4 Copyright (C) Andrew Tridgell 2006
6 This program is free software; you can redistribute it and/or modify
7 it under the terms of the GNU General Public License as published by
8 the Free Software Foundation; either version 3 of the License, or
9 (at your option) any later version.
11 This program is distributed in the hope that it will be useful,
12 but WITHOUT ANY WARRANTY; without even the implied warranty of
13 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
14 GNU General Public License for more details.
16 You should have received a copy of the GNU General Public License
17 along with this program; if not, see <http://www.gnu.org/licenses/>.
21 #include "lib/events/events.h"
22 #include "lib/tdb/include/tdb.h"
23 #include "system/network.h"
24 #include "system/filesys.h"
25 #include "system/wait.h"
26 #include "../include/ctdb_private.h"
28 int LogLevel = DEBUG_NOTICE;
29 int this_log_level = 0;
32 return error string for last error
34 const char *ctdb_errstr(struct ctdb_context *ctdb)
41 remember an error message
43 void ctdb_set_error(struct ctdb_context *ctdb, const char *fmt, ...)
46 talloc_free(ctdb->err_msg);
48 ctdb->err_msg = talloc_vasprintf(ctdb, fmt, ap);
49 DEBUG(DEBUG_ERR,("ctdb error: %s\n", ctdb->err_msg));
54 a fatal internal error occurred - no hope for recovery
56 void ctdb_fatal(struct ctdb_context *ctdb, const char *msg)
58 DEBUG(DEBUG_ALERT,("ctdb fatal error: %s\n", msg));
65 int ctdb_parse_address(struct ctdb_context *ctdb,
66 TALLOC_CTX *mem_ctx, const char *str,
67 struct ctdb_address *address)
72 se = getservbyname("ctdb", "tcp");
75 address->address = talloc_strdup(mem_ctx, str);
76 CTDB_NO_MEMORY(ctdb, address->address);
79 address->port = CTDB_PORT;
81 address->port = ntohs(se->s_port);
88 check if two addresses are the same
90 bool ctdb_same_address(struct ctdb_address *a1, struct ctdb_address *a2)
92 return strcmp(a1->address, a2->address) == 0 && a1->port == a2->port;
97 hash function for mapping data to a VNN - taken from tdb
99 uint32_t ctdb_hash(const TDB_DATA *key)
101 uint32_t value; /* Used to compute the hash value. */
102 uint32_t i; /* Used to cycle through random values. */
104 /* Set the initial value from the key size. */
105 for (value = 0x238F13AF * key->dsize, i=0; i < key->dsize; i++)
106 value = (value + (key->dptr[i] << (i*5 % 24)));
108 return (1103515243 * value + 12345);
112 a type checking varient of idr_find
114 static void *_idr_find_type(struct idr_context *idp, int id, const char *type, const char *location)
116 void *p = idr_find(idp, id);
117 if (p && talloc_check_name(p, type) == NULL) {
118 DEBUG(DEBUG_ERR,("%s idr_find_type expected type %s but got %s\n",
119 location, type, talloc_get_name(p)));
127 update a max latency number
129 void ctdb_latency(struct ctdb_db_context *ctdb_db, const char *name, double *latency, struct timeval t)
131 double l = timeval_elapsed(&t);
136 if (ctdb_db->ctdb->tunable.log_latency_ms !=0) {
137 if (l*1000 > ctdb_db->ctdb->tunable.log_latency_ms) {
138 DEBUG(DEBUG_WARNING, ("High latency %fs for operation %s on database %s\n", l*1000000, name, ctdb_db->db_name));
