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/tevent/tevent.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 "system/shmem.h"
27 #include "../include/ctdb_private.h"
29 int LogLevel = DEBUG_NOTICE;
30 int this_log_level = 0;
33 return error string for last error
35 const char *ctdb_errstr(struct ctdb_context *ctdb)
42 remember an error message
44 void ctdb_set_error(struct ctdb_context *ctdb, const char *fmt, ...)
47 talloc_free(ctdb->err_msg);
49 ctdb->err_msg = talloc_vasprintf(ctdb, fmt, ap);
50 DEBUG(DEBUG_ERR,("ctdb error: %s\n", ctdb->err_msg));
55 a fatal internal error occurred - no hope for recovery
57 void ctdb_fatal(struct ctdb_context *ctdb, const char *msg)
59 DEBUG(DEBUG_ALERT,("ctdb fatal error: %s\n", msg));
66 int ctdb_parse_address(struct ctdb_context *ctdb,
67 TALLOC_CTX *mem_ctx, const char *str,
68 struct ctdb_address *address)
73 se = getservbyname("ctdb", "tcp");
76 address->address = talloc_strdup(mem_ctx, str);
77 CTDB_NO_MEMORY(ctdb, address->address);
80 address->port = CTDB_PORT;
82 address->port = ntohs(se->s_port);
89 check if two addresses are the same
91 bool ctdb_same_address(struct ctdb_address *a1, struct ctdb_address *a2)
93 return strcmp(a1->address, a2->address) == 0 && a1->port == a2->port;
98 hash function for mapping data to a VNN - taken from tdb
100 uint32_t ctdb_hash(const TDB_DATA *key)
102 uint32_t value; /* Used to compute the hash value. */
103 uint32_t i; /* Used to cycle through random values. */
105 /* Set the initial value from the key size. */
106 for (value = 0x238F13AF * key->dsize, i=0; i < key->dsize; i++)
107 value = (value + (key->dptr[i] << (i*5 % 24)));
109 return (1103515243 * value + 12345);
113 a type checking varient of idr_find
115 static void *_idr_find_type(struct idr_context *idp, int id, const char *type, const char *location)
117 void *p = idr_find(idp, id);
118 if (p && talloc_check_name(p, type) == NULL) {
119 DEBUG(DEBUG_ERR,("%s idr_find_type expected type %s but got %s\n",
120 location, type, talloc_get_name(p)));
128 update a max latency number
130 void ctdb_latency(struct ctdb_db_context *ctdb_db, const char *name, double *latency, struct timeval t)
132 double l = timeval_elapsed(&t);
137 if (ctdb_db->ctdb->tunable.log_latency_ms !=0) {
138 if (l*1000 > ctdb_db->ctdb->tunable.log_latency_ms) {
139 DEBUG(DEBUG_WARNING, ("High latency %.6fs for operation %s on database %s\n", l, name, ctdb_db->db_name));
145 update a reclock latency number
147 void ctdb_reclock_latency(struct ctdb_context *ctdb, const char *name, double *latency, double l)
153 if (ctdb->tunable.reclock_latency_ms !=0) {
154 if (l*1000 > ctdb->tunable.reclock_latency_ms) {
155 DEBUG(DEBUG_ERR, ("High RECLOCK latency %fs for operation %s\n", l, name));
160 uint32_t ctdb_reqid_new(struct ctdb_context *ctdb, void *state)
162 int id = idr_get_new_above(ctdb->idr, state, ctdb->lastid+1, INT_MAX);
164 DEBUG(DEBUG_NOTICE, ("Reqid wrap!\n"));
165 id = idr_get_new(ctdb->idr, state, INT_MAX);
171 void *_ctdb_reqid_find(struct ctdb_context *ctdb, uint32_t reqid, const char *type, const char *location)
