summaryrefslogtreecommitdiffstats
diff options
context:
space:
mode:
Diffstat
-rw-r--r--plugins/check_icmp.c1381
1 files changed, 1381 insertions, 0 deletions
diff --git a/plugins/check_icmp.c b/plugins/check_icmp.c
new file mode 100644
index 0000000..45bfcc5
--- /dev/null
+++ b/plugins/check_icmp.c
@@ -0,0 +1,1381 @@
1/*
2 * check_icmp - A hack of fping2 to work with nagios.
3 * This way we don't have to use the output parser.
4 *
5 * VIEWING NOTES:
6 * This file was formatted with tab indents at a tab stop of 4.
7 *
8 * It is highly recommended that your editor is set to this
9 * tab stop setting for viewing and editing.
10 *
11 * COPYLEFT;
12 * This programs copyright status is currently undetermined. Much of
13 * the code in it comes from the fping2 program which used to be licensed
14 * under the Stanford General Software License (available at
15 * http://graphics.stanford.edu/software/license.html). It is presently
16 * unclear what license (if any) applies to the original code at the
17 * moment.
18 *
19 * The fping website can be found at http://www.fping.com
20 */
21
22#include <stdio.h>
23#include <errno.h>
24#include <time.h>
25#include <signal.h>
26
27#include <unistd.h>
28
29#include <stdlib.h>
30
31#include <string.h>
32#include <stddef.h>
33
34#include <sys/types.h>
35#include <sys/time.h>
36#include <sys/socket.h>
37
38#include <sys/file.h>
39
40#include <netinet/in_systm.h>
41#include <netinet/in.h>
42
43/* Linux has bizarre ip.h and ip_icmp.h */
44/* Taken from the fping distro. Thank you. */
45#if defined( __linux__ )
46#include "linux.h"
47#else
48#include <netinet/ip.h>
49#include <netinet/ip_icmp.h>
50#endif /* defined(__linux__) */
51
52#include <arpa/inet.h>
53#include <netdb.h>
54
55/* RS6000 has sys/select.h */
56#ifdef HAVE_SYS_SELECT_H
57#include <sys/select.h>
58#endif /* HAVE_SYS_SELECT_H */
59
60/* rta threshold values can't be larger than MAXTTL seconds */
61#ifndef MAXTTL
62# define MAXTTL 255
63#endif
64#ifndef IPDEFTTL
65# define IPDEFTTL 64
66#endif
67
68/*** externals ***/
69extern char *optarg;
70extern int optind, opterr;
71
72/*** Constants ***/
73#define REV_DATE "2004-09-06"
74#define EMAIL "ae@op5.se"
75#define VERSION "0.8"
76
77#ifndef INADDR_NONE
78# define INADDR_NONE 0xffffffU
79#endif
80
81/*** Ping packet defines ***/
82/* data added after ICMP header for our nefarious purposes */
83typedef struct ping_data {
84 unsigned int ping_count; /* counts up to -[n|p] count or 1 */
85 struct timeval ping_ts; /* time sent */
86} PING_DATA;
87
88#define MIN_PING_DATA sizeof(PING_DATA)
89#define MAX_IP_PACKET 65536 /* (theoretical) max IP packet size */
90#define SIZE_IP_HDR 20
91#define SIZE_ICMP_HDR ICMP_MINLEN /* from ip_icmp.h */
92#define MAX_PING_DATA (MAX_IP_PACKET - SIZE_IP_HDR - SIZE_ICMP_HDR)
93
94/*
95 * Interval is the minimum amount of time between sending a ping packet to
96 * any host.
97 *
98 * Perhost_interval is the minimum amount of time between sending a ping
99 * packet to a particular responding host
100 *
101 * Timeout is the initial amount of time between sending a ping packet to
102 * a particular non-responding host.
103 *
104 * Retry is the number of ping packets to send to a non-responding host
105 * before giving up (in is-it-alive mode).
106 *
107 * Backoff factor is how much longer to wait on successive retries.
108 */
109#ifndef DEFAULT_INTERVAL
110#define DEFAULT_INTERVAL 25 /* default time between packets (msec) */
111#endif
112
113#ifndef DEFAULT_RETRY
114#define DEFAULT_RETRY 1 /* number of times to retry a host */
115#endif
116
117#ifndef DEFAULT_TIMEOUT
118# define DEFAULT_TIMEOUT 1000
119#endif
120
121#ifndef DEFAULT_BACKOFF_FACTOR
122#define DEFAULT_BACKOFF_FACTOR 1.5 /* exponential timeout factor */
123#endif
124#define MIN_BACKOFF_FACTOR 1.0 /* exponential timeout factor */
125#define MAX_BACKOFF_FACTOR 5.0 /* exponential timeout factor */
126
127#ifndef DNS_TIMEOUT
128#define DNS_TIMEOUT 1000 /* time in usec for dns retry */
129#endif
130
131#ifndef MAX_RTA_THRESHOLD_VALUE
132# define MAX_RTA_THRESHOLD_VALUE 120*1000000 /* 2 minutes should be enough */
133#endif
134#ifndef MIN_RTA_THRESHOLD_VALUE
135# define MIN_RTA_THRESHOLD_VALUE 10000 /* minimum RTA threshold value */
136#endif
137
138/* sized so as to be like traditional ping */
139#define DEFAULT_PING_DATA_SIZE (MIN_PING_DATA + 44)
140
141/* maxima and minima */
142#define MAX_COUNT 50 /* max count even if we're root */
143#define MAX_RETRY 5
144#define MIN_INTERVAL 25 /* msecs */
145#define MIN_TIMEOUT 50 /* msecs */
146
147/* response time array flags */
148#define RESP_WAITING -1
149#define RESP_UNUSED -2
150
151#define ICMP_UNREACH_MAXTYPE 15
152
153/* entry used to keep track of each host we are pinging */
154struct host_entry {
155 int i; /* index into array */
156 char *name; /* name as given by user */
157 char *host; /* text description of host */
158 struct sockaddr_in saddr; /* internet address */
159 unsigned short **pr; /* TCP port range to check for connectivity */
160 struct timeval last_send_time; /* time of last packet sent */
161 unsigned int num_sent; /* number of ping packets sent */
162 unsigned int num_recv; /* number of pings received */
163 unsigned int total_time; /* sum of response times */
164 unsigned int status; /* this hosts status */
165 unsigned int running; /* unset when through sending */
166 unsigned int waiting; /* waiting for response */
167 int *resp_times; /* individual response times */
168 struct host_entry *prev, *next; /* doubly linked list */
169};
170
171typedef struct host_entry HOST_ENTRY;
172
173struct host_name_list {
174 char *entry;
175 struct host_name_list *next;
176};
177
178/* threshold structure */
179struct threshold {
180 unsigned int pl; /* packet loss */
181 unsigned int rta; /* roundtrip time average */
