1 /**
2 * collectd - src/collectd.c
3 * Copyright (C) 2005-2007 Florian octo Forster
4 *
5 * Permission is hereby granted, free of charge, to any person obtaining a
6 * copy of this software and associated documentation files (the "Software"),
7 * to deal in the Software without restriction, including without limitation
8 * the rights to use, copy, modify, merge, publish, distribute, sublicense,
9 * and/or sell copies of the Software, and to permit persons to whom the
10 * Software is furnished to do so, subject to the following conditions:
11 *
12 * The above copyright notice and this permission notice shall be included in
13 * all copies or substantial portions of the Software.
14 *
15 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
16 * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
17 * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
18 * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
19 * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
20 * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER
21 * DEALINGS IN THE SOFTWARE.
22 *
23 * Authors:
24 * Florian octo Forster <octo at collectd.org>
25 * Alvaro Barcellos <alvaro.barcellos at gmail.com>
26 **/
28 #include "collectd.h"
29 #include "common.h"
31 #include "plugin.h"
32 #include "configfile.h"
34 #include <sys/types.h>
35 #include <sys/socket.h>
36 #include <sys/un.h>
37 #include <netdb.h>
39 #include <pthread.h>
41 #if HAVE_LOCALE_H
42 # include <locale.h>
43 #endif
45 #if HAVE_STATGRAB_H
46 # include <statgrab.h>
47 #endif
49 #ifndef COLLECTD_LOCALE
50 # define COLLECTD_LOCALE "C"
51 #endif
53 /*
54 * Global variables
55 */
56 char hostname_g[DATA_MAX_NAME_LEN];
57 cdtime_t interval_g;
58 int pidfile_from_cli = 0;
59 int timeout_g;
60 #if HAVE_LIBKSTAT
61 kstat_ctl_t *kc;
62 #endif /* HAVE_LIBKSTAT */
64 static int loop = 0;
66 static void *do_flush (void __attribute__((unused)) *arg)
67 {
68 INFO ("Flushing all data.");
69 plugin_flush (/* plugin = */ NULL,
70 /* timeout = */ 0,
71 /* ident = */ NULL);
72 INFO ("Finished flushing all data.");
73 pthread_exit (NULL);
74 return NULL;
75 }
77 static void sig_int_handler (int __attribute__((unused)) signal)
78 {
79 loop++;
80 }
82 static void sig_term_handler (int __attribute__((unused)) signal)
83 {
84 loop++;
85 }
87 static void sig_usr1_handler (int __attribute__((unused)) signal)
88 {
89 pthread_t thread;
90 pthread_attr_t attr;
92 /* flushing the data might take a while,
93 * so it should be done asynchronously */
94 pthread_attr_init (&attr);
95 pthread_attr_setdetachstate (&attr, PTHREAD_CREATE_DETACHED);
96 pthread_create (&thread, &attr, do_flush, NULL);
97 pthread_attr_destroy (&attr);
98 }
100 static int init_hostname (void)
101 {
102 const char *str;
104 struct addrinfo ai_hints;
105 struct addrinfo *ai_list;
106 struct addrinfo *ai_ptr;
107 int status;
109 str = global_option_get ("Hostname");
110 if (str != NULL)
111 {
112 sstrncpy (hostname_g, str, sizeof (hostname_g));
113 return (0);
114 }
116 if (gethostname (hostname_g, sizeof (hostname_g)) != 0)
117 {
118 fprintf (stderr, "`gethostname' failed and no "
119 "hostname was configured.\n");
120 return (-1);
121 }
123 str = global_option_get ("FQDNLookup");
124 if (IS_FALSE (str))
125 return (0);
127 memset (&ai_hints, '\0', sizeof (ai_hints));
128 ai_hints.ai_flags = AI_CANONNAME;
130 status = getaddrinfo (hostname_g, NULL, &ai_hints, &ai_list);
131 if (status != 0)
132 {
133 ERROR ("Looking up \"%s\" failed. You have set the "
134 "\"FQDNLookup\" option, but I cannot resolve "
135 "my hostname to a fully qualified domain "
136 "name. Please fix the network "
137 "configuration.", hostname_g);
138 return (-1);
139 }
141 for (ai_ptr = ai_list; ai_ptr != NULL; ai_ptr = ai_ptr->ai_next)
142 {
143 if (ai_ptr->ai_canonname == NULL)
144 continue;
