrecordflags;
5710 for (i=0; i < m->count; i++) {
5714 /* we do not want the header splitted, so we pass NULL*/
5715 rec = ctdb_marshall_loop_next(m, rec, &reqid,
5718 ctdb_dumpdb_record(key, data, &c);
5721 printf("Dumped %d records\n", i);
5722 talloc_free(tmp_ctx);
5727 * wipe a database from a file
5729 static int control_wipedb(struct ctdb_context *ctdb, int argc,
5732 const char *db_name;
5734 TALLOC_CTX *tmp_ctx = talloc_new(ctdb);
5736 struct ctdb_db_context *ctdb_db;
5737 struct ctdb_node_map_old *nodemap = NULL;
5738 struct ctdb_vnn_map *vnnmap = NULL;
5740 struct ctdb_transdb w;
5742 uint32_t generation;
5745 assert_single_node_only();
5748 DEBUG(DEBUG_ERR,("Invalid arguments\n"));
5752 if (!db_exists(ctdb, argv[0], NULL, &db_name, &flags)) {
5756 ctdb_db = ctdb_attach(ctdb, TIMELIMIT(), db_name, flags & CTDB_DB_FLAGS_PERSISTENT, 0);
5757 if (ctdb_db == NULL) {
5758 DEBUG(DEBUG_ERR, ("Unable to attach to database '%s'\n",
5760 talloc_free(tmp_ctx);
5764 ret = ctdb_ctrl_getnodemap(ctdb, TIMELIMIT(), options.pnn, ctdb,
5767 DEBUG(DEBUG_ERR, ("Unable to get nodemap from node %u\n",
5769 talloc_free(tmp_ctx);
5773 ret = ctdb_ctrl_getvnnmap(ctdb, TIMELIMIT(), options.pnn, tmp_ctx,
5776 DEBUG(DEBUG_ERR, ("Unable to get vnnmap from node %u\n",
5778 talloc_free(tmp_ctx);
5782 /* freeze all nodes */
5783 nodes = list_of_active_nodes(ctdb, nodemap, tmp_ctx, true);
5784 for (i=1; i<=NUM_DB_PRIORITIES; i++) {
5785 ret = ctdb_client_async_control(ctdb, CTDB_CONTROL_FREEZE,
5792 DEBUG(DEBUG_ERR, ("Unable to freeze nodes.\n"));
5793 ctdb_ctrl_setrecmode(ctdb, TIMELIMIT(), options.pnn,
5794 CTDB_RECOVERY_ACTIVE);
5795 talloc_free(tmp_ctx);
5800 generation = vnnmap->generation;
5801 data.dptr = (void *)&generation;
5802 data.dsize = sizeof(generation);
5804 /* start a cluster wide transaction */
5805 nodes = list_of_active_nodes(ctdb, nodemap, tmp_ctx, true);
5806 ret = ctdb_client_async_control(ctdb, CTDB_CONTROL_TRANSACTION_START,
5808 TIMELIMIT(), false, data,
5812 DEBUG(DEBUG_ERR, ("Unable to start cluster wide "
5813 "transactions.\n"));
5817 w.db_id = ctdb_db->db_id;
5820 data.dptr = (void *)&w;
5821 data.dsize = sizeof(w);
5823 /* wipe all the remote databases. */
5824 nodes = list_of_active_nodes(ctdb, nodemap, tmp_ctx, true);
5825 if (ctdb_client_async_control(ctdb, CTDB_CONTROL_WIPE_DATABASE,
5827 TIMELIMIT(), false, data,
5830 DEBUG(DEBUG_ERR, ("Unable to wipe database.\n"));
5831 ctdb_ctrl_setrecmode(ctdb, TIMELIMIT(), options.pnn, CTDB_RECOVERY_ACTIVE);
5832 talloc_free(tmp_ctx);
5836 data.dptr = (void *)&ctdb_db->db_id;
5837 data.dsize = sizeof(ctdb_db->db_id);
5839 /* mark the database as healthy */
5840 nodes = list_of_active_nodes(ctdb, nodemap, tmp_ctx, true);
5841 if (ctdb_client_async_control(ctdb, CTDB_CONTROL_DB_SET_HEALTHY,
5843 TIMELIMIT(), false, data,
5846 DEBUG(DEBUG_ERR, ("Failed to mark database as healthy.\n"));
5847 ctdb_ctrl_setrecmode(ctdb, TIMELIMIT(), options.pnn, CTDB_RECOVERY_ACTIVE);
5848 talloc_free(tmp_ctx);
5852 data.dptr = (void *)&generation;
5853 data.dsize = sizeof(generation);
5855 /* commit all the changes */
5856 if (ctdb_client_async_control(ctdb, CTDB_CONTROL_TRANSACTION_COMMIT,
5858 TIMELIMIT(), false, data,
5861 DEBUG(DEBUG_ERR, ("Unable to commit databases.\n"));
5862 ctdb_ctrl_setrecmode(ctdb, TIMELIMIT(), options.pnn, CTDB_RECOVERY_ACTIVE);
5863 talloc_free(tmp_ctx);
5867 /* thaw all nodes */
5868 nodes = list_of_active_nodes(ctdb, nodemap, tmp_ctx, true);
5869 if (ctdb_client_async_control(ctdb, CTDB_CONTROL_THAW,
5875 DEBUG(DEBUG_ERR, ("Unable to thaw nodes.\n"));