182};
183typedef struct threshold threshold;
184
185/*****************************************************************************
186 * Global Variables *
187 *****************************************************************************/
188
189HOST_ENTRY *rrlist = NULL; /* linked list of hosts be pinged */
190HOST_ENTRY **table = NULL; /* array of pointers to items in the list */
191HOST_ENTRY *cursor;
192
193char *prog; /* our name */
194int ident; /* our pid, for marking icmp packets */
195int sock; /* socket */
196u_int debug = 0;
197
198/* threshold value defaults;
199 * WARNING; 60% packetloss or 200 msecs round trip average
200 * CRITICAL; 80% packetloss or 500 msecs round trip average */
201threshold warn = {60, 200 * 1000};
202threshold crit = {80, 500 * 1000};
203
204/* times get *100 because all times are calculated in 10 usec units, not ms */
205unsigned int retry = DEFAULT_RETRY;
206u_int timeout = DEFAULT_TIMEOUT * 100;
207u_int interval = DEFAULT_INTERVAL * 100;
208float backoff = DEFAULT_BACKOFF_FACTOR;
209u_int select_time; /* calculated using maximum threshold rta value */
210u_int ping_data_size = DEFAULT_PING_DATA_SIZE;
211u_int ping_pkt_size;
212unsigned int count = 5;
213unsigned int trials = 1;
214
215/* global stats */
216int total_replies = 0;
217int num_jobs = 0; /* number of hosts still to do */
218int num_hosts = 0; /* total number of hosts */
219int num_alive = 0; /* total number alive */
220int num_unreachable = 0; /* total number unreachable */
221int num_noaddress = 0; /* total number of addresses not found */
222int num_timeout = 0; /* number of timed out packets */
223int num_pingsent = 0; /* total pings sent */
224int num_pingreceived = 0; /* total pings received */
225int num_othericmprcvd = 0; /* total non-echo-reply ICMP received */
226
227struct timeval current_time; /* current time (pseudo) */
228struct timeval start_time;
229struct timeval end_time;
230struct timeval last_send_time; /* time last ping was sent */
231struct timezone tz;
232
233/* switches */
234int generate_flag = 0; /* flag for IP list generation */
235int stats_flag, unreachable_flag, alive_flag;
236int elapsed_flag, version_flag, count_flag;
237int name_flag, addr_flag, backoff_flag;
238int multif_flag;
239
240/*** prototypes ***/
241void add_name(char *);
242void add_addr(char *, char *, struct in_addr);
243char *na_cat(char *, struct in_addr);
244char *cpystr(char *);
245void crash(char *);
246char *get_host_by_address(struct in_addr);
247int in_cksum(u_short *, int);
248void u_sleep(int);
249int recvfrom_wto(int, char *, int, struct sockaddr *, int);
250void remove_job(HOST_ENTRY *);
251void send_ping(int, HOST_ENTRY *);
252long timeval_diff(struct timeval *, struct timeval *);
253void usage(void);
254int wait_for_reply(int);
255void finish(void);
256int handle_random_icmp(struct icmp *, struct sockaddr_in *);
257char *sprint_tm(int);
258int get_threshold(char *, threshold *);
259
260/*** the various exit-states */
261enum {
262 STATE_OK = 0,
263 STATE_WARNING,
264 STATE_CRITICAL,
265 STATE_UNKNOWN,
266 STATE_DEPENDANT,
267 STATE_OOB
268};
269/* the strings that correspond to them */
270char *status_string[STATE_OOB] = {
271 "OK",
272 "WARNING",
273 "CRITICAL",
274 "UNKNOWN",
275 "DEPENDANT"
276};
277
278int status = STATE_OK;
279int fin_stat = STATE_OK;
280
281/*****************************************************************************
282 * Code block start *
283 *****************************************************************************/
284int main(int argc, char **argv)
285{
286 int c;
287 u_int lt, ht;
288 int advance;
289 struct protoent *proto;
290 uid_t uid;
291 struct host_name_list *host_ptr, *host_base_ptr;
292
293 if(strchr(argv[0], '/')) prog = strrchr(argv[0], '/') + 1;
294 else prog = argv[0];
295
296 /* check if we are root */
297 if(geteuid()) {
298 printf("Root access needed (for raw sockets)\n");
299 exit(STATE_UNKNOWN);
300 }
301
302 /* confirm that ICMP is available on this machine */
303 if((proto = getprotobyname("icmp")) == NULL)
304 crash("icmp: unknown protocol");
305
306 /* create raw socket for ICMP calls (ping) */
307 sock = socket(AF_INET, SOCK_RAW, proto->p_proto);
308
309 if(sock < 0)
310 crash("can't create raw socket");
311
312 /* drop privileges now that we have the socket */
313 if((uid = getuid())) {
314 seteuid(uid);
315 }
316
317 if(argc < 2) usage();
318
319 ident = getpid() & 0xFFFF;
320
321 if(!(host_base_ptr = malloc(sizeof(struct host_name_list)))) {
322 crash("Unable to allocate memory for host name list\n");
323 }
324 host_ptr = host_base_ptr;
325
326 backoff_flag = 0;
327 opterr = 1;
328
329 /* get command line options
330 * -H denotes a host (actually ignored and picked up later)
331 * -h for help
332 * -V or -v for version
333 * -d to display hostnames rather than addresses
334 * -t sets timeout for packets and tcp connects
335 * -r defines retries (persistence)
336 * -p or -n sets packet count (5)
337 * -b sets packet size (56)
338 * -w sets warning threshhold (200,40%)
339 * -c sets critical threshhold (500,80%)
340 * -i sets interval for both packet transmissions and connect attempts
341 */
342#define OPT_STR "amH:hvVDdAp:n:b:r:t:i:w:c:"
343 while((c = getopt(argc, argv, OPT_STR)) != EOF) {
344 switch (c) {
345 case 'H':
346 if(!(host_ptr->entry = malloc(strlen(optarg) + 1))) {
347 crash("Failed to allocate memory for hostname");
348 }
349 memset(host_ptr->entry, 0, strlen(optarg) + 1);
350 host_ptr->entry = memcpy(host_ptr->entry, optarg, strlen(optarg));
351 if(!(host_ptr->next = malloc(sizeof(struct host_name_list))))
352 crash("Failed to allocate memory for hostname");
353 host_ptr = host_ptr->next;
354 host_ptr->next = NULL;
355// add_name(optarg);
356 break;
357 /* this is recognized, but silently ignored.