175 p = _idr_find_type(ctdb->idr, reqid, type, location);
177 DEBUG(DEBUG_WARNING, ("Could not find idr:%u\n",reqid));
184 void ctdb_reqid_remove(struct ctdb_context *ctdb, uint32_t reqid)
188 ret = idr_remove(ctdb->idr, reqid);
190 DEBUG(DEBUG_ERR, ("Removing idr that does not exist\n"));
196 form a ctdb_rec_data record from a key/data pair
198 note that header may be NULL. If not NULL then it is included in the data portion
201 struct ctdb_rec_data *ctdb_marshall_record(TALLOC_CTX *mem_ctx, uint32_t reqid,
203 struct ctdb_ltdb_header *header,
207 struct ctdb_rec_data *d;
209 length = offsetof(struct ctdb_rec_data, data) + key.dsize +
210 data.dsize + (header?sizeof(*header):0);
211 d = (struct ctdb_rec_data *)talloc_size(mem_ctx, length);
217 d->keylen = key.dsize;
218 memcpy(&d->data[0], key.dptr, key.dsize);
220 d->datalen = data.dsize + sizeof(*header);
221 memcpy(&d->data[key.dsize], header, sizeof(*header));
222 memcpy(&d->data[key.dsize+sizeof(*header)], data.dptr, data.dsize);
224 d->datalen = data.dsize;
225 memcpy(&d->data[key.dsize], data.dptr, data.dsize);
231 /* helper function for marshalling multiple records */
232 struct ctdb_marshall_buffer *ctdb_marshall_add(TALLOC_CTX *mem_ctx,
233 struct ctdb_marshall_buffer *m,
237 struct ctdb_ltdb_header *header,
240 struct ctdb_rec_data *r;
241 size_t m_size, r_size;
242 struct ctdb_marshall_buffer *m2;
244 r = ctdb_marshall_record(mem_ctx, reqid, key, header, data);
251 m = talloc_zero_size(mem_ctx, offsetof(struct ctdb_marshall_buffer, data));
258 m_size = talloc_get_size(m);
259 r_size = talloc_get_size(r);
261 m2 = talloc_realloc_size(mem_ctx, m, m_size + r_size);
267 memcpy(m_size + (uint8_t *)m2, r, r_size);
276 /* we've finished marshalling, return a data blob with the marshalled records */
277 TDB_DATA ctdb_marshall_finish(struct ctdb_marshall_buffer *m)
280 data.dptr = (uint8_t *)m;
281 data.dsize = talloc_get_size(m);
286 loop over a marshalling buffer
288 - pass r==NULL to start
289 - loop the number of times indicated by m->count
291 struct ctdb_rec_data *ctdb_marshall_loop_next(struct ctdb_marshall_buffer *m, struct ctdb_rec_data *r,
293 struct ctdb_ltdb_header *header,
294 TDB_DATA *key, TDB_DATA *data)
297 r = (struct ctdb_rec_data *)&m->data[0];
299 r = (struct ctdb_rec_data *)(r->length + (uint8_t *)r);
307 key->dptr = &r->data[0];
308 key->dsize = r->keylen;
311 data->dptr = &r->data[r->keylen];
312 data->dsize = r->datalen;
313 if (header != NULL) {
314 data->dptr += sizeof(*header);
315 data->dsize -= sizeof(*header);
319 if (header != NULL) {
320 if (r->datalen < sizeof(*header)) {
323 *header = *(struct ctdb_ltdb_header *)&r->data[r->keylen];
330 if possible, make this task very high priority
332 void ctdb_high_priority(struct ctdb_context *ctdb)
335 if (nice(-20) == -1 && errno != 0) {
336 DEBUG(DEBUG_WARNING,("Unable to renice self: %s\n",
339 DEBUG(DEBUG_NOTICE,("Scheduler says I'm nice: %i\n",
340 getpriority(PRIO_PROCESS, getpid())));
345 make ourselves slightly nicer: eg. a ctdb child.