5876 ctdb_ctrl_setrecmode(ctdb, TIMELIMIT(), options.pnn, CTDB_RECOVERY_ACTIVE);
5877 talloc_free(tmp_ctx);
5881 DEBUG(DEBUG_ERR, ("Database wiped.\n"));
5883 talloc_free(tmp_ctx);
5890 static int control_dumpmemory(struct ctdb_context *ctdb, int argc, const char **argv)
5896 TALLOC_CTX *tmp_ctx = talloc_new(ctdb);
5897 ret = ctdb_control(ctdb, options.pnn, 0, CTDB_CONTROL_DUMP_MEMORY,
5898 0, tdb_null, tmp_ctx, &data, &res, NULL, &errmsg);
5899 if (ret != 0 || res != 0) {
5900 DEBUG(DEBUG_ERR,("Failed to dump memory - %s\n", errmsg));
5901 talloc_free(tmp_ctx);
5904 sys_write(1, data.dptr, data.dsize);
5905 talloc_free(tmp_ctx);
5910 handler for memory dumps
5912 static void mem_dump_handler(uint64_t srvid, TDB_DATA data, void *private_data)
5914 sys_write(1, data.dptr, data.dsize);
5919 dump memory usage on the recovery daemon
5921 static int control_rddumpmemory(struct ctdb_context *ctdb, int argc, const char **argv)
5925 struct ctdb_srvid_message rd;
5927 rd.pnn = ctdb_get_pnn(ctdb);
5928 rd.srvid = getpid();
5930 /* register a message port for receiveing the reply so that we
5931 can receive the reply
5933 ctdb_client_set_message_handler(ctdb, rd.srvid, mem_dump_handler, NULL);
5936 data.dptr = (uint8_t *)&rd;
5937 data.dsize = sizeof(rd);
5939 ret = ctdb_client_send_message(ctdb, options.pnn, CTDB_SRVID_MEM_DUMP, data);
5941 DEBUG(DEBUG_ERR,("Failed to send memdump request message to %u\n", options.pnn));
5945 /* this loop will terminate when we have received the reply */
5947 tevent_loop_once(ctdb->ev);
5954 send a message to a srvid
5956 static int control_msgsend(struct ctdb_context *ctdb, int argc, const char **argv)
5958 unsigned long srvid;
5966 srvid = strtoul(argv[0], NULL, 0);
5968 data.dptr = (uint8_t *)discard_const(argv[1]);
5969 data.dsize= strlen(argv[1]);
5971 ret = ctdb_client_send_message(ctdb, CTDB_BROADCAST_CONNECTED, srvid, data);
5973 DEBUG(DEBUG_ERR,("Failed to send memdump request message to %u\n", options.pnn));
5981 handler for msglisten
5983 static void msglisten_handler(uint64_t srvid, TDB_DATA data,
5988 printf("Message received: ");
5989 for (i=0;i<data.dsize;i++) {
5990 printf("%c", data.dptr[i]);
5996 listen for messages on a messageport
5998 static int control_msglisten(struct ctdb_context *ctdb, int argc, const char **argv)
6004 /* register a message port and listen for messages
6006 ctdb_client_set_message_handler(ctdb, srvid, msglisten_handler, NULL);
6007 printf("Listening for messages on srvid:%d\n", (int)srvid);
6010 tevent_loop_once(ctdb->ev);
6017 list all nodes in the cluster
6018 we parse the nodes file directly
6020 static int control_listnodes(struct ctdb_context *ctdb, int argc, const char **argv)
6022 TALLOC_CTX *mem_ctx = talloc_new(NULL);
6023 struct ctdb_node_map_old *node_map;
6026 assert_single_node_only();
6028 node_map = read_nodes_file(mem_ctx);
6029 if (node_map == NULL) {
6030 talloc_free(mem_ctx);
6034 for (i = 0; i < node_map->num; i++) {
6037 if (node_map->nodes[i].flags & NODE_FLAGS_DELETED) {
6040 addr = ctdb_addr_to_str(&node_map->nodes[i].addr);
6041 if (options.machinereadable){
6042 printm(":%d:%s:\n", node_map->nodes[i].pnn, addr);
6044 printf("%s\n", addr);
6047 talloc_free(mem_ctx);
6052 /**********************************************************************/
6053 /* reload the nodes file on all nodes */
6055 static void get_nodes_files_callback(struct ctdb_context *ctdb,
6056 uint32_t node_pnn, int32_t res,
6057 TDB_DATA outdata, void *callback_data)
6059 struct ctdb_node_map_old **maps =
6060 talloc_get_type(callback_data, struct ctdb_node_map_old *);
6062 if (outdata.dsize < offsetof(struct ctdb_node_map_old, nodes) ||