358 * host(s) are added later on */
359
360 break;
361 case 'w':
362 if(get_threshold(optarg, &warn)) {
363 printf("Illegal threshold pair specified for -%c", c);
364 usage();
365 }
366 break;
367
368 case 'c':
369 if(get_threshold(optarg, &crit)) {
370 printf("Illegal threshold pair specified for -%c", c);
371 usage();
372 }
373 break;
374
375 case 't':
376 if(!(timeout = (u_int) strtoul(optarg, NULL, 0) * 100)) {
377 printf("option -%c requires integer argument\n", c);
378 usage();
379 }
380 break;
381
382 case 'r':
383 if(!(retry = (u_int) strtoul(optarg, NULL, 0))) {
384 printf("option -%c requires integer argument\n", c);
385 usage();
386 }
387 break;
388
389 case 'i':
390 if(!(interval = (u_int) strtoul(optarg, NULL, 0) * 100)) {
391 printf("option -%c requires positive non-zero integer argument\n", c);
392 usage();
393 }
394 break;
395
396 case 'p':
397 case 'n':
398 if(!(count = (u_int) strtoul(optarg, NULL, 0))) {
399 printf("option -%c requires positive non-zero integer argument\n", c);
400 usage();
401 }
402 break;
403
404 case 'b':
405 if(!(ping_data_size = (u_int) strtoul(optarg, NULL, 0))) {
406 printf("option -%c requires integer argument\n", c);
407 usage();
408 }
409 break;
410
411 case 'h':
412 usage();
413 break;
414
415 case 'e':
416 elapsed_flag = 1;
417 break;
418
419 case 'm':
420 multif_flag = 1;
421 break;
422
423 case 'd':
424 name_flag = 1;
425 break;
426
427 case 'A':
428 addr_flag = 1;
429 break;
430
431 case 's':
432 stats_flag = 1;
433 break;
434
435 case 'u':
436 unreachable_flag = 1;
437 break;
438
439 case 'a':
440 alive_flag = 1;
441 break;
442
443 case 'v':
444 printf("%s: Version %s $Date$\n", prog, VERSION, REV_DATE);
445 printf("%s: comments to %s\n", prog, EMAIL);
446 exit(STATE_OK);
447
448 case 'g':
449 /* use IP list generation */
450 /* mutex with file input or command line targets */
451 generate_flag = 1;
452 break;
453
454 default:
455 printf("option flag -%c specified, but not recognized\n", c);
456 usage();
457 break;
458 }
459 }
460
461 /* arguments are parsed, so now we validate them */
462
463 if(count > 1) count_flag = 1;
464
465 /* set threshold values to 10usec units (inherited from fping.c) */
466 crit.rta = crit.rta / 10;
467 warn.rta = warn.rta / 10;
468 select_time = crit.rta;
469 /* this isn't critical, but will most likely not be what the user expects
470 * so we tell him/her about it, but keep running anyways */
471 if(warn.pl > crit.pl || warn.rta > crit.rta) {
472 select_time = warn.rta;
473 printf("(WARNING threshold > CRITICAL threshold) :: ");
474 fflush(stdout);
475 }
476
477 /* A timeout smaller than maximum rta threshold makes no sense */
478 if(timeout < crit.rta) timeout = crit.rta;
479 else if(timeout < warn.rta) timeout = warn.rta;
480
481 if((interval < MIN_INTERVAL * 100 || retry > MAX_RETRY) && getuid()) {
482 printf("%s: these options are too risky for mere mortals.\n", prog);
483 printf("%s: You need i >= %u and r < %u\n",
484 prog, MIN_INTERVAL, MAX_RETRY);
485 printf("Current settings; i = %d, r = %d\n",
486 interval / 100, retry);
487 usage();
488 }
489
490 if((ping_data_size > MAX_PING_DATA) || (ping_data_size < MIN_PING_DATA)) {
491 printf("%s: data size %u not valid, must be between %u and %u\n",
492 prog, ping_data_size, MIN_PING_DATA, MAX_PING_DATA);
493 usage();
494
495 }
496
497 if((backoff > MAX_BACKOFF_FACTOR) || (backoff < MIN_BACKOFF_FACTOR)) {
498 printf("%s: backoff factor %.1f not valid, must be between %.1f and %.1f\n",
499 prog, backoff, MIN_BACKOFF_FACTOR, MAX_BACKOFF_FACTOR);
500 usage();
501
502 }
503
504 if(count > MAX_COUNT) {
505 printf("%s: count %u not valid, must be less than %u\n",
506 prog, count, MAX_COUNT);
507 usage();
508 }
509
510 if(count_flag) {
511 alive_flag = unreachable_flag = 0;
512 }
513
514 trials = (count > retry + 1) ? count : retry + 1;
515
516 /* handle host names supplied on command line or in a file */
517 /* if the generate_flag is on, then generate the IP list */
518 argv = &argv[optind];
519
520 /* cover allowable conditions */
521
522 /* generate requires command line parameters beyond the switches */
523 if(generate_flag && !*argv) {
524 printf("generate flag requires command line parameters beyond switches\n");
