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);
77 address->port = CTDB_PORT;
79 address->port = ntohs(se->s_port);
86 check if two addresses are the same
88 bool ctdb_same_address(struct ctdb_address *a1, struct ctdb_address *a2)
90 return strcmp(a1->address, a2->address) == 0 && a1->port == a2->port;
95 hash function for mapping data to a VNN - taken from tdb
97 uint32_t ctdb_hash(const TDB_DATA *key)
99 uint32_t value; /* Used to compute the hash value. */
100 uint32_t i; /* Used to cycle through random values. */
102 /* Set the initial value from the key size. */
103 for (value = 0x238F13AF * key->dsize, i=0; i < key->dsize; i++)
104 value = (value + (key->dptr[i] << (i*5 % 24)));
106 return (1103515243 * value + 12345);
110 a type checking varient of idr_find
112 static void *_idr_find_type(struct idr_context *idp, int id, const char *type, const char *location)
114 void *p = idr_find(idp, id);
115 if (p && talloc_check_name(p, type) == NULL) {
116 DEBUG(DEBUG_ERR,("%s idr_find_type expected type %s but got %s\n",
117 location, type, talloc_get_name(p)));
125 update a max latency number
127 void ctdb_latency(struct ctdb_db_context *ctdb_db, const char *name, double *latency, struct timeval t)
129 double l = timeval_elapsed(&t);
134 if (ctdb_db->ctdb->tunable.log_latency_ms !=0) {
135 if (l*1000 > ctdb_db->ctdb->tunable.log_latency_ms) {
136 DEBUG(DEBUG_WARNING, ("High latency %fs for operation %s on database %s\n", l*1000000, name, ctdb_db->db_name));
141 uint32_t ctdb_reqid_new(struct ctdb_context *ctdb, void *state)
145 id = ctdb->idr_cnt++ & 0xFFFF;
146 id |= (idr_get_new(ctdb->idr, state, 0xFFFF)<<16);
150 void *_ctdb_reqid_find(struct ctdb_context *ctdb, uint32_t reqid, const char *type, const char *location)
154 p = _idr_find_type(ctdb->idr, (reqid>>16)&0xFFFF, type, location);
156 DEBUG(DEBUG_WARNING, ("Could not find idr:%u\n",reqid));
163 void ctdb_reqid_remove(struct ctdb_context *ctdb, uint32_t reqid)
167 ret = idr_remove(ctdb->idr, (reqid>>16)&0xFFFF);
169 DEBUG(DEBUG_ERR, ("Removing idr that does not exist\n"));
175 form a ctdb_rec_data record from a key/data pair
177 note that header may be NULL. If not NULL then it is included in the data portion
180 struct ctdb_rec_data *ctdb_marshall_record(TALLOC_CTX *mem_ctx, uint32_t reqid,
182 struct ctdb_ltdb_header *header,
186 struct ctdb_rec_data *d;
188 length = offsetof(struct ctdb_rec_data, data) + key.dsize +
189 data.dsize + (header?sizeof(*header):0);
190 d = (struct ctdb_rec_data *)talloc_size(mem_ctx, length);
196 d->keylen = key.dsize;
197 memcpy(&d->data[0], key.dptr, key.dsize);
199 d->datalen = data.dsize + sizeof(*header);
200 memcpy(&d->data[key.dsize], header, sizeof(*header));
201 memcpy(&d->data[key.dsize+sizeof(*header)], data.dptr, data.dsize);
203 d->datalen = data.dsize;
204 memcpy(&d->data[key.dsize], data.dptr, data.dsize);
210 /* helper function for marshalling multiple records */
211 struct ctdb_marshall_buffer *ctdb_marshall_add(TALLOC_CTX *mem_ctx,
212 struct ctdb_marshall_buffer *m,
216 struct ctdb_ltdb_header *header,
219 struct ctdb_rec_data *r;
220 size_t m_size, r_size;
221 struct ctdb_marshall_buffer *m2;
223 r = ctdb_marshall_record(mem_ctx, reqid, key, header, data);
230 m = talloc_zero_size(mem_ctx, offsetof(struct ctdb_marshall_buffer, data));
237 m_size = talloc_get_size(m);
238 r_size = talloc_get_size(r);
240 m2 = talloc_realloc_size(mem_ctx, m, m_size + r_size);
246 memcpy(m_size + (uint8_t *)m2, r, r_size);
255 /* we've finished marshalling, return a data blob with the marshalled records */
256 TDB_DATA ctdb_marshall_finish(struct ctdb_marshall_buffer *m)
259 data.dptr = (uint8_t *)m;
260 data.dsize = talloc_get_size(m);