6063 outdata.dptr == NULL) {
6065 (__location__ " Invalid return data: %u %p\n",
6066 (unsigned)outdata.dsize, outdata.dptr));
6070 if (node_pnn >= talloc_array_length(maps)) {
6072 (__location__ " unexpected PNN %u\n", node_pnn));
6076 maps[node_pnn] = talloc_memdup(maps, outdata.dptr, outdata.dsize);
6079 static void get_nodes_files_fail_callback(struct ctdb_context *ctdb,
6080 uint32_t node_pnn, int32_t res,
6081 TDB_DATA outdata, void *callback_data)
6084 ("ERROR: Failed to get nodes file from node %u\n", node_pnn));
6087 static struct ctdb_node_map_old **
6088 ctdb_get_nodes_files(struct ctdb_context *ctdb,
6089 TALLOC_CTX *mem_ctx,
6090 struct timeval timeout,
6091 struct ctdb_node_map_old *nodemap)
6095 struct ctdb_node_map_old **maps;
6097 maps = talloc_zero_array(mem_ctx, struct ctdb_node_map_old *, nodemap->num);
6098 CTDB_NO_MEMORY_NULL(ctdb, maps);
6100 nodes = list_of_connected_nodes(ctdb, nodemap, mem_ctx, true);
6102 ret = ctdb_client_async_control(ctdb, CTDB_CONTROL_GET_NODES_FILE,
6103 nodes, 0, TIMELIMIT(),
6105 get_nodes_files_callback,
6106 get_nodes_files_fail_callback,
6116 static bool node_files_are_identical(struct ctdb_node_map_old *nm1,
6117 struct ctdb_node_map_old *nm2)
6121 if (nm1->num != nm2->num) {
6124 for (i = 0; i < nm1->num; i++) {
6125 if (memcmp(&nm1->nodes[i], &nm2->nodes[i],
6126 sizeof(struct ctdb_node_and_flags)) != 0) {
6134 static bool check_all_node_files_are_identical(struct ctdb_context *ctdb,
6135 TALLOC_CTX *mem_ctx,
6136 struct timeval timeout,
6137 struct ctdb_node_map_old *nodemap,
6138 struct ctdb_node_map_old *file_nodemap)
6140 static struct ctdb_node_map_old **maps;
6144 maps = ctdb_get_nodes_files(ctdb, mem_ctx, timeout, nodemap);
6149 for (i = 0; i < talloc_array_length(maps); i++) {
6150 if (maps[i] == NULL) {
6153 if (!node_files_are_identical(file_nodemap, maps[i])) {
6155 ("ERROR: Node file on node %u differs from current node (%u)\n",
6156 i, ctdb_get_pnn(ctdb)));
6165 reload the nodes file on the local node
6167 static bool sanity_check_nodes_file_changes(TALLOC_CTX *mem_ctx,
6168 struct ctdb_node_map_old *nodemap,
6169 struct ctdb_node_map_old *file_nodemap)
6172 bool should_abort = false;
6173 bool have_changes = false;
6175 for (i=0; i<nodemap->num; i++) {
6176 if (i >= file_nodemap->num) {
6178 ("ERROR: Node %u (%s) missing from nodes file\n",
6179 nodemap->nodes[i].pnn,
6180 ctdb_addr_to_str(&nodemap->nodes[i].addr)));
6181 should_abort = true;
6184 if (nodemap->nodes[i].flags & NODE_FLAGS_DELETED &&
6185 file_nodemap->nodes[i].flags & NODE_FLAGS_DELETED) {
6186 /* Node remains deleted */
6188 ("Node %u is unchanged (DELETED)\n",
6189 nodemap->nodes[i].pnn));
6190 } else if (!(nodemap->nodes[i].flags & NODE_FLAGS_DELETED) &&
6191 !(file_nodemap->nodes[i].flags & NODE_FLAGS_DELETED)) {
6192 /* Node not newly nor previously deleted */
6193 if (!ctdb_same_ip(&nodemap->nodes[i].addr,
6194 &file_nodemap->nodes[i].addr)) {
6196 ("ERROR: Node %u has changed IP address (was %s, now %s)\n",
6197 nodemap->nodes[i].pnn,
6198 /* ctdb_addr_to_str() returns a static */
6199 talloc_strdup(mem_ctx,
6200 ctdb_addr_to_str(&nodemap->nodes[i].addr)),
6201 ctdb_addr_to_str(&file_nodemap->nodes[i].addr)));
6202 should_abort = true;
6205 ("Node %u is unchanged\n",
6206 nodemap->nodes[i].pnn));
6207 if (nodemap->nodes[i].flags & NODE_FLAGS_DISCONNECTED) {
6208 DEBUG(DEBUG_WARNING,
6209 ("WARNING: Node %u is disconnected. You MUST fix this node manually!\n",
6210 nodemap->nodes[i].pnn));
6213 } else if (file_nodemap->nodes[i].flags & NODE_FLAGS_DELETED) {
6214 /* Node is being deleted */