146 sstrncpy (hostname_g, ai_ptr->ai_canonname, sizeof (hostname_g));
147 break;
148 }
150 freeaddrinfo (ai_list);
151 return (0);
152 } /* int init_hostname */
154 static int init_global_variables (void)
155 {
156 char const *str;
158 interval_g = cf_get_default_interval ();
159 assert (interval_g > 0);
160 DEBUG ("interval_g = %.3f;", CDTIME_T_TO_DOUBLE (interval_g));
162 str = global_option_get ("Timeout");
163 if (str == NULL)
164 str = "2";
165 timeout_g = atoi (str);
166 if (timeout_g <= 1)
167 {
168 fprintf (stderr, "Cannot set the timeout to a correct value.\n"
169 "Please check your settings.\n");
170 return (-1);
171 }
172 DEBUG ("timeout_g = %i;", timeout_g);
174 if (init_hostname () != 0)
175 return (-1);
176 DEBUG ("hostname_g = %s;", hostname_g);
178 return (0);
179 } /* int init_global_variables */
181 static int change_basedir (const char *orig_dir)
182 {
183 char *dir;
184 size_t dirlen;
185 int status;
187 dir = strdup (orig_dir);
188 if (dir == NULL)
189 {
190 char errbuf[1024];
191 ERROR ("strdup failed: %s",
192 sstrerror (errno, errbuf, sizeof (errbuf)));
193 return (-1);
194 }
196 dirlen = strlen (dir);
197 while ((dirlen > 0) && (dir[dirlen - 1] == '/'))
198 dir[--dirlen] = '\0';
200 if (dirlen <= 0)
201 return (-1);
203 status = chdir (dir);
204 if (status == 0)
205 {
206 free (dir);
207 return (0);
208 }
209 else if (errno != ENOENT)
210 {
211 char errbuf[1024];
212 ERROR ("change_basedir: chdir (%s): %s", dir,
213 sstrerror (errno, errbuf, sizeof (errbuf)));
214 free (dir);
215 return (-1);
216 }
218 status = mkdir (dir, S_IRWXU | S_IRWXG | S_IRWXO);
219 if (status != 0)
220 {
221 char errbuf[1024];
222 ERROR ("change_basedir: mkdir (%s): %s", dir,
223 sstrerror (errno, errbuf, sizeof (errbuf)));
224 free (dir);
225 return (-1);
226 }
228 status = chdir (dir);
229 if (status != 0)
230 {
231 char errbuf[1024];
232 ERROR ("change_basedir: chdir (%s): %s", dir,
233 sstrerror (errno, errbuf, sizeof (errbuf)));
234 free (dir);
235 return (-1);
236 }
238 free (dir);
239 return (0);
240 } /* static int change_basedir (char *dir) */
242 #if HAVE_LIBKSTAT
243 static void update_kstat (void)
244 {
245 if (kc == NULL)
246 {
247 if ((kc = kstat_open ()) == NULL)
248 ERROR ("Unable to open kstat control structure");
249 }
250 else
251 {
252 kid_t kid;
253 kid = kstat_chain_update (kc);
254 if (kid > 0)
255 {
256 INFO ("kstat chain has been updated");
257 plugin_init_all ();
258 }
259 else if (kid < 0)
260 ERROR ("kstat chain update failed");
261 /* else: everything works as expected */
262 }
264 return;
265 } /* static void update_kstat (void) */
266 #endif /* HAVE_LIBKSTAT */
268 /* TODO
269 * Remove all settings but `-f' and `-C'
270 */
271 static void exit_usage (int status)
272 {
273 printf ("Usage: "PACKAGE_NAME" [OPTIONS]\n\n"
275 "Available options:\n"
276 " General:\n"
277 " -C <file> Configuration file.\n"
278 " Default: "CONFIGFILE"\n"
279 " -t Test config and exit.\n"
280 " -T Test plugin read and exit.\n"
281 " -P <file> PID-file.\n"
282 " Default: "PIDFILE"\n"
283 #if COLLECT_DAEMON
284 " -f Don't fork to the background.\n"
285 #endif
286 " -h Display help (this message)\n"
287 "\nBuiltin defaults:\n"
288 " Config file "CONFIGFILE"\n"
289 " PID file "PIDFILE"\n"
290 " Plugin directory "PLUGINDIR"\n"
291 " Data directory "PKGLOCALSTATEDIR"\n"
292 "\n"PACKAGE_NAME" "PACKAGE_VERSION", http://collectd.org/\n"
293 "by Florian octo Forster <octo@collectd.org>\n"
294 "for contributions see `AUTHORS'\n");
295 exit (status);
296 } /* static void exit_usage (int status) */
298 static int do_init (void)
299 {
300 #if HAVE_SETLOCALE
301 if (setlocale (LC_NUMERIC, COLLECTD_LOCALE) == NULL)
302 WARNING ("setlocale (\"%s\") failed.", COLLECTD_LOCALE);
303 #endif
305 #if HAVE_LIBKSTAT
306 kc = NULL;
307 update_kstat ();
308 #endif