525 usage();
526 }
527
528 if(*argv && !generate_flag) {
529 while(*argv) {
530 if(!(host_ptr->entry = malloc(strlen(*argv) + 1))) {
531 crash("Failed to allocate memory for hostname");
532 }
533 memset(host_ptr->entry, 0, strlen(*argv) + 1);
534 host_ptr->entry = memcpy(host_ptr->entry, *argv, strlen(*argv));
535 if(!(host_ptr->next = malloc(sizeof(struct host_name_list))))
536 crash("Failed to allocate memory for hostname");
537 host_ptr = host_ptr->next;
538 host_ptr->next = NULL;
539
540// add_name(*argv);
541 argv++;
542 }
543 }
544
545 // now add all the hosts
546 host_ptr = host_base_ptr;
547 while(host_ptr->next) {
548 add_name(host_ptr->entry);
549 host_ptr = host_ptr->next;
550 }
551
552 if(!num_hosts) {
553 printf("No hosts to work with!\n\n");
554 usage();
555 }
556
557 /* allocate array to hold outstanding ping requests */
558 table = (HOST_ENTRY **) malloc(sizeof(HOST_ENTRY *) * num_hosts);
559 if(!table) crash("Can't malloc array of hosts");
560
561 cursor = rrlist;
562
563 for(num_jobs = 0; num_jobs < num_hosts; num_jobs++) {
564 table[num_jobs] = cursor;
565 cursor->i = num_jobs;
566
567 cursor = cursor->next;
568 } /* FOR */
569
570 ping_pkt_size = ping_data_size + SIZE_ICMP_HDR;
571
572 signal(SIGINT, (void *)finish);
573
574 gettimeofday(&start_time, &tz);
575 current_time = start_time;
576
577 last_send_time.tv_sec = current_time.tv_sec - 10000;
578
579 cursor = rrlist;
580 advance = 0;
581
582 /* main loop */
583 while(num_jobs) {
584 /* fetch all packets that receive within time boundaries */
585 while(num_pingsent &&
586 cursor &&
587 cursor->num_sent > cursor->num_recv &&
588 wait_for_reply(sock)) ;
589
590 if(cursor && advance) {
591 cursor = cursor->next;
592 }
593
594 gettimeofday(&current_time, &tz);
595 lt = timeval_diff(&current_time, &last_send_time);
596 ht = timeval_diff(&current_time, &cursor->last_send_time);
597
598 advance = 1;
599
600 /* if it's OK to send while counting or looping or starting */
601 if(lt > interval) {
602 /* send if starting or looping */
603 if((cursor->num_sent == 0)) {
604 send_ping(sock, cursor);
605 continue;
606 } /* IF */
607
608 /* send if counting and count not exceeded */
609 if(count_flag) {
610 if(cursor->num_sent < count) {
611 send_ping(sock, cursor);
612 continue;
613 } /* IF */
614 } /* IF */
615 } /* IF */
616
617 /* is-it-alive mode, and timeout exceeded while waiting for a reply */
618 /* and we haven't exceeded our retries */
619 if((lt > interval) && !count_flag && !cursor->num_recv &&
620 (ht > timeout) && (cursor->waiting < retry + 1)) {
621 num_timeout++;
622
623 /* try again */
624 send_ping(sock, cursor);
625 continue;
626 } /* IF */
627
628 /* didn't send, can we remove? */
629
630 /* remove if counting and count exceeded */
631 if(count_flag) {
632 if((cursor->num_sent >= count)) {
633 remove_job(cursor);
634 continue;
635 } /* IF */
636 } /* IF */
637 else {
638 /* normal mode, and we got one */
639 if(cursor->num_recv) {
640 remove_job(cursor);
641 continue;
642 } /* IF */
643
644 /* normal mode, and timeout exceeded while waiting for a reply */
645 /* and we've run out of retries, so node is unreachable */
646 if((ht > timeout) && (cursor->waiting >= retry + 1)) {
647 num_timeout++;
648 remove_job(cursor);
649 continue;
650
651 } /* IF */
652 } /* ELSE */
653
654 /* could send to this host, so keep considering it */
655 if(ht > interval) {
656 advance = 0;
657 }
658 } /* WHILE */
659
660 finish();
661 return 0;
662} /* main() */
663
664/************************************************************
665 * Description:
666 *
667 * Main program clean up and exit point
668 ************************************************************/
669void finish()
670{
671 int i;
672 HOST_ENTRY *h;
673
674 gettimeofday(&end_time, &tz);
675
676 /* tot up unreachables */
677 for(i=0; i<num_hosts; i++) {
678 h = table[i];
679
680 if(!h->num_recv) {
681 num_unreachable++;
682 status = fin_stat = STATE_CRITICAL;
683 if(num_hosts == 1) {
684 printf("CRITICAL - %s is down (lost 100%%)|"
685 "rta=;%d;%d;; pl=100%%;%d;%d;;\n",
686 h->host,
687 warn.rta / 100, crit.rta / 100,
688 warn.pl, crit.pl);
689 }
690 else {