6216 ("Node %u is DELETED\n",
6217 nodemap->nodes[i].pnn));
6218 have_changes = true;
6219 if (!(nodemap->nodes[i].flags & NODE_FLAGS_DISCONNECTED)) {
6221 ("ERROR: Node %u is still connected\n",
6222 nodemap->nodes[i].pnn));
6223 should_abort = true;
6225 } else if (nodemap->nodes[i].flags & NODE_FLAGS_DELETED) {
6226 /* Node was previously deleted */
6228 ("Node %u is UNDELETED\n", nodemap->nodes[i].pnn));
6229 have_changes = true;
6235 ("ERROR: Nodes will not be reloaded due to previous error\n"));
6236 talloc_free(mem_ctx);
6240 /* Leftover nodes in file are NEW */
6241 for (; i < file_nodemap->num; i++) {
6242 DEBUG(DEBUG_NOTICE, ("Node %u is NEW\n",
6243 file_nodemap->nodes[i].pnn));
6244 have_changes = true;
6247 return have_changes;
6250 static void reload_nodes_fail_callback(struct ctdb_context *ctdb,
6251 uint32_t node_pnn, int32_t res,
6252 TDB_DATA outdata, void *callback_data)
6254 DEBUG(DEBUG_WARNING,
6255 ("WARNING: Node %u failed to reload nodes. You MUST fix this node manually!\n",
6259 static int control_reload_nodes_file(struct ctdb_context *ctdb, int argc, const char **argv)
6262 struct ctdb_node_map_old *nodemap=NULL;
6263 TALLOC_CTX *tmp_ctx = talloc_new(NULL);
6264 struct ctdb_node_map_old *file_nodemap;
6268 assert_current_node_only(ctdb);
6270 /* Load both the current nodemap and the contents of the local
6271 * nodes file. Compare and sanity check them before doing
6274 ret = ctdb_ctrl_getnodemap(ctdb, TIMELIMIT(), CTDB_CURRENT_NODE, ctdb, &nodemap);
6276 DEBUG(DEBUG_ERR, ("Unable to get nodemap from local node\n"));
6280 file_nodemap = read_nodes_file(tmp_ctx);
6281 if (file_nodemap == NULL) {
6282 DEBUG(DEBUG_ERR,("Failed to read nodes file\n"));
6283 talloc_free(tmp_ctx);
6287 if (!check_all_node_files_are_identical(ctdb, tmp_ctx, TIMELIMIT(),
6288 nodemap, file_nodemap)) {
6292 if (!sanity_check_nodes_file_changes(tmp_ctx, nodemap, file_nodemap)) {
6294 ("No change in nodes file, skipping unnecessary reload\n"));
6295 talloc_free(tmp_ctx);
6299 /* Now make the changes */
6300 conn = list_of_connected_nodes(ctdb, nodemap, tmp_ctx, true);
6301 for (i = 0; i < talloc_array_length(conn); i++) {
6302 DEBUG(DEBUG_NOTICE, ("Reloading nodes file on node %u\n",
6306 /* Another timeout could be used, such as ReRecoveryTimeout or
6307 * a new one for this purpose. However, this is the simplest
6309 timeout = options.timelimit;
6310 srvid_broadcast(ctdb, CTDB_SRVID_DISABLE_RECOVERIES, &timeout,
6311 "Disable recoveries", true);
6314 ret = ctdb_client_async_control(ctdb, CTDB_CONTROL_RELOAD_NODES_FILE,
6315 conn, 0, TIMELIMIT(),
6317 NULL, reload_nodes_fail_callback,
6321 srvid_broadcast(ctdb, CTDB_SRVID_DISABLE_RECOVERIES, &timeout,
6322 "Enable recoveries", true);
6324 talloc_free(tmp_ctx);
6330 static const struct {
6332 int (*fn)(struct ctdb_context *, int, const char **);
6334 bool without_daemon; /* can be run without daemon running ? */
6337 } ctdb_commands[] = {
6338 { "version", control_version, true, true, "show version of ctdb" },
6339 { "status", control_status, true, false, "show node status" },
6340 { "uptime", control_uptime, true, false, "show node uptime" },
6341 { "ping", control_ping, true, false, "ping all nodes" },
6342 { "runstate", control_runstate, true, false, "get/check runstate of a node", "[setup|first_recovery|startup|running]" },
6343 { "getvar", control_getvar, true, false, "get a tunable variable", "<name>"},
6344 { "setvar", control_setvar, true, false, "set a tunable variable", "<name> <value>"},
6345 { "listvars", control_listvars, true, false, "list tunable variables"},
6346 { "statistics", control_statistics, false, false, "show statistics" },