310 #if HAVE_LIBSTATGRAB
311 if (sg_init (
312 # if HAVE_LIBSTATGRAB_0_90
313 0
314 # endif
315 ))
316 {
317 ERROR ("sg_init: %s", sg_str_error (sg_get_error ()));
318 return (-1);
319 }
321 if (sg_drop_privileges ())
322 {
323 ERROR ("sg_drop_privileges: %s", sg_str_error (sg_get_error ()));
324 return (-1);
325 }
326 #endif
328 plugin_init_all ();
330 return (0);
331 } /* int do_init () */
334 static int do_loop (void)
335 {
336 cdtime_t interval = cf_get_default_interval ();
337 cdtime_t wait_until;
339 wait_until = cdtime () + interval;
341 while (loop == 0)
342 {
343 struct timespec ts_wait = { 0, 0 };
344 cdtime_t now;
346 #if HAVE_LIBKSTAT
347 update_kstat ();
348 #endif
350 /* Issue all plugins */
351 plugin_read_all ();
353 now = cdtime ();
354 if (now >= wait_until)
355 {
356 WARNING ("Not sleeping because the next interval is "
357 "%.3f seconds in the past!",
358 CDTIME_T_TO_DOUBLE (now - wait_until));
359 wait_until = now + interval;
360 continue;
361 }
363 CDTIME_T_TO_TIMESPEC (wait_until - now, &ts_wait);
364 wait_until = wait_until + interval;
366 while ((loop == 0) && (nanosleep (&ts_wait, &ts_wait) != 0))
367 {
368 if (errno != EINTR)
369 {
370 char errbuf[1024];
371 ERROR ("nanosleep failed: %s",
372 sstrerror (errno, errbuf,
373 sizeof (errbuf)));
374 return (-1);
375 }
376 }
377 } /* while (loop == 0) */
379 return (0);
380 } /* int do_loop */
382 static int do_shutdown (void)
383 {
384 plugin_shutdown_all ();
385 return (0);
386 } /* int do_shutdown */
388 #if COLLECT_DAEMON
389 static int pidfile_create (void)
390 {
391 FILE *fh;
392 const char *file = global_option_get ("PIDFile");
394 if ((fh = fopen (file, "w")) == NULL)
395 {
396 char errbuf[1024];
397 ERROR ("fopen (%s): %s", file,
398 sstrerror (errno, errbuf, sizeof (errbuf)));
399 return (1);
400 }
402 fprintf (fh, "%i\n", (int) getpid ());
403 fclose(fh);
405 return (0);
406 } /* static int pidfile_create (const char *file) */
408 static int pidfile_remove (void)
409 {
410 const char *file = global_option_get ("PIDFile");
411 if (file == NULL)
412 return 0;
414 return (unlink (file));
415 } /* static int pidfile_remove (const char *file) */
416 #endif /* COLLECT_DAEMON */
418 #ifdef KERNEL_LINUX
419 int notify_upstart (void)
420 {
421 char const *upstart_job = getenv("UPSTART_JOB");
423 if (upstart_job == NULL)
424 return 0;
426 if (strcmp(upstart_job, "collectd") != 0)
427 {
428 WARNING ("Environment specifies unexpected UPSTART_JOB=\"%s\", expected \"collectd\". Ignoring the variable.", upstart_job);
429 return 0;
430 }
432 NOTICE("Upstart detected, stopping now to signal readyness.");
433 raise(SIGSTOP);
434 unsetenv("UPSTART_JOB");
436 return 1;
437 }
439 int notify_systemd (void)
440 {
441 int fd;
442 const char *notifysocket;
443 struct sockaddr_un su;
444 size_t su_size;
445 char buffer[] = "READY=1\n";
447 notifysocket = getenv ("NOTIFY_SOCKET");
448 if (notifysocket == NULL)
449 return 0;
451 if ((strlen (notifysocket) < 2)
452 || ((notifysocket[0] != '@') && (notifysocket[0] != '/')))
453 {
454 ERROR ("invalid notification socket NOTIFY_SOCKET=\"%s\": path must be absolute", notifysocket);
455 return 0;
456 }
457 NOTICE ("Systemd detected, trying to signal readyness.");
459 unsetenv ("NOTIFY_SOCKET");
461 fd = socket (AF_UNIX, SOCK_DGRAM, /* protocol = */ 0);
462 if (fd < 0) {
463 char errbuf[1024];
464 ERROR ("creating UNIX socket failed: %s",
465 sstrerror (errno, errbuf, sizeof (errbuf)));
466 return 0;
467 }
469 memset (&su, 0, sizeof (su));
470 su.sun_family = AF_UNIX;
471 if (notifysocket[0] != '@')
472 {
473 /* regular UNIX socket */
474 sstrncpy (su.sun_path, notifysocket, sizeof (su.sun_path));
475 su_size = sizeof (su);
476 }
477 else
478 {
479 #if KERNEL_LINUX
480 /* Linux abstract namespace socket: specify address as "\0foo", i.e.