691 printf("%s is down (lost 100%%)", h->host);
692 }
693 }
694 else {
695 /* reset the status */
696 status = STATE_OK;
697
698 /* check for warning before critical, for debugging purposes */
699 if(warn.rta <= h->total_time / h->num_recv) {
700/* printf("warn.rta exceeded\n");
701*/ status = STATE_WARNING;
702 }
703 if(warn.pl <= ((h->num_sent - h->num_recv) * 100) / h->num_sent) {
704/* printf("warn.pl exceeded (pl=%d)\n",
705 ((h->num_sent - h->num_recv) * 100) / h->num_sent);
706*/ status = STATE_WARNING;
707 }
708 if(crit.rta <= h->total_time / h->num_recv) {
709/* printf("crit.rta exceeded\n");
710*/ status = STATE_CRITICAL;
711 }
712 if(crit.pl <= ((h->num_sent - h->num_recv) * 100) / h->num_sent) {
713/* printf("crit.pl exceeded (pl=%d)\n",
714 ((h->num_sent - h->num_recv) * 100) / h->num_sent);
715*/ status = STATE_CRITICAL;
716 }
717
718 if(num_hosts == 1 || status != STATE_OK) {
719 printf("%s - %s: rta %s ms, lost %d%%",
720 status_string[status], h->host,
721 sprint_tm(h->total_time / h->num_recv),
722 h->num_sent > 0 ? ((h->num_sent - h->num_recv) * 100) / h->num_sent : 0
723 );
724 /* perfdata only available for single-host stuff */
725 if(num_hosts == 1) {
726 printf("|rta=%sms;%d;%d;; pl=%d%%;%d;%d;;\n",
727 sprint_tm(h->total_time / h->num_recv), warn.rta / 100, crit.rta / 100,
728 h->num_sent > 0 ? ((h->num_sent - h->num_recv) * 100) / h->num_sent : 0, warn.pl, crit.pl
729 );
730 }
731 else printf(" :: ");
732 }
733
734 /* fin_stat should always hold the WORST state */
735 if(fin_stat != STATE_CRITICAL && status != STATE_OK) {
736 fin_stat = status;
737 }
738 }
739 }
740
741 if(num_noaddress) {
742 printf("No hostaddress specified.\n");
743 usage();
744 }
745 else if(num_alive != num_hosts) {
746 /* for future multi-check support */
747 /*printf("num_alive != num_hosts (%d : %d)\n", num_alive, num_hosts);*/
748 fin_stat = STATE_CRITICAL;
749 }
750
751 if(num_hosts > 1) {
752 if(num_alive == num_hosts) {
753 printf("OK - All %d hosts are alive\n", num_hosts);
754 }
755 else {
756 printf("CRITICAL - %d of %d hosts are alive\n", num_alive, num_hosts);
757 }
758 }
759 exit(fin_stat);
760}
761
762
763void send_ping(int lsock, HOST_ENTRY *h)
764{
765 char *buffer;
766 struct icmp *icp;
767 PING_DATA *pdp;
768 int n;
769
770 buffer = (char *)malloc((size_t) ping_pkt_size);
771 if(!buffer)
772 crash("can't malloc ping packet");
773
774 memset(buffer, 0, ping_pkt_size * sizeof(char));
775 icp = (struct icmp *)buffer;
776
777 gettimeofday(&h->last_send_time, &tz);
778
779 icp->icmp_type = ICMP_ECHO;
780 icp->icmp_code = 0;
781 icp->icmp_cksum = 0;
782 icp->icmp_seq = h->i;
783 icp->icmp_id = ident;
784
785 pdp = (PING_DATA *) (buffer + SIZE_ICMP_HDR);
786 pdp->ping_ts = h->last_send_time;
787 pdp->ping_count = h->num_sent;
788
789 icp->icmp_cksum = in_cksum((u_short *) icp, ping_pkt_size);
790
791 n = sendto(lsock, buffer, ping_pkt_size, 0,
792 (struct sockaddr *)&h->saddr, sizeof(struct sockaddr_in));
793
794 if(n < 0 || (unsigned int)n != ping_pkt_size) {
795 if(unreachable_flag) {
796 printf("%s error while sending ping: %s\n",
797 h->host, strerror(errno));
798 } /* IF */
799
800 num_unreachable++;
801 remove_job(h);
802 } /* IF */
803 else {
804 /* mark this trial as outstanding */
805 h->resp_times[h->num_sent] = RESP_WAITING;
806
807 h->num_sent++;
808 h->waiting++;
809 num_pingsent++;
810 last_send_time = h->last_send_time;
811 } /* ELSE */
812
813 free(buffer);
814} /* send_ping() */
815
816int wait_for_reply(int lsock)
817{
818 int result;
819 static char buffer[4096];
820 struct sockaddr_in response_addr;
821 struct ip *ip;
822 int hlen;
823 struct icmp *icp;
824 int n;
825 HOST_ENTRY *h = NULL;
826 long this_reply;
827 int this_count;
828 struct timeval sent_time;
829
830 result = recvfrom_wto(lsock, buffer, sizeof(buffer),
831 (struct sockaddr *)&response_addr, select_time);
832
833 if(result < 0) return 0; /* timeout */
834
835 ip = (struct ip *)buffer;
836
837#if defined( __alpha__ ) && __STDC__ && !defined( __GLIBC__ )
838 /* The alpha headers are decidedly broken.
839 * Using an ANSI compiler, it provides ip_vhl instead of ip_hl and
840 * ip_v. So, to get ip_hl, we mask off the bottom four bits.