6347 { "statisticsreset", control_statistics_reset, true, false, "reset statistics"},
6348 { "stats", control_stats, false, false, "show rolling statistics", "[number of history records]" },
6349 { "ip", control_ip, false, false, "show which public ip's that ctdb manages" },
6350 { "ipinfo", control_ipinfo, true, false, "show details about a public ip that ctdb manages", "<ip>" },
6351 { "ifaces", control_ifaces, true, false, "show which interfaces that ctdb manages" },
6352 { "setifacelink", control_setifacelink, true, false, "set interface link status", "<iface> <status>" },
6353 { "process-exists", control_process_exists, true, false, "check if a process exists on a node", "<pid>"},
6354 { "getdbmap", control_getdbmap, true, false, "show the database map" },
6355 { "getdbstatus", control_getdbstatus, true, false, "show the status of a database", "<dbname|dbid>" },
6356 { "catdb", control_catdb, true, false, "dump a ctdb database" , "<dbname|dbid>"},
6357 { "cattdb", control_cattdb, true, false, "dump a local tdb database" , "<dbname|dbid>"},
6358 { "getmonmode", control_getmonmode, true, false, "show monitoring mode" },
6359 { "getcapabilities", control_getcapabilities, true, false, "show node capabilities" },
6360 { "pnn", control_pnn, true, false, "show the pnn of the currnet node" },
6361 { "lvs", control_lvs, true, false, "show lvs configuration" },
6362 { "lvsmaster", control_lvsmaster, true, false, "show which node is the lvs master" },
6363 { "disablemonitor", control_disable_monmode,true, false, "set monitoring mode to DISABLE" },
6364 { "enablemonitor", control_enable_monmode, true, false, "set monitoring mode to ACTIVE" },
6365 { "setdebug", control_setdebug, true, false, "set debug level", "<EMERG|ALERT|CRIT|ERR|WARNING|NOTICE|INFO|DEBUG>" },
6366 { "getdebug", control_getdebug, true, false, "get debug level" },
6367 { "attach", control_attach, true, false, "attach to a database", "<dbname> [persistent]" },
6368 { "detach", control_detach, false, false, "detach from a database", "<dbname|dbid> [<dbname|dbid> ...]" },
6369 { "dumpmemory", control_dumpmemory, true, false, "dump memory map to stdout" },
6370 { "rddumpmemory", control_rddumpmemory, true, false, "dump memory map from the recovery daemon to stdout" },
6371 { "getpid", control_getpid, true, false, "get ctdbd process ID" },
6372 { "disable", control_disable, true, false, "disable a nodes public IP" },
6373 { "enable", control_enable, true, false, "enable a nodes public IP" },
6374 { "stop", control_stop, true, false, "stop a node" },
6375 { "continue", control_continue, true, false, "re-start a stopped node" },
6376 { "ban", control_ban, true, false, "ban a node from the cluster", "<bantime>"},
6377 { "unban", control_unban, true, false, "unban a node" },
6378 { "showban", control_showban, true, false, "show ban information"},
6379 { "shutdown", control_shutdown, true, false, "shutdown ctdbd" },
6380 { "recover", control_recover, true, false, "force recovery" },
6381 { "sync", control_ipreallocate, false, false, "wait until ctdbd has synced all state changes" },
6382 { "ipreallocate", control_ipreallocate, false, false, "force the recovery daemon to perform a ip reallocation procedure" },
6383 { "thaw", control_thaw, true, false, "thaw databases", "[priority:1-3]" },
6384 { "isnotrecmaster", control_isnotrecmaster, false, false, "check if the local node is recmaster or not" },
6385 { "killtcp", kill_tcp, false, false, "kill a tcp connection.", "[<srcip:port> <dstip:port>]" },
6386 { "gratiousarp", control_gratious_arp, false, false, "send a gratious arp", "<ip> <interface>" },