481 * start with a null byte. Since null bytes have no special meaning in
482 * that case, we have to set su_size correctly to cover only the bytes
483 * that are part of the address. */
484 sstrncpy (su.sun_path, notifysocket, sizeof (su.sun_path));
485 su.sun_path[0] = 0;
486 su_size = sizeof (sa_family_t) + strlen (notifysocket);
487 if (su_size > sizeof (su))
488 su_size = sizeof (su);
489 #else
490 ERROR ("Systemd socket uses Linux abstract namespace notation (\"%s\"), "
491 "but I don't appear to be running on Linux.", notifysocket);
492 return 0;
493 #endif
494 }
496 if (sendto (fd, buffer, strlen (buffer), MSG_NOSIGNAL, (void *) &su, (socklen_t) su_size) < 0)
497 {
498 char errbuf[1024];
499 ERROR ("sendto(\"%s\") failed: %s", notifysocket,
500 sstrerror (errno, errbuf, sizeof (errbuf)));
501 close(fd);
502 return 0;
503 }
505 close(fd);
506 return 1;
507 }
508 #endif /* KERNEL_LINUX */
510 int main (int argc, char **argv)
511 {
512 struct sigaction sig_int_action;
513 struct sigaction sig_term_action;
514 struct sigaction sig_usr1_action;
515 struct sigaction sig_pipe_action;
516 char *configfile = CONFIGFILE;
517 int test_config = 0;
518 int test_readall = 0;
519 const char *basedir;
520 #if COLLECT_DAEMON
521 struct sigaction sig_chld_action;
522 pid_t pid;
523 int daemonize = 1;
524 #endif
525 int exit_status = 0;
527 /* read options */
528 while (1)
529 {
530 int c;
532 c = getopt (argc, argv, "htTC:"
533 #if COLLECT_DAEMON
534 "fP:"
535 #endif
536 );
538 if (c == -1)
539 break;
541 switch (c)
542 {
543 case 'C':
544 configfile = optarg;
545 break;
546 case 't':
547 test_config = 1;
548 break;
549 case 'T':
550 test_readall = 1;
551 global_option_set ("ReadThreads", "-1");
552 #if COLLECT_DAEMON
553 daemonize = 0;
554 #endif /* COLLECT_DAEMON */
555 break;
556 #if COLLECT_DAEMON
557 case 'P':
558 global_option_set ("PIDFile", optarg);
559 pidfile_from_cli = 1;
560 break;
561 case 'f':
562 daemonize = 0;
563 break;
564 #endif /* COLLECT_DAEMON */
565 case 'h':
566 exit_usage (0);
567 break;
568 default:
569 exit_usage (1);
570 } /* switch (c) */
571 } /* while (1) */
573 if (optind < argc)
574 exit_usage (1);
576 plugin_init_ctx ();
578 /*
579 * Read options from the config file, the environment and the command
580 * line (in that order, with later options overwriting previous ones in
581 * general).
582 * Also, this will automatically load modules.
583 */
584 if (cf_read (configfile))
585 {
586 fprintf (stderr, "Error: Reading the config file failed!\n"
587 "Read the syslog for details.\n");
588 return (1);
589 }
591 /*
592 * Change directory. We do this _after_ reading the config and loading
593 * modules to relative paths work as expected.