347 void ctdb_reduce_priority(struct ctdb_context *ctdb)
350 if (nice(10) == -1 && errno != 0) {
351 DEBUG(DEBUG_WARNING,("Unable to lower priority: %s\n",
356 void set_nonblocking(int fd)
359 v = fcntl(fd, F_GETFL, 0);
360 fcntl(fd, F_SETFL, v | O_NONBLOCK);
363 void set_close_on_exec(int fd)
366 v = fcntl(fd, F_GETFD, 0);
367 fcntl(fd, F_SETFD, v | FD_CLOEXEC);
371 bool parse_ipv4(const char *s, unsigned port, struct sockaddr_in *sin)
373 sin->sin_family = AF_INET;
374 sin->sin_port = htons(port);
376 if (inet_pton(AF_INET, s, &sin->sin_addr) != 1) {
377 DEBUG(DEBUG_ERR, (__location__ " Failed to translate %s into sin_addr\n", s));
384 static bool parse_ipv6(const char *s, const char *ifaces, unsigned port, ctdb_sock_addr *saddr)
386 saddr->ip6.sin6_family = AF_INET6;
387 saddr->ip6.sin6_port = htons(port);
388 saddr->ip6.sin6_flowinfo = 0;
389 saddr->ip6.sin6_scope_id = 0;
391 if (inet_pton(AF_INET6, s, &saddr->ip6.sin6_addr) != 1) {
392 DEBUG(DEBUG_ERR, (__location__ " Failed to translate %s into sin6_addr\n", s));
396 if (ifaces && IN6_IS_ADDR_LINKLOCAL(&saddr->ip6.sin6_addr)) {
397 if (strchr(ifaces, ',')) {
398 DEBUG(DEBUG_ERR, (__location__ " Link local address %s "
399 "is specified for multiple ifaces %s\n",
403 saddr->ip6.sin6_scope_id = if_nametoindex(ifaces);
411 bool parse_ip_port(const char *addr, ctdb_sock_addr *saddr)
413 TALLOC_CTX *tmp_ctx = talloc_new(NULL);
419 s = talloc_strdup(tmp_ctx, addr);
421 DEBUG(DEBUG_ERR, (__location__ " Failed strdup()\n"));
422 talloc_free(tmp_ctx);
428 DEBUG(DEBUG_ERR, (__location__ " This addr: %s does not contain a port number\n", s));
429 talloc_free(tmp_ctx);
433 port = strtoul(p+1, &endp, 10);
434 if (endp == NULL || *endp != 0) {
435 /* trailing garbage */
436 DEBUG(DEBUG_ERR, (__location__ " Trailing garbage after the port in %s\n", s));
437 talloc_free(tmp_ctx);
443 /* now is this a ipv4 or ipv6 address ?*/
444 ret = parse_ip(s, NULL, port, saddr);
446 talloc_free(tmp_ctx);
453 bool parse_ip(const char *addr, const char *ifaces, unsigned port, ctdb_sock_addr *saddr)
458 /* now is this a ipv4 or ipv6 address ?*/
459 p = index(addr, ':');
461 ret = parse_ipv4(addr, port, &saddr->ip);
463 ret = parse_ipv6(addr, ifaces, port, saddr);
472 bool parse_ip_mask(const char *str, const char *ifaces, ctdb_sock_addr *addr, unsigned *mask)
474 TALLOC_CTX *tmp_ctx = talloc_new(NULL);
480 s = talloc_strdup(tmp_ctx, str);
482 DEBUG(DEBUG_ERR, (__location__ " Failed strdup()\n"));
483 talloc_free(tmp_ctx);
489 DEBUG(DEBUG_ERR, (__location__ " This addr: %s does not contain a mask\n", s));
490 talloc_free(tmp_ctx);
494 *mask = strtoul(p+1, &endp, 10);
495 if (endp == NULL || *endp != 0) {
496 /* trailing garbage */
497 DEBUG(DEBUG_ERR, (__location__ " Trailing garbage after the mask in %s\n", s));
498 talloc_free(tmp_ctx);
504 /* now is this a ipv4 or ipv6 address ?*/
505 ret = parse_ip(s, ifaces, 0, addr);
507 talloc_free(tmp_ctx);