6387 { "tickle", tickle_tcp, false, false, "send a tcp tickle ack", "<srcip:port> <dstip:port>" },
6388 { "gettickles", control_get_tickles, false, false, "get the list of tickles registered for this ip", "<ip> [<port>]" },
6389 { "addtickle", control_add_tickle, false, false, "add a tickle for this ip", "<ip>:<port> <ip>:<port>" },
6391 { "deltickle", control_del_tickle, false, false, "delete a tickle from this ip", "<ip>:<port> <ip>:<port>" },
6393 { "regsrvid", regsrvid, false, false, "register a server id", "<pnn> <type> <id>" },
6394 { "unregsrvid", unregsrvid, false, false, "unregister a server id", "<pnn> <type> <id>" },
6395 { "chksrvid", chksrvid, false, false, "check if a server id exists", "<pnn> <type> <id>" },
6396 { "getsrvids", getsrvids, false, false, "get a list of all server ids"},
6397 { "check_srvids", check_srvids, false, false, "check if a srvid exists", "<id>+" },
6398 { "listnodes", control_listnodes, false, true, "list all nodes in the cluster"},
6399 { "reloadnodes", control_reload_nodes_file, false, false, "reload the nodes file and restart the transport on all nodes"},
6400 { "moveip", control_moveip, false, false, "move/failover an ip address to another node", "<ip> <node>"},
6401 { "rebalanceip", control_rebalanceip, false, false, "release an ip from the node and let recd rebalance it", "<ip>"},
6402 { "addip", control_addip, true, false, "add a ip address to a node", "<ip/mask> <iface>"},
6403 { "delip", control_delip, false, false, "delete an ip address from a node", "<ip>"},
6404 { "eventscript", control_eventscript, true, false, "run the eventscript with the given parameters on a node", "<arguments>"},
6405 { "backupdb", control_backupdb, false, false, "backup the database into a file.", "<dbname|dbid> <file>"},
6406 { "restoredb", control_restoredb, false, false, "restore the database from a file.", "<file> [dbname]"},
6407 { "dumpdbbackup", control_dumpdbbackup, false, true, "dump database backup from a file.", "<file>"},
6408 { "wipedb", control_wipedb, false, false, "wipe the contents of a database.", "<dbname|dbid>"},
6409 { "recmaster", control_recmaster, true, false, "show the pnn for the recovery master."},
6410 { "scriptstatus", control_scriptstatus, true, false, "show the status of the monitoring scripts (or all scripts)", "[all]"},
6411 { "enablescript", control_enablescript, true, false, "enable an eventscript", "<script>"},
6412 { "disablescript", control_disablescript, true, false, "disable an eventscript", "<script>"},
6413 { "natgwlist", control_natgwlist, false, true, "show the nodes belonging to this natgw configuration"},
6414 { "xpnn", control_xpnn, false, true, "find the pnn of the local node without talking to the daemon (unreliable)" },
6415 { "getreclock", control_getreclock, true, false, "Show the reclock file of a node"},
6416 { "setreclock", control_setreclock, true, false, "Set/clear the reclock file of a node", "[filename]"},
6417 { "setnatgwstate", control_setnatgwstate, false, false, "Set NATGW state to on/off", "{on|off}"},
6418 { "setlmasterrole", control_setlmasterrole, false, false, "Set LMASTER role to on/off", "{on|off}"},
6419 { "setrecmasterrole", control_setrecmasterrole, false, false, "Set RECMASTER role to on/off", "{on|off}"},
6420 { "setdbprio", control_setdbprio, false, false, "Set DB priority", "<dbname|dbid> <prio:1-3>"},
6421 { "getdbprio", control_getdbprio, false, false, "Get DB priority", "<dbname|dbid>"},
6422 { "setdbreadonly", control_setdbreadonly, false, false, "Set DB readonly capable", "<dbname|dbid>"},