144 update a reclock latency number
146 void ctdb_reclock_latency(struct ctdb_context *ctdb, const char *name, double *latency, double l)
152 if (ctdb->tunable.reclock_latency_ms !=0) {
153 if (l*1000 > ctdb->tunable.reclock_latency_ms) {
154 DEBUG(DEBUG_ERR, ("High RECLOCK latency %fs for operation %s\n", l, name));
159 uint32_t ctdb_reqid_new(struct ctdb_context *ctdb, void *state)
163 id = ctdb->idr_cnt++ & 0xFFFF;
164 id |= (idr_get_new(ctdb->idr, state, 0xFFFF)<<16);
168 void *_ctdb_reqid_find(struct ctdb_context *ctdb, uint32_t reqid, const char *type, const char *location)
172 p = _idr_find_type(ctdb->idr, (reqid>>16)&0xFFFF, type, location);
174 DEBUG(DEBUG_WARNING, ("Could not find idr:%u\n",reqid));
181 void ctdb_reqid_remove(struct ctdb_context *ctdb, uint32_t reqid)
185 ret = idr_remove(ctdb->idr, (reqid>>16)&0xFFFF);
187 DEBUG(DEBUG_ERR, ("Removing idr that does not exist\n"));
193 form a ctdb_rec_data record from a key/data pair
195 note that header may be NULL. If not NULL then it is included in the data portion
198 struct ctdb_rec_data *ctdb_marshall_record(TALLOC_CTX *mem_ctx, uint32_t reqid,
200 struct ctdb_ltdb_header *header,
204 struct ctdb_rec_data *d;
206 length = offsetof(struct ctdb_rec_data, data) + key.dsize +
207 data.dsize + (header?sizeof(*header):0);
208 d = (struct ctdb_rec_data *)talloc_size(mem_ctx, length);
214 d->keylen = key.dsize;
215 memcpy(&d->data[0], key.dptr, key.dsize);
217 d->datalen = data.dsize + sizeof(*header);
218 memcpy(&d->data[key.dsize], header, sizeof(*header));
219 memcpy(&d->data[key.dsize+sizeof(*header)], data.dptr, data.dsize);
221 d->datalen = data.dsize;
222 memcpy(&d->data[key.dsize], data.dptr, data.dsize);
228 /* helper function for marshalling multiple records */
229 struct ctdb_marshall_buffer *ctdb_marshall_add(TALLOC_CTX *mem_ctx,
230 struct ctdb_marshall_buffer *m,
234 struct ctdb_ltdb_header *header,
237 struct ctdb_rec_data *r;
238 size_t m_size, r_size;
239 struct ctdb_marshall_buffer *m2;
241 r = ctdb_marshall_record(mem_ctx, reqid, key, header, data);
248 m = talloc_zero_size(mem_ctx, offsetof(struct ctdb_marshall_buffer, data));
255 m_size = talloc_get_size(m);
256 r_size = talloc_get_size(r);
258 m2 = talloc_realloc_size(mem_ctx, m, m_size + r_size);
264 memcpy(m_size + (uint8_t *)m2, r, r_size);
273 /* we've finished marshalling, return a data blob with the marshalled records */
274 TDB_DATA ctdb_marshall_finish(struct ctdb_marshall_buffer *m)
277 data.dptr = (uint8_t *)m;
278 data.dsize = talloc_get_size(m);
283 loop over a marshalling buffer
285 - pass r==NULL to start
286 - loop the number of times indicated by m->count
288 struct ctdb_rec_data *ctdb_marshall_loop_next(struct ctdb_marshall_buffer *m, struct ctdb_rec_data *r,
290 struct ctdb_ltdb_header *header,
291 TDB_DATA *key, TDB_DATA *data)
294 r = (struct ctdb_rec_data *)&m->data[0];
296 r = (struct ctdb_rec_data *)(r->length + (uint8_t *)r);
304 key->dptr = &r->data[0];
305 key->dsize = r->keylen;
308 data->dptr = &r->data[r->keylen];
309 data->dsize = r->datalen;
310 if (header != NULL) {