512 This is used to canonicalize a ctdb_sock_addr structure.
514 void ctdb_canonicalize_ip(const ctdb_sock_addr *ip, ctdb_sock_addr *cip)
516 char prefix[12] = { 0,0,0,0,0,0,0,0,0,0,0xff,0xff };
518 memcpy(cip, ip, sizeof (*cip));
520 if ( (ip->sa.sa_family == AF_INET6)
521 && !memcmp(&ip->ip6.sin6_addr, prefix, 12)) {
522 memset(cip, 0, sizeof(*cip));
523 #ifdef HAVE_SOCK_SIN_LEN
524 cip->ip.sin_len = sizeof(*cip);
526 cip->ip.sin_family = AF_INET;
527 cip->ip.sin_port = ip->ip6.sin6_port;
528 memcpy(&cip->ip.sin_addr, &ip->ip6.sin6_addr.s6_addr32[3], 4);
532 bool ctdb_same_ip(const ctdb_sock_addr *tip1, const ctdb_sock_addr *tip2)
534 ctdb_sock_addr ip1, ip2;
536 ctdb_canonicalize_ip(tip1, &ip1);
537 ctdb_canonicalize_ip(tip2, &ip2);
539 if (ip1.sa.sa_family != ip2.sa.sa_family) {
543 switch (ip1.sa.sa_family) {
545 return ip1.ip.sin_addr.s_addr == ip2.ip.sin_addr.s_addr;
547 return !memcmp(&ip1.ip6.sin6_addr.s6_addr[0],
548 &ip2.ip6.sin6_addr.s6_addr[0],
551 DEBUG(DEBUG_ERR, (__location__ " CRITICAL Can not compare sockaddr structures of type %u\n", ip1.sa.sa_family));
559 compare two ctdb_sock_addr structures
561 bool ctdb_same_sockaddr(const ctdb_sock_addr *ip1, const ctdb_sock_addr *ip2)
563 return ctdb_same_ip(ip1, ip2) && ip1->ip.sin_port == ip2->ip.sin_port;
566 char *ctdb_addr_to_str(ctdb_sock_addr *addr)
568 static char cip[128] = "";
570 switch (addr->sa.sa_family) {
572 inet_ntop(addr->ip.sin_family, &addr->ip.sin_addr, cip, sizeof(cip));
575 inet_ntop(addr->ip6.sin6_family, &addr->ip6.sin6_addr, cip, sizeof(cip));
578 DEBUG(DEBUG_ERR, (__location__ " ERROR, unknown family %u\n", addr->sa.sa_family));
584 unsigned ctdb_addr_to_port(ctdb_sock_addr *addr)
586 switch (addr->sa.sa_family) {
588 return ntohs(addr->ip.sin_port);
591 return ntohs(addr->ip6.sin6_port);
594 DEBUG(DEBUG_ERR, (__location__ " ERROR, unknown family %u\n", addr->sa.sa_family));
600 void ctdb_block_signal(int signum)
604 sigaddset(&set,signum);
605 sigprocmask(SIG_BLOCK,&set,NULL);
608 void ctdb_unblock_signal(int signum)
612 sigaddset(&set,signum);
613 sigprocmask(SIG_UNBLOCK,&set,NULL);
616 struct debug_levels debug_levels[] = {
617 {DEBUG_EMERG, "EMERG"},
618 {DEBUG_ALERT, "ALERT"},
619 {DEBUG_CRIT, "CRIT"},
621 {DEBUG_WARNING, "WARNING"},
622 {DEBUG_NOTICE, "NOTICE"},
623 {DEBUG_INFO, "INFO"},
624 {DEBUG_DEBUG, "DEBUG"},
628 const char *get_debug_by_level(int32_t level)
632 for (i=0; debug_levels[i].description != NULL; i++) {
633 if (debug_levels[i].level == level) {
634 return debug_levels[i].description;
640 int32_t get_debug_by_desc(const char *desc)
644 for (i=0; debug_levels[i].description != NULL; i++) {
645 if (!strcmp(debug_levels[i].description, desc)) {
646 return debug_levels[i].level;
653 /* we don't lock future pages here; it would increase the chance that
654 * we'd fail to mmap later on. */
655 void ctdb_lockdown_memory(struct ctdb_context *ctdb)
658 /* Extra stack, please! */
660 memset(dummy, 0, sizeof(dummy));
662 if (ctdb->valgrinding) {
666 /* Avoid compiler optimizing out dummy. */
667 mlock(dummy, sizeof(dummy));
668 if (mlockall(MCL_CURRENT) != 0) {
669 DEBUG(DEBUG_WARNING,("Failed to lock memory: %s'\n",
675 const char *ctdb_eventscript_call_names[] = {