6423 { "setdbsticky", control_setdbsticky, false, false, "Set DB sticky-records capable", "<dbname|dbid>"},
6424 { "msglisten", control_msglisten, false, false, "Listen on a srvid port for messages", "<msg srvid>"},
6425 { "msgsend", control_msgsend, false, false, "Send a message to srvid", "<srvid> <message>"},
6426 { "pfetch", control_pfetch, false, false, "fetch a record from a persistent database", "<dbname|dbid> <key> [<file>]" },
6427 { "pstore", control_pstore, false, false, "write a record to a persistent database", "<dbname|dbid> <key> <file containing record>" },
6428 { "pdelete", control_pdelete, false, false, "delete a record from a persistent database", "<dbname|dbid> <key>" },
6429 { "ptrans", control_ptrans, false, false, "update a persistent database (from stdin)", "<dbname|dbid>" },
6430 { "tfetch", control_tfetch, false, true, "fetch a record from a [c]tdb-file [-v]", "<tdb-file> <key> [<file>]" },
6431 { "tstore", control_tstore, false, true, "store a record (including ltdb header)", "<tdb-file> <key> <data> [<rsn> <dmaster> <flags>]" },
6432 { "readkey", control_readkey, true, false, "read the content off a database key", "<dbname|dbid> <key>" },
6433 { "writekey", control_writekey, true, false, "write to a database key", "<dbname|dbid> <key> <value>" },
6434 { "checktcpport", control_chktcpport, false, true, "check if a service is bound to a specific tcp port or not", "<port>" },
6435 { "rebalancenode", control_rebalancenode, false, false, "mark nodes as forced IP rebalancing targets", "[<pnn-list>]"},
6436 { "getdbseqnum", control_getdbseqnum, false, false, "get the sequence number off a database", "<dbname|dbid>" },
6437 { "nodestatus", control_nodestatus, true, false, "show and return node status", "[<pnn-list>]" },
6438 { "dbstatistics", control_dbstatistics, false, false, "show db statistics", "<dbname|dbid>" },
6439 { "reloadips", control_reloadips, false, false, "reload the public addresses file on specified nodes" , "[<pnn-list>]" },
6440 { "ipiface", control_ipiface, false, true, "Find which interface an ip address is hosted on", "<ip>" },
6446 static void usage(void)
6450 "Usage: ctdb [options] <control>\n" \
6452 " -n <node> choose node number, or 'all' (defaults to local node)\n"
6453 " -Y generate machine readable output\n"
6454 " -x <char> specify delimiter for machine readable output\n"
6455 " -v generate verbose output\n"
6456 " -t <timelimit> set timelimit for control in seconds (default %u)\n", options.timelimit);
6457 printf("Controls:\n");
6458 for (i=0;i<ARRAY_SIZE(ctdb_commands);i++) {
6459 printf(" %-15s %-27s %s\n",
6460 ctdb_commands[i].name,
6461 ctdb_commands[i].args?ctdb_commands[i].args:"",
6462 ctdb_commands[i].msg);
6468 static void ctdb_alarm(int sig)
6470 printf("Maximum runtime exceeded - exiting\n");
6477 int main(int argc, const char *argv[])
6479 struct ctdb_context *ctdb;
6480 char *nodestring = NULL;
6481 int machineparsable = 0;
6482 struct poptOption popt_options[] = {
6485 { "timelimit", 't', POPT_ARG_INT, &options.timelimit, 0, "timelimit", "integer" },
6486 { "node", 'n', POPT_ARG_STRING, &nodestring, 0, "node", "integer|all" },
6487 { "machinereadable", 'Y', POPT_ARG_NONE, &options.machinereadable, 0, "enable machine readable output", NULL },
6488 { NULL, 'x', POPT_ARG_STRING, &options.machineseparator, 0, "specify separator for machine readable output", "char" },
6489 { NULL, 'X', POPT_ARG_NONE, &machineparsable, 0, "enable machine parsable output with separator |", NULL },
6490 { "verbose", 'v', POPT_ARG_NONE, &options.verbose, 0, "enable verbose output", NULL },