594 */
595 if ((basedir = global_option_get ("BaseDir")) == NULL)
596 {
597 fprintf (stderr, "Don't have a basedir to use. This should not happen. Ever.");
598 return (1);
599 }
600 else if (change_basedir (basedir))
601 {
602 fprintf (stderr, "Error: Unable to change to directory `%s'.\n", basedir);
603 return (1);
604 }
606 /*
607 * Set global variables or, if that failes, exit. We cannot run with
608 * them being uninitialized. If nothing is configured, then defaults
609 * are being used. So this means that the user has actually done
610 * something wrong.
611 */
612 if (init_global_variables () != 0)
613 return (1);
615 if (test_config)
616 return (0);
618 #if COLLECT_DAEMON
619 /*
620 * fork off child
621 */
622 memset (&sig_chld_action, '\0', sizeof (sig_chld_action));
623 sig_chld_action.sa_handler = SIG_IGN;
624 sigaction (SIGCHLD, &sig_chld_action, NULL);
626 /*
627 * Only daemonize if we're not being supervised
628 * by upstart or systemd (when using Linux).
629 */
630 if (daemonize
631 #ifdef KERNEL_LINUX
632 && notify_upstart() == 0 && notify_systemd() == 0
633 #endif
634 )
635 {
636 if ((pid = fork ()) == -1)
637 {
638 /* error */
639 char errbuf[1024];
640 fprintf (stderr, "fork: %s",
641 sstrerror (errno, errbuf,
642 sizeof (errbuf)));
643 return (1);
644 }
645 else if (pid != 0)
646 {
647 /* parent */
648 /* printf ("Running (PID %i)\n", pid); */
649 return (0);
650 }
652 /* Detach from session */
653 setsid ();
655 /* Write pidfile */
656 if (pidfile_create ())
657 exit (2);
659 /* close standard descriptors */
660 close (2);
661 close (1);
662 close (0);
664 if (open ("/dev/null", O_RDWR) != 0)
665 {
666 ERROR ("Error: Could not connect `STDIN' to `/dev/null'");
667 return (1);
668 }
669 if (dup (0) != 1)
670 {
671 ERROR ("Error: Could not connect `STDOUT' to `/dev/null'");
672 return (1);
673 }
674 if (dup (0) != 2)
675 {
676 ERROR ("Error: Could not connect `STDERR' to `/dev/null'");
677 return (1);
678 }
679 } /* if (daemonize) */
680 #endif /* COLLECT_DAEMON */
682 memset (&sig_pipe_action, '\0', sizeof (sig_pipe_action));
683 sig_pipe_action.sa_handler = SIG_IGN;
684 sigaction (SIGPIPE, &sig_pipe_action, NULL);
686 /*
687 * install signal handlers
688 */
689 memset (&sig_int_action, '\0', sizeof (sig_int_action));
690 sig_int_action.sa_handler = sig_int_handler;
691 if (0 != sigaction (SIGINT, &sig_int_action, NULL)) {
692 char errbuf[1024];
693 ERROR ("Error: Failed to install a signal handler for signal INT: %s",
694 sstrerror (errno, errbuf, sizeof (errbuf)));
695 return (1);
696 }
698 memset (&sig_term_action, '\0', sizeof (sig_term_action));
699 sig_term_action.sa_handler = sig_term_handler;
700 if (0 != sigaction (SIGTERM, &sig_term_action, NULL)) {
701 char errbuf[1024];
702 ERROR ("Error: Failed to install a signal handler for signal TERM: %s",
703 sstrerror (errno, errbuf, sizeof (errbuf)));
704 return (1);
705 }
707 memset (&sig_usr1_action, '\0', sizeof (sig_usr1_action));
708 sig_usr1_action.sa_handler = sig_usr1_handler;
709 if (0 != sigaction (SIGUSR1, &sig_usr1_action, NULL)) {
710 char errbuf[1024];
711 ERROR ("Error: Failed to install a signal handler for signal USR1: %s",
712 sstrerror (errno, errbuf, sizeof (errbuf)));
713 return (1);
714 }
716 /*
717 * run the actual loops
718 */
719 do_init ();
721 if (test_readall)
722 {
723 if (plugin_read_all_once () != 0)
724 exit_status = 1;
725 }
726 else
727 {
728 INFO ("Initialization complete, entering read-loop.");
729 do_loop ();
730 }
732 /* close syslog */
733 INFO ("Exiting normally.");
735 do_shutdown ();
737 #if COLLECT_DAEMON
738 if (daemonize)
739 pidfile_remove ();
740 #endif /* COLLECT_DAEMON */
742 return (exit_status);
743 } /* int main */