311 data->dptr += sizeof(*header);
312 data->dsize -= sizeof(*header);
316 if (header != NULL) {
317 if (r->datalen < sizeof(*header)) {
320 *header = *(struct ctdb_ltdb_header *)&r->data[r->keylen];
327 if possible, make this task very high priority
329 void ctdb_high_priority(struct ctdb_context *ctdb)
332 if (nice(-20) == -1 && errno != 0) {
333 DEBUG(DEBUG_WARNING,("Unable to renice self: %s\n",
336 DEBUG(DEBUG_NOTICE,("Scheduler says I'm nice: %i\n",
337 getpriority(PRIO_PROCESS, getpid())));
341 void set_nonblocking(int fd)
344 v = fcntl(fd, F_GETFL, 0);
345 fcntl(fd, F_SETFL, v | O_NONBLOCK);
348 void set_close_on_exec(int fd)
351 v = fcntl(fd, F_GETFD, 0);
352 fcntl(fd, F_SETFD, v | FD_CLOEXEC);
356 bool parse_ipv4(const char *s, unsigned port, struct sockaddr_in *sin)
358 sin->sin_family = AF_INET;
359 sin->sin_port = htons(port);
361 if (inet_pton(AF_INET, s, &sin->sin_addr) != 1) {
362 DEBUG(DEBUG_ERR, (__location__ " Failed to translate %s into sin_addr\n", s));
369 static bool parse_ipv6(const char *s, const char *iface, unsigned port, ctdb_sock_addr *saddr)
371 saddr->ip6.sin6_family = AF_INET6;
372 saddr->ip6.sin6_port = htons(port);
373 saddr->ip6.sin6_flowinfo = 0;
374 saddr->ip6.sin6_scope_id = 0;
376 if (inet_pton(AF_INET6, s, &saddr->ip6.sin6_addr) != 1) {
377 DEBUG(DEBUG_ERR, (__location__ " Failed to translate %s into sin6_addr\n", s));
381 if (iface && IN6_IS_ADDR_LINKLOCAL(&saddr->ip6.sin6_addr)) {
382 saddr->ip6.sin6_scope_id = if_nametoindex(iface);
390 bool parse_ip_port(const char *addr, ctdb_sock_addr *saddr)
392 TALLOC_CTX *tmp_ctx = talloc_new(NULL);
398 s = talloc_strdup(tmp_ctx, addr);
400 DEBUG(DEBUG_ERR, (__location__ " Failed strdup()\n"));
401 talloc_free(tmp_ctx);
407 DEBUG(DEBUG_ERR, (__location__ " This addr: %s does not contain a port number\n", s));
408 talloc_free(tmp_ctx);
412 port = strtoul(p+1, &endp, 10);
413 if (endp == NULL || *endp != 0) {
414 /* trailing garbage */
415 DEBUG(DEBUG_ERR, (__location__ " Trailing garbage after the port in %s\n", s));
416 talloc_free(tmp_ctx);
422 /* now is this a ipv4 or ipv6 address ?*/
423 ret = parse_ip(s, NULL, port, saddr);
425 talloc_free(tmp_ctx);
432 bool parse_ip(const char *addr, const char *iface, unsigned port, ctdb_sock_addr *saddr)
437 /* now is this a ipv4 or ipv6 address ?*/
438 p = index(addr, ':');
440 ret = parse_ipv4(addr, port, &saddr->ip);
442 ret = parse_ipv6(addr, iface, port, saddr);
451 bool parse_ip_mask(const char *str, const char *iface, ctdb_sock_addr *addr, unsigned *mask)
453 TALLOC_CTX *tmp_ctx = talloc_new(NULL);
459 s = talloc_strdup(tmp_ctx, str);
461 DEBUG(DEBUG_ERR, (__location__ " Failed strdup()\n"));
462 talloc_free(tmp_ctx);
468 DEBUG(DEBUG_ERR, (__location__ " This addr: %s does not contain a mask\n", s));
469 talloc_free(tmp_ctx);
473 *mask = strtoul(p+1, &endp, 10);
474 if (endp == NULL || *endp != 0) {
475 /* trailing garbage */
476 DEBUG(DEBUG_ERR, (__location__ " Trailing garbage after the mask in %s\n", s));
477 talloc_free(tmp_ctx);
483 /* now is this a ipv4 or ipv6 address ?*/
484 ret = parse_ip(s, iface, 0, addr);
486 talloc_free(tmp_ctx);