265 loop over a marshalling buffer
267 - pass r==NULL to start
268 - loop the number of times indicated by m->count
270 struct ctdb_rec_data *ctdb_marshall_loop_next(struct ctdb_marshall_buffer *m, struct ctdb_rec_data *r,
272 struct ctdb_ltdb_header *header,
273 TDB_DATA *key, TDB_DATA *data)
276 r = (struct ctdb_rec_data *)&m->data[0];
278 r = (struct ctdb_rec_data *)(r->length + (uint8_t *)r);
286 key->dptr = &r->data[0];
287 key->dsize = r->keylen;
290 data->dptr = &r->data[r->keylen];
291 data->dsize = r->datalen;
292 if (header != NULL) {
293 data->dptr += sizeof(*header);
294 data->dsize -= sizeof(*header);
298 if (header != NULL) {
299 if (r->datalen < sizeof(*header)) {
302 *header = *(struct ctdb_ltdb_header *)&r->data[r->keylen];
314 if possible, make this task real time
316 void ctdb_set_scheduler(struct ctdb_context *ctdb)
318 #if HAVE_SCHED_SETSCHEDULER
319 struct sched_param p;
320 if (ctdb->saved_scheduler_param == NULL) {
321 ctdb->saved_scheduler_param = talloc_size(ctdb, sizeof(p));
324 if (sched_getparam(0, (struct sched_param *)ctdb->saved_scheduler_param) == -1) {
325 DEBUG(DEBUG_ERR,("Unable to get old scheduler params\n"));
329 p = *(struct sched_param *)ctdb->saved_scheduler_param;
330 p.sched_priority = 1;
332 if (sched_setscheduler(0, SCHED_FIFO, &p) == -1) {
333 DEBUG(DEBUG_CRIT,("Unable to set scheduler to SCHED_FIFO (%s)\n",
336 DEBUG(DEBUG_NOTICE,("Set scheduler to SCHED_FIFO\n"));
342 restore previous scheduler parameters
344 void ctdb_restore_scheduler(struct ctdb_context *ctdb)
346 #if HAVE_SCHED_SETSCHEDULER
347 if (ctdb->saved_scheduler_param == NULL) {
348 ctdb_fatal(ctdb, "No saved scheduler parameters\n");
350 if (sched_setscheduler(0, SCHED_OTHER, (struct sched_param *)ctdb->saved_scheduler_param) == -1) {
351 ctdb_fatal(ctdb, "Unable to restore old scheduler parameters\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 *iface, 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 (iface && IN6_IS_ADDR_LINKLOCAL(&saddr->ip6.sin6_addr)) {
397 saddr->ip6.sin6_scope_id = if_nametoindex(iface);
405 bool parse_ip_port(const char *addr, ctdb_sock_addr *saddr)
407 TALLOC_CTX *tmp_ctx = talloc_new(NULL);
413 s = talloc_strdup(tmp_ctx, addr);
415 DEBUG(DEBUG_ERR, (__location__ " Failed strdup()\n"));
416 talloc_free(tmp_ctx);
422 DEBUG(DEBUG_ERR, (__location__ " This addr: %s does not contain a port number\n", s));
423 talloc_free(tmp_ctx);
427 port = strtoul(p+1, &endp, 10);
428 if (endp == NULL || *endp != 0) {
429 /* trailing garbage */
430 DEBUG(DEBUG_ERR, (__location__ " Trailing garbage after the port in %s\n", s));
431 talloc_free(tmp_ctx);
437 /* now is this a ipv4 or ipv6 address ?*/
438 ret = parse_ip(s, NULL, port, saddr);
440 talloc_free(tmp_ctx);
447 bool parse_ip(const char *addr, const char *iface, unsigned port, ctdb_sock_addr *saddr)
452 /* now is this a ipv4 or ipv6 address ?*/
453 p = index(addr, ':');
455 ret = parse_ipv4(addr, port, &saddr->ip);
457 ret = parse_ipv6(addr, iface, port, saddr);
466 bool parse_ip_mask(const char *str, const char *iface, ctdb_sock_addr *addr, unsigned *mask)
468 TALLOC_CTX *tmp_ctx = talloc_new(NULL);
474 s = talloc_strdup(tmp_ctx, str);
476 DEBUG(DEBUG_ERR, (__location__ " Failed strdup()\n"));
477 talloc_free(tmp_ctx);
483 DEBUG(DEBUG_ERR, (__location__ " This addr: %s does not contain a mask\n", s));
484 talloc_free(tmp_ctx);
488 *mask = strtoul(p+1, &endp, 10);
489 if (endp == NULL || *endp != 0) {
490 /* trailing garbage */
491 DEBUG(DEBUG_ERR, (__location__ " Trailing garbage after the mask in %s\n", s));
492 talloc_free(tmp_ctx);
498 /* now is this a ipv4 or ipv6 address ?*/
499 ret = parse_ip(s, iface, 0, addr);
501 talloc_free(tmp_ctx);