841 */
842 hlen = (ip->ip_vhl & 0x0F) << 2;
843#else
844 hlen = ip->ip_hl << 2;
845#endif /* defined(__alpha__) && __STDC__ */
846
847 if(result < hlen + ICMP_MINLEN) {
848 printf("received packet too short for ICMP (%d bytes from %s)\n", result,
849 inet_ntoa(response_addr.sin_addr));
850
851 return (1); /* too short */
852 } /* IF */
853
854 icp = (struct icmp *)(buffer + hlen);
855 if(icp->icmp_type != ICMP_ECHOREPLY) {
856 /* handle some problem */
857 if(handle_random_icmp(icp, &response_addr))
858 num_othericmprcvd++;
859
860 return 1;
861 } /* IF */
862
863 if(icp->icmp_id != ident)
864 return 1; /* packet received, but not the one we are looking for! */
865
866 num_pingreceived++;
867
868 if(icp->icmp_seq >= (n_short) num_hosts)
869 return(1); /* packet received, don't worry about it anymore */
870
871 n = icp->icmp_seq;
872 h = table[n];
873
874 /* received ping is cool, so process it */
875
876 gettimeofday(&current_time, &tz);
877 h->waiting = 0;
878 h->num_recv++;
879
880 memcpy(&sent_time, icp->icmp_data + offsetof(PING_DATA, ping_ts),
881 sizeof(sent_time));
882 memcpy(&this_count, icp->icmp_data, sizeof(this_count));
883
884 this_reply = timeval_diff(&current_time, &sent_time);
885 h->total_time += this_reply;
886 total_replies++;
887
888 /* note reply time in array, probably */
889 if((this_count >= 0) && ((unsigned int)this_count < trials)) {
890 if(h->resp_times[this_count] != RESP_WAITING) {
891 printf("%s : duplicate for [%d], %d bytes, %s ms",
892 h->host, this_count, result, sprint_tm(this_reply));
893
894 if(response_addr.sin_addr.s_addr != h->saddr.sin_addr.s_addr)
895 printf(" [<- %s]\n", inet_ntoa(response_addr.sin_addr));
896 } /* IF */
897 else h->resp_times[this_count] = this_reply;
898 } /* IF */
899 else {
900 /* count is out of bounds?? */
901 printf("%s : duplicate for [%d], %d bytes, %s ms\n",
902 h->host, this_count, result, sprint_tm(this_reply));
903 } /* ELSE */
904
905 if(h->num_recv == 1) {
906 num_alive++;
907 } /* IF */
908
909 return num_jobs;
910} /* wait_for_reply() */
911
912int handle_random_icmp(struct icmp *p, struct sockaddr_in *addr)
913{
914 struct icmp *sent_icmp;
915 u_char *c;
916 HOST_ENTRY *h;
917
918 c = (u_char *) p;
919 switch (p->icmp_type) {
920 case ICMP_UNREACH:
921 sent_icmp = (struct icmp *)(c + 28);
922
923 if((sent_icmp->icmp_type == ICMP_ECHO) &&
924 (sent_icmp->icmp_id == ident) &&
925 (sent_icmp->icmp_seq < (n_short) num_hosts)) {
926 /* this is a response to a ping we sent */
927 h = table[sent_icmp->icmp_seq];
928
929 if(p->icmp_code > ICMP_UNREACH_MAXTYPE) {
930 printf("ICMP Unreachable (Invalid Code) from %s for ICMP Echo sent to %s",
931 inet_ntoa(addr->sin_addr), h->host);
932
933 } /* IF */
934 else {
935 printf("ICMP Unreachable from %s for ICMP Echo sent to %s",
936 inet_ntoa(addr->sin_addr), h->host);
937
938 } /* ELSE */
939
940 if(inet_addr(h->host) == INADDR_NONE)
941 printf(" (%s)", inet_ntoa(h->saddr.sin_addr));
942
943 printf("\n");
944
945 } /* IF */
946
947 return 1;
948
949 case ICMP_SOURCEQUENCH:
950 case ICMP_REDIRECT:
951 case ICMP_TIMXCEED:
952 case ICMP_PARAMPROB:
953 sent_icmp = (struct icmp *)(c + 28);
954 if((sent_icmp->icmp_type = ICMP_ECHO) &&
955 (sent_icmp->icmp_id = ident) &&
956 (sent_icmp->icmp_seq < (n_short) num_hosts)) {
957 /* this is a response to a ping we sent */
958 h = table[sent_icmp->icmp_seq];
959 printf("ICMP Unreachable from %s for ICMP Echo sent to %s",
960 inet_ntoa(addr->sin_addr), h->host);
961
962 if(inet_addr(h->host) == INADDR_NONE)
963 printf(" (%s)", inet_ntoa(h->saddr.sin_addr));
964
965 printf("\n");
966 } /* IF */
967
968 return 2;
969
970 /* no way to tell whether any of these are sent due to our ping */
971 /* or not (shouldn't be, of course), so just discard */
972 case ICMP_TSTAMP:
973 case ICMP_TSTAMPREPLY:
974 case ICMP_IREQ:
975 case ICMP_IREQREPLY:
976 case ICMP_MASKREQ:
977 case ICMP_MASKREPLY:
978 default:
979 return 0;
980
981 } /* SWITCH */
982
983} /* handle_random_icmp() */
984
985int in_cksum(u_short * p, int n)
986{
987 register u_short answer;
988 register long sum = 0;
989 u_short odd_byte = 0;
990
991 while(n > 1) {
992 sum += *p++;
993 n -= 2;
994 } /* WHILE */
995
996 /* mop up an odd byte, if necessary */
997 if(n == 1) {
998 *(u_char *) (&odd_byte) = *(u_char *) p;
999 sum += odd_byte;
1000 } /* IF */
1001
1002 sum = (sum >> 16) + (sum & 0xffff); /* add hi 16 to low 16 */
1003 sum += (sum >> 16); /* add carry */
1004 answer = ~sum; /* ones-complement, truncate */
1005
1006 return (answer);
1007
1008} /* in_cksum() */
1009
1010void add_name(char *name)
1011{
1012 struct hostent *host_ent;
1013 int ipaddress;
1014 struct in_addr *ipa = (struct in_addr *)&ipaddress;
1015 struct in_addr *host_add;
1016 char *nm;
1017 int i = 0;
1018
1019 if((ipaddress = inet_addr(name)) != -1) {