491 This is used to canonicalize a ctdb_sock_addr structure.
493 void ctdb_canonicalize_ip(const ctdb_sock_addr *ip, ctdb_sock_addr *cip)
495 char prefix[12] = { 0,0,0,0,0,0,0,0,0,0,0xff,0xff };
497 memcpy(cip, ip, sizeof (*cip));
499 if ( (ip->sa.sa_family == AF_INET6)
500 && !memcmp(&ip->ip6.sin6_addr, prefix, 12)) {
501 memset(cip, 0, sizeof(*cip));
502 #ifdef HAVE_SOCK_SIN_LEN
503 cip->ip.sin_len = sizeof(*cip);
505 cip->ip.sin_family = AF_INET;
506 cip->ip.sin_port = ip->ip6.sin6_port;
507 memcpy(&cip->ip.sin_addr, &ip->ip6.sin6_addr.s6_addr32[3], 4);
511 bool ctdb_same_ip(const ctdb_sock_addr *tip1, const ctdb_sock_addr *tip2)
513 ctdb_sock_addr ip1, ip2;
515 ctdb_canonicalize_ip(tip1, &ip1);
516 ctdb_canonicalize_ip(tip2, &ip2);
518 if (ip1.sa.sa_family != ip2.sa.sa_family) {
522 switch (ip1.sa.sa_family) {
524 return ip1.ip.sin_addr.s_addr == ip2.ip.sin_addr.s_addr;
526 return !memcmp(&ip1.ip6.sin6_addr.s6_addr[0],
527 &ip2.ip6.sin6_addr.s6_addr[0],
530 DEBUG(DEBUG_ERR, (__location__ " CRITICAL Can not compare sockaddr structures of type %u\n", ip1.sa.sa_family));
538 compare two ctdb_sock_addr structures
540 bool ctdb_same_sockaddr(const ctdb_sock_addr *ip1, const ctdb_sock_addr *ip2)
542 return ctdb_same_ip(ip1, ip2) && ip1->ip.sin_port == ip2->ip.sin_port;
545 char *ctdb_addr_to_str(ctdb_sock_addr *addr)
547 static char cip[128] = "";
549 switch (addr->sa.sa_family) {
551 inet_ntop(addr->ip.sin_family, &addr->ip.sin_addr, cip, sizeof(cip));
554 inet_ntop(addr->ip6.sin6_family, &addr->ip6.sin6_addr, cip, sizeof(cip));
557 DEBUG(DEBUG_ERR, (__location__ " ERROR, unknown family %u\n", addr->sa.sa_family));
563 unsigned ctdb_addr_to_port(ctdb_sock_addr *addr)
565 switch (addr->sa.sa_family) {
567 return ntohs(addr->ip.sin_port);
570 return ntohs(addr->ip6.sin6_port);
573 DEBUG(DEBUG_ERR, (__location__ " ERROR, unknown family %u\n", addr->sa.sa_family));
579 void ctdb_block_signal(int signum)
583 sigaddset(&set,signum);
584 sigprocmask(SIG_BLOCK,&set,NULL);
587 void ctdb_unblock_signal(int signum)
591 sigaddset(&set,signum);
592 sigprocmask(SIG_UNBLOCK,&set,NULL);
595 struct debug_levels debug_levels[] = {
596 {DEBUG_EMERG, "EMERG"},
597 {DEBUG_ALERT, "ALERT"},
598 {DEBUG_CRIT, "CRIT"},
600 {DEBUG_WARNING, "WARNING"},
601 {DEBUG_NOTICE, "NOTICE"},
602 {DEBUG_INFO, "INFO"},
603 {DEBUG_DEBUG, "DEBUG"},
607 const char *get_debug_by_level(int32_t level)
611 for (i=0; debug_levels[i].description != NULL; i++) {
612 if (debug_levels[i].level == level) {
613 return debug_levels[i].description;
619 int32_t get_debug_by_desc(const char *desc)
623 for (i=0; debug_levels[i].description != NULL; i++) {
624 if (!strcmp(debug_levels[i].description, desc)) {
625 return debug_levels[i].level;
632 const char *ctdb_eventscript_call_names[] = {