506 This is used to canonicalize a ctdb_sock_addr structure.
508 void ctdb_canonicalize_ip(const ctdb_sock_addr *ip, ctdb_sock_addr *cip)
510 char prefix[12] = { 0,0,0,0,0,0,0,0,0,0,0xff,0xff };
512 memcpy(cip, ip, sizeof (*cip));
514 if ( (ip->sa.sa_family == AF_INET6)
515 && !memcmp(&ip->ip6.sin6_addr, prefix, 12)) {
516 memset(cip, 0, sizeof(*cip));
517 #ifdef HAVE_SOCK_SIN_LEN
518 cip->ip.sin_len = sizeof(*cip);
520 cip->ip.sin_family = AF_INET;
521 cip->ip.sin_port = ip->ip6.sin6_port;
522 memcpy(&cip->ip.sin_addr, &ip->ip6.sin6_addr.s6_addr32[3], 4);
526 bool ctdb_same_ip(const ctdb_sock_addr *tip1, const ctdb_sock_addr *tip2)
528 ctdb_sock_addr ip1, ip2;
530 ctdb_canonicalize_ip(tip1, &ip1);
531 ctdb_canonicalize_ip(tip2, &ip2);
533 if (ip1.sa.sa_family != ip2.sa.sa_family) {
537 switch (ip1.sa.sa_family) {
539 return ip1.ip.sin_addr.s_addr == ip2.ip.sin_addr.s_addr;
541 return !memcmp(&ip1.ip6.sin6_addr.s6_addr[0],
542 &ip2.ip6.sin6_addr.s6_addr[0],
545 DEBUG(DEBUG_ERR, (__location__ " CRITICAL Can not compare sockaddr structures of type %u\n", ip1.sa.sa_family));
553 compare two ctdb_sock_addr structures
555 bool ctdb_same_sockaddr(const ctdb_sock_addr *ip1, const ctdb_sock_addr *ip2)
557 return ctdb_same_ip(ip1, ip2) && ip1->ip.sin_port == ip2->ip.sin_port;
560 char *ctdb_addr_to_str(ctdb_sock_addr *addr)
562 static char cip[128] = "";
564 switch (addr->sa.sa_family) {
566 inet_ntop(addr->ip.sin_family, &addr->ip.sin_addr, cip, sizeof(cip));
569 inet_ntop(addr->ip6.sin6_family, &addr->ip6.sin6_addr, cip, sizeof(cip));
572 DEBUG(DEBUG_ERR, (__location__ " ERROR, unknown family %u\n", addr->sa.sa_family));
578 unsigned ctdb_addr_to_port(ctdb_sock_addr *addr)
580 switch (addr->sa.sa_family) {
582 return ntohs(addr->ip.sin_port);
585 return ntohs(addr->ip6.sin6_port);
588 DEBUG(DEBUG_ERR, (__location__ " ERROR, unknown family %u\n", addr->sa.sa_family));
594 void ctdb_block_signal(int signum)
598 sigaddset(&set,signum);
599 sigprocmask(SIG_BLOCK,&set,NULL);
602 void ctdb_unblock_signal(int signum)
606 sigaddset(&set,signum);
607 sigprocmask(SIG_UNBLOCK,&set,NULL);