1020 /* input name is an IP addr, go with it */
1021 if(name_flag) {
1022 if(addr_flag)
1023 add_addr(name, na_cat(get_host_by_address(*ipa), *ipa), *ipa);
1024 else {
1025 nm = cpystr(get_host_by_address(*ipa));
1026 add_addr(name, nm, *ipa);
1027
1028 } /* ELSE */
1029 } /* IF */
1030 else add_addr(name, name, *ipa);
1031
1032 return;
1033 } /* IF */
1034
1035 /* input name is not an IP addr, maybe it's a host name */
1036 host_ent = gethostbyname(name);
1037 if(host_ent == NULL) {
1038 if(h_errno == TRY_AGAIN) {
1039 u_sleep(DNS_TIMEOUT);
1040 host_ent = gethostbyname(name);
1041 } /* IF */
1042
1043 if(host_ent == NULL) {
1044 printf("%s address not found\n", name);
1045 num_noaddress++;
1046 return;
1047 } /* IF */
1048 } /* IF */
1049
1050 host_add = (struct in_addr *)*(host_ent->h_addr_list);
1051 if(host_add == NULL) {
1052 printf("%s has no address data\n", name);
1053 num_noaddress++;
1054 return;
1055 } /* IF */
1056 else {
1057 /* it is indeed a hostname with a real address */
1058 while(host_add) {
1059 if(name_flag && addr_flag)
1060 add_addr(name, na_cat(name, *host_add), *host_add);
1061 else if(addr_flag) {
1062 nm = cpystr(inet_ntoa(*host_add));
1063 add_addr(name, nm, *host_add);
1064 } /* ELSE IF */
1065 else {
1066 add_addr(name, name, *host_add);
1067 }
1068
1069 if(!multif_flag) break;
1070
1071 host_add = (struct in_addr *)(host_ent->h_addr_list[++i]);
1072 } /* WHILE */
1073 } /* ELSE */
1074} /* add_name() */
1075
1076
1077char *na_cat(char *name, struct in_addr ipaddr)
1078{
1079 char *nm, *as;
1080
1081 as = inet_ntoa(ipaddr);
1082 nm = (char *)malloc(strlen(name) + strlen(as) + 4);
1083
1084 if(!nm)
1085 crash("can't allocate some space for a string");
1086
1087 strcpy(nm, name);
1088 strcat(nm, " (");
1089 strcat(nm, as);
1090 strcat(nm, ")");
1091
1092 return (nm);
1093
1094} /* na_cat() */
1095
1096
1097void add_addr(char *name, char *host, struct in_addr ipaddr)
1098{
1099 HOST_ENTRY *p;
1100 unsigned int n;
1101 int *i;
1102
1103 if(!(p = (HOST_ENTRY *) malloc(sizeof(HOST_ENTRY)))) {
1104 crash("can't allocate HOST_ENTRY");
1105 }
1106
1107 memset((char *)p, 0, sizeof(HOST_ENTRY));
1108
1109 p->name = name;
1110 p->host = host;
1111 p->saddr.sin_family = AF_INET;
1112 p->saddr.sin_addr = ipaddr;
1113 p->running = 1;
1114
1115 /* array for response time results */
1116 if(!(i = (int *)malloc(trials * sizeof(int)))) {
1117 crash("can't allocate resp_times array");
1118 }
1119
1120 for(n = 1; n < trials; n++)
1121 i[n] = RESP_UNUSED;
1122
1123 p->resp_times = i;
1124
1125 if(!rrlist) {
1126 rrlist = p;
1127 p->next = p;
1128 p->prev = p;
1129 } /* IF */
1130 else {
1131 p->next = rrlist;
1132 p->prev = rrlist->prev;
1133 p->prev->next = p;
1134 p->next->prev = p;
1135 } /* ELSE */
1136
1137 num_hosts++;
1138} /* add_addr() */
1139
1140
1141void remove_job(HOST_ENTRY * h)
1142{
1143 h->running = 0;
1144 h->waiting = 0;
1145 num_jobs--;
1146
1147
1148 if(num_jobs) {
1149 /* remove us from list of active jobs */
1150 h->prev->next = h->next;
1151 h->next->prev = h->prev;
1152 if(h == cursor) cursor = h->next;
1153 } /* IF */
1154 else {
1155 cursor = NULL;
1156 rrlist = NULL;
1157 } /* ELSE */
1158
1159} /* remove_job() */
1160
1161
1162char *get_host_by_address(struct in_addr in)
1163{
1164 struct hostent *h;
1165 h = gethostbyaddr((char *)&in, sizeof(struct in_addr), AF_INET);
1166
1167 if(h == NULL || h->h_name == NULL)
1168 return inet_ntoa(in);
1169 else
1170 return (char *)h->h_name;
1171
1172} /* get_host_by_address() */
1173
1174
1175char *cpystr(char *string)
1176{
1177 char *dst;
1178
1179 if(string) {
1180 dst = (char *)malloc(1 + strlen(string));
1181 if(!dst) crash("malloc() failed!");
1182
1183 strcpy(dst, string);
1184 return dst;
1185
1186 } /* IF */
1187 else return NULL;
1188
1189} /* cpystr() */
1190
1191
1192void crash(char *msg)
1193{
1194 if(errno || h_errno) {
1195 if(errno)
1196 printf("%s: %s : %s\n", prog, msg, strerror(errno));
1197 if(h_errno)
1198 printf("%s: %s : A network error occurred\n", prog, msg);
1199 }
1200 else printf("%s: %s\n", prog, msg);
1201
1202 exit(STATE_UNKNOWN);
1203} /* crash() */
1204
1205
1206long timeval_diff(struct timeval *a, struct timeval *b)
1207{
1208 double temp;
1209
1210 temp = (((a->tv_sec * 1000000) + a->tv_usec) -
1211 ((b->tv_sec * 1000000) + b->tv_usec)) / 10;
1212
1213 return (long)temp;
1214
1215} /* timeval_diff() */
1216
1217
1218char *sprint_tm(int t)
1219{
1220 static char buf[10];
1221
1222 /* <= 0.99 ms */
1223 if(t < 100) {
1224 sprintf(buf, "0.%02d", t);
1225 return (buf);
1226 } /* IF */
1227
1228 /* 1.00 - 9.99 ms */
1229 if(t < 1000) {
1230 sprintf(buf, "%d.%02d", t / 100, t % 100);
1231 return (buf);
1232 } /* IF */
1233
1234 /* 10.0 - 99.9 ms */
1235 if(t < 10000) {
1236 sprintf(buf, "%d.%d", t / 100, (t % 100) / 10);
1237 return (buf);
1238 } /* IF */