6491 { "maxruntime", 'T', POPT_ARG_INT, &options.maxruntime, 0, "die if runtime exceeds this limit (in seconds)", "integer" },
6492 { "print-emptyrecords", 0, POPT_ARG_NONE, &options.printemptyrecords, 0, "print the empty records when dumping databases (catdb, cattdb, dumpdbbackup)", NULL },
6493 { "print-datasize", 0, POPT_ARG_NONE, &options.printdatasize, 0, "do not print record data when dumping databases, only the data size", NULL },
6494 { "print-lmaster", 0, POPT_ARG_NONE, &options.printlmaster, 0, "print the record's lmaster in catdb", NULL },
6495 { "print-hash", 0, POPT_ARG_NONE, &options.printhash, 0, "print the record's hash when dumping databases", NULL },
6496 { "print-recordflags", 0, POPT_ARG_NONE, &options.printrecordflags, 0, "print the record flags in catdb and dumpdbbackup", NULL },
6500 const char **extra_argv;
6504 struct tevent_context *ev;
6505 const char *control;
6509 /* set some defaults */
6510 options.maxruntime = 0;
6511 options.timelimit = 10;
6512 options.pnn = CTDB_CURRENT_NODE;
6514 pc = poptGetContext(argv[0], argc, argv, popt_options, POPT_CONTEXT_KEEP_FIRST);
6516 while ((opt = poptGetNextOpt(pc)) != -1) {
6519 DEBUG(DEBUG_ERR, ("Invalid option %s: %s\n",
6520 poptBadOption(pc, 0), poptStrerror(opt)));
6525 /* setup the remaining options for the main program to use */
6526 extra_argv = poptGetArgs(pc);
6529 while (extra_argv[extra_argc]) extra_argc++;
6532 if (extra_argc < 1) {
6536 if (options.maxruntime == 0) {
6537 const char *ctdb_timeout;
6538 ctdb_timeout = getenv("CTDB_TIMEOUT");
6539 if (ctdb_timeout != NULL) {
6540 options.maxruntime = strtoul(ctdb_timeout, NULL, 0);
6542 /* default timeout is 120 seconds */
6543 options.maxruntime = 120;
6547 if (machineparsable) {
6548 options.machineseparator = "|";
6550 if (options.machineseparator != NULL) {
6551 if (strlen(options.machineseparator) != 1) {
6552 printf("Invalid separator \"%s\" - "
6553 "must be single character\n",
6554 options.machineseparator);
6559 options.machinereadable = true;
6560 } else if (options.machinereadable) {
6561 options.machineseparator = ":";
6564 signal(SIGALRM, ctdb_alarm);
6565 alarm(options.maxruntime);
6567 control = extra_argv[0];
6569 /* Default value for CTDB_BASE - don't override */
6570 setenv("CTDB_BASE", CTDB_ETCDIR, 0);
6572 ev = tevent_context_init(NULL);
6574 DEBUG(DEBUG_ERR, ("Failed to initialize event system\n"));
6578 for (i=0;i<ARRAY_SIZE(ctdb_commands);i++) {
6579 if (strcmp(control, ctdb_commands[i].name) == 0) {
6584 if (i == ARRAY_SIZE(ctdb_commands)) {
6585 DEBUG(DEBUG_ERR, ("Unknown control '%s'\n", control));
6589 if (ctdb_commands[i].without_daemon == true) {
6590 if (nodestring != NULL) {
6591 DEBUG(DEBUG_ERR, ("Can't specify node(s) with \"ctdb %s\"\n", control));
6594 return ctdb_commands[i].fn(NULL, extra_argc-1, extra_argv+1);
6597 /* initialise ctdb */
6598 ctdb = ctdb_cmdline_client(ev, TIMELIMIT());
6602 DEBUG(DEBUG_ERR, ("Failed to init ctdb\n"));
6604 pnn = find_node_xpnn();
6607 ("Is this node part of a CTDB cluster?\n"));
6612 /* setup the node number(s) to contact */
6613 if (!parse_nodestring(ctdb, ctdb, nodestring, CTDB_CURRENT_NODE, false,
6614 &options.nodes, &options.pnn)) {
6618 if (options.pnn == CTDB_CURRENT_NODE) {
6619 options.pnn = options.nodes[0];
6622 if (ctdb_commands[i].auto_all &&
6623 ((options.pnn == CTDB_BROADCAST_ALL) ||
6624 (options.pnn == CTDB_MULTICAST))) {
6628 for (j = 0; j < talloc_array_length(options.nodes); j++) {
6629 options.pnn = options.nodes[j];
6630 ret |= ctdb_commands[i].fn(ctdb, extra_argc-1, extra_argv+1);
6633 ret = ctdb_commands[i].fn(ctdb, extra_argc-1, extra_argv+1);
6638 (void)poptFreeContext(pc);