1239
1240 /* >= 100 ms */
1241 sprintf(buf, "%d", t / 100);
1242 return (buf);
1243} /* sprint_tm() */
1244
1245
1246/*
1247 * select() is posix, so we expect it to be around
1248 */
1249void u_sleep(int u_sec)
1250{
1251 int nfound;
1252 struct timeval to;
1253 fd_set readset, writeset;
1254
1255 to.tv_sec = u_sec / 1000000;
1256 to.tv_usec = u_sec - (to.tv_sec * 1000000);
1257/* printf("u_sleep :: to.tv_sec: %d, to_tv_usec: %d\n",
1258 (int)to.tv_sec, (int)to.tv_usec);
1259*/
1260 FD_ZERO(&writeset);
1261 FD_ZERO(&readset);
1262 nfound = select(0, &readset, &writeset, NULL, &to);
1263 if(nfound < 0)
1264 crash("select() in u_sleep:");
1265
1266 return;
1267} /* u_sleep() */
1268
1269
1270/************************************************************
1271 * Description:
1272 *
1273 * receive with timeout
1274 * returns length of data read or -1 if timeout
1275 * crash on any other errrors
1276 ************************************************************/
1277/* TODO: add MSG_DONTWAIT to recvfrom flags (currently 0) */
1278int recvfrom_wto(int sock, char *buf, int len, struct sockaddr *saddr, int timo)
1279{
1280 int nfound = 0, slen, n;
1281 struct timeval to;
1282 fd_set readset, writeset;
1283
1284 to.tv_sec = timo / 1000000;
1285 to.tv_usec = (timo - (to.tv_sec * 1000000)) * 10;
1286
1287/* printf("to.tv_sec: %d, to.tv_usec: %d\n", (int)to.tv_sec, (int)to.tv_usec);
1288*/
1289
1290 FD_ZERO(&readset);
1291 FD_ZERO(&writeset);
1292 FD_SET(sock, &readset);
1293 nfound = select(sock + 1, &readset, &writeset, NULL, &to);
1294 if(nfound < 0) crash("select() in recvfrom_wto");
1295
1296 if(nfound == 0) return -1; /* timeout */
1297
1298 if(nfound) {
1299 slen = sizeof(struct sockaddr);
1300 n = recvfrom(sock, buf, len, 0, saddr, &slen);
1301 if(n < 0) crash("recvfrom");
1302 return(n);
1303 }
1304
1305 return(0); /* 0 bytes read, so return it */
1306} /* recvfrom_wto() */
1307
1308
1309/*
1310 * u = micro
1311 * m = milli
1312 * s = seconds
1313 */
1314int get_threshold(char *str, threshold *th)
1315{
1316 unsigned int i, factor = 0;
1317 char *p = NULL;
1318
1319 if(!str || !strlen(str) || !th) return -1;
1320
1321 for(i=0; i<strlen(str); i++) {
1322 /* we happily accept decimal points in round trip time thresholds,
1323 * but we ignore them quite blandly. The new way of specifying higher
1324 * precision is to specify 'u' (for microseconds),
1325 * 'm' (for millisecs - default) or 's' for seconds. */
1326 if(!p && !factor) {
1327 if(str[i] == 's') factor = 1000000; /* seconds */
1328 else if(str[i] == 'm') factor = 1000; /* milliseconds */
1329 else if(str[i] == 'u') factor = 1; /* microseconds */
1330 }
1331
1332 if(str[i] == '%') str[i] = '\0';
1333 else if(str[i] == ',' && !p && i != (strlen(str) - 1)) {
1334 p = &str[i+1];
1335 str[i] = '\0';
1336 }
1337 }
1338
1339 /* default to milliseconds */
1340 if(!factor) factor = 1000;
1341
1342 if(!p || !strlen(p)) return -1;
1343 th->rta = (unsigned int)strtoul(str, NULL, 0) * factor;
1344 th->pl = (unsigned int)strtoul(p, NULL, 0);
1345 return 0;
1346}
1347
1348/* make a blahblah */
1349void usage(void)
1350{
1351 printf("\nUsage: %s [options] [targets]\n", prog);
1352 printf(" -H host target host\n");
1353 printf(" -b n ping packet size in bytes (default %d)\n", ping_data_size);
1354 printf(" -n|p n number of pings to send to each target (default %d)\n", count);
1355 printf(" -r n number of retries (default %d)\n", retry);
1356 printf(" -t n timeout value (in msec) (default %d)\n", timeout / 100);
1357 printf(" -i n packet interval (in msec) (default %d)\n", DEFAULT_INTERVAL);
1358/* XXX - possibly on todo-list
1359 printf(" -m ping multiple interfaces on target host\n");
1360 printf(" -a show targets that are alive\n");
1361 printf(" -d show dead targets\n");
1362*/ printf(" -v show version\n");
1363 printf(" -D increase debug output level\n");
1364 printf(" -w warning threshold pair, given as RTA[ums],PL[%%]\n");
1365 printf(" -c critical threshold pair, given as RTA[ums],PL[%%]\n");
1366 printf("\n");
1367 printf("Note:\n");
1368 printf("* This program requires root privileges to run properly.\n");
1369 printf(" If it is run as setuid root it will halt with an error if;\n");
1370 printf(" interval < 25 || retries > 5\n\n");
1371 printf("* Threshold pairs are given as such;\n");
1372 printf(" 100,40%%\n");
1373 printf(" to set a threshold value pair of 100 milliseconds and 40%% packetloss\n");
1374 printf(" The '%%' sign is optional, and if rta value is suffixed by;\n");
1375 printf(" s, rta time is set in seconds\n");
1376 printf(" m, rta time will be set in milliseconds (this is default)\n");
1377 printf(" u, rta time will be set in microseconds\n");
1378 printf(" Decimal points are silently ignored for sideways compatibility.\n");
1379 printf("\n");
1380 exit(3);
1381} /